Nvidia’s RTX 30-series graphics cards made a big splash when they began arriving all the way back in 2020. Those products delivered a huge performance leap in their day, but time marches on for us all. The oldest Ampere cards are just a few months away from their sixth birthdays, and even though Nvidia has continued to support 30-series cards with its latest Game Ready driver optimizations and DLSS model upgrades, other signature GeForce features like DLSS Frame Generation are never coming to Ampere.

Even where new software features are technically supported, Ampere comes with big asterisks. DLSS 4.5 is the first upscaling model to take advantage of FP8 acceleration that’s exclusive to RTX 40- and 50-series Tensor Cores. RTX 30-series cards can still technically run DLSS 4.5 upscaling models, but the improved image quality they offer now demands a significant performance penalty from Ampere compared to past DLSS versions. And if you want to experiment with frame generation, you have to deal with the lower image quality of AMD’s cross-platform FSR 3.1 framegen tech, assuming it’s available at all in a given title.

Those software limitations aren’t insurmountable obstacles to a good gaming experience, but VRAM is a different story. Ampere cards arrived when games were less hungry for VRAM than they are today, and even the RTX 3080 has just 10GB to play with. Unless you bought into the highest end of the Ampere range, you’re likely feeling constrained by your card’s 8GB of VRAM with max settings in the latest games at resolutions higher than 1080p.

If any or all of those limitations have you itching for a more powerful, more flexible modern GPU, and you’d rather not navigate our GPU Hierarchy to figure out what constitutes a true upgrade, worry not. We’ve done the hard number-crunching work and thought through the most common gaming scenarios to arrive at the best upgrade path for each common Ampere card.

So what defines an upgrade for the purposes of this guide? First and foremost, we want to see at least a 1.5x improvement in overall raster performance from GPU to GPU as a baseline, and larger leaps are even better. The architectural advances of Nvidia’s latest Blackwell GPUs naturally mean you’ll enjoy improved RT gaming performance from our picks, as well.

Whatever your individual feelings for upscaling and framegen might be, you’ll enjoy greater freedom to play with DLSS 4.5 and MFG on the latest GeForces. Our path-traced performance results with Pragmata illustrate why you should use those features to the fullest, but it’s ultimately your choice. All told, raster gaming performance boosts still matter most, so that’s our hard line.

If you’re already in the Nvidia fold, we expect that you want to stay there, so we’ve made our picks assuming as much. But where a given Radeon card might make sense, we’ve included it as a suggestion if you’re open to switching.

Monitoring the situation

Before we talk about specific upgrade paths for your GPU, we need to take a moment and consider the monitor you’re using with it. If you only have a 1080p or 60Hz monitor, a fixed-refresh-rate panel, or all of the above, your graphics card likely shouldn’t be your first or only upgrade. It’s overwhelmingly likely that you won’t enjoy a perceptibly smoother or lower-latency gaming experience on a 60Hz monitor than you currently do with the graphics card you already own.

The continuing development of high-quality upscaling tech means that monitor resolution is no longer a hard wall for gaming smoothness and responsiveness. Instead, it’s a hard cap on the image quality you can achieve. To get the most out of DLSS (or FSR), you really want to give those upscalers the highest output resolution and highest refresh rate to work with that you can. Upscaling from lower resolution to 1080p just isn’t worth it anymore unless you absolutely must, while upscaling to 4K using DLSS 4.5 Performance requires only a small frame-rate penalty relative to native 1080p output.

Along with more and more affordable OLED options, we’re seeing plenty of dual-mode monitors that can offer high-refresh-rate 4K output alongside even faster 1080p modes for downright affordable prices. And broadly compatible variable-refresh-rate tech is now in virtually every gaming monitor, so you can likely enable G-Sync or G-Sync Compatible modes with the GPU you already have.

Best of all, displays are one of the few PC gaming products that don’t rely heavily on advanced logic chips or DRAM to work, so prices for gaming monitors have remained reasonable even as everything else has gotten eye-wateringly expensive. If your monitor is older than your Ampere GPU, it’s likely high time for an upgrade. Start there first.

RTX 3080 Ti, RTX 3090, and RTX 3090 Ti: Wait for a compelling upgrade

If you’re one of the lucky gamers with an RTX 3080 Ti, RTX 3090, or RTX 3090 Ti, you can rest easy knowing that your graphics card has plenty of life left in it. Any upgrade right now is elective rather than essential, especially if you’re already using DLSS upscaling. The RTX 3080 Ti’s 12GB of VRAM is the only conceivable pain point we can see in this upper tier of Ampere.

The only cards that are likely to feel like major upgrades for you are the RTX 4090 or RTX 5090, and unless you’re willing to compete with local LLM trailblazers for a used 4090 or put down nearly 2X MSRP for a new 5090, your best bet is to hold onto your current card unless you’re really feeling limited by its ray-tracing or path-tracing horsepower, the lack of FP8 Tensor Core acceleration for DLSS 4.5, or the lack of high-quality framegen support.

If you’re only gaming, 24GB of VRAM isn’t doing much for you, and you might ponder selling your RTX 3090 or 3090 Ti to one of those same LLM enthusiasts while the market is hot and putting the proceeds toward a new RTX 5080, which is substantially faster and more power-efficient than those cards and gives you full-speed access to DLSS 4.5 and MFG. But both of those things are nice to have rather than essentials.

RTX 3080: Upgrade if you’re feeling the VRAM pinch

The RTX 3080 isn’t constrained so much by its compute horsepower, which remains strong in pure raster gaming, as it is by its 10GB of VRAM. Especially if you’re trying to game at 1440p or 4K with max settings and minimal upscaling, you are likely finding that your 3080’s VRAM is the biggest obstacle to achieving the best combo of performance and image quality nowadays.

Unfortunately, the most tangible upgrades for the 3080 are the currently pricey RTX 5070 Ti or the chronically overpriced RTX 5080. Both of those cards will feel much faster for both raster and ray-traced games, they enable full-speed DLSS 4.5 upscaling for practically free performance boosts, and they give you access to framegen and MFG juice that the 3080 doesn’t support at all.

But you’d have to be really hurting for an upgrade to shell out for either of those cards, given their stiff premiums over MSRP right now. A faster, higher-resolution monitor paired with DLSS 4 upscaling (Preset K in Nvidia App override language) is a cheaper, easier path if you haven’t already tried it.

But if you’re a 3080 gamer looking for that “wow” moment from a new GPU, the RTX 5070 Ti and 5080 are the way, and their prices are what they are.

RTX 3070 and RTX 3070 Ti: Upgrade now

As with the RTX 3080, the biggest performance constraint for the RTX 3070 and 3070 Ti these days isn’t necessarily raw compute, but VRAM. Nvidia only ever offered these cards in 8GB flavors, and they still offer solid 1080p gaming performance with the latest titles. But if you’ve tried to max out settings at 1440p or above, you’ve likely felt the squeeze of that limited VRAM pool.

And VRAM-hungry RT gaming is inadvisable on the 3070 and 3070 Ti, as you’re going to be leaning hard on DLSS to even get to a fuzzy 1080p output. It’s just not worth it.

The RTX 5070 neatly solves all of the problems. It delivers the large baseline performance boost we want to see for a true upgrade, it has 12GB of VRAM for more demanding games at higher resolutions, and its support for DLSS 4.5 and MFG unlocks the latest tools for achieving high output image quality and smoothness. And it does all of this at a relatively reasonable price, even in today’s graphics card market.

If you don’t bleed green, you can get even more VRAM and a bit higher performance with the impressively fast and efficient Radeon RX 9070, whose prices also haven’t risen too terribly amid the current RAMpocalypse.

But that move is a bit of a leap of faith given the spotty support for FSR 4 AI upscaling and frame gen in the latest titles, and it also comes with the risk that you’ll be entirely locked out of next-gen features like path tracing, as we’ve seen in the latest Capcom games. If none of that sounds bothersome to you, though, the RX 9070 is worth a look as a possible option.

RTX 3060 Ti: Upgrade now

Like the RTX 3070 and 3070 Ti, Nvidia only ever offered the RTX 3060 Ti in an 8GB flavor, and that’s a tough enough limitation these days. But the 3060 Ti’s somewhat lower compute horsepower is a correspondingly greater liability as games march ever forward.

As with the RTX 3070 and 3070 Ti, your best upgrade bet is the RTX 5070. You’ll feel an even bigger boost in performance than you will with the 3070 duo, and you get more VRAM and better DLSS support than your existing card to go with it. Easy.

RTX 3060 12GB: Upgrade if you can make the 5070 leap

The RTX 3060 12GB keeps going and going thanks to its unusually large VRAM pool for a budget-friendly GPU. But that VRAM is paired with just so-so compute horsepower that’s really showing its age in the latest games.

The most natural upgrade for the 3060 in normal times would be the RTX 5060 Ti 16GB, which delivers the performance leap we want at a reasonable MSRP. But the ongoing RAMpocalypse has made those cards too scarce and too pricey to recommend. And the RTX 5060 Ti 8GB’s performance is all too likely to fall off a cliff due to its small VRAM pool, so we’d never suggest it as an upgrade.

Sorry to repeat ourselves, but the best step up from the 3060 12GB in today’s chaotic market is the RTX 5070, assuming you can afford it. It more than doubles the 3060 12GB’s gaming performance even at 1080p, and that gap becomes even more pronounced at 1440p and beyond.

Critically, the 5070 doesn’t leave you with less VRAM than you already have. Add in vastly better RT performance and support for DLSS Multi Frame Generation, and an RTX 5070 is a truly transformative gaming upgrade.

If you need a more budget-friendly upgrade than the RTX 5070 amid the RAMpocalypse, your only reasonable choice is the Radeon RX 9060 XT 16GB. That card delivers the 1.5x basic performance improvement that we want to see for the money, it runs FSR 4 AI upscaling and framegen where it’s available, and it slots into roughly the same power envelope as the RTX 3060 12GB. But you’ll need to be comfortable with leaving the Nvidia fold, and that might be too big a leap for some.

RTX 3050: Upgrade now

As the entry-level Ampere card, the RTX 3050 was already wimpy when it hit the market, and time hasn’t treated it well. It lands well below the 60 FPS mark in our overall standings, even at 1080p, and we’ve found that its baseline performance is so low that DLSS doesn’t improve things much.

If you’re still gaming on an RTX 3050 and have the freedom not to, it’s dead simple to find a compelling upgrade. Even the humble RTX 5060 handily doubles the 3050’s average frame rate across our tests at 1080p, and assuming you don’t run into VRAM limitations, the 5060 can even deliver a smooth 1440p gaming experience if 60 FPS is an acceptable baseline on average. That major performance boost barely comes with increased power requirements, so you won’t have to budget for a PSU upgrade, either.

But the 5060’s 8GB of VRAM means that you might still run into performance pitfalls in modern games, especially if you want to try out RT and DLSS framegen. If you’re only playing Counter-Strike 2, Fortnite, or Apex Legends, that’s less of a problem than it might be if you’re keen for the latest AAA experiences or PlayStation ports. But you’ll always be thinking about

If you’re looking for a card with greater longevity than the 8GB 5060 and aren’t married to the Nvidia ecosystem, we’d also check out the Radeon RX 9060 XT 16GB. It’s even faster than the 5060 in our tests and will allow you to start properly exploring ray tracing in titles that support it. It offers high-quality FSR 4 upscaling and framegen, and it won’t crush a small or aging PSU.

Bottom line

Nvidia graphics cards are lasting longer than ever, thanks to common architecture capabilities like RT and Tensor Cores, in addition to a continuously improving DLSS software suite that boosts both performance and image quality. But even with those benefits, games are continuously advancing, too, and today’s titles are starting to put pressure on even the top-end RTX 30-series GPUs of yore.

The AI boom and the accompanying RAMpocalypse both make this a tough time to upgrade for gamers’ wallets, but if you’re still using an RTX 30-series GPU and are starting to feel the upgrade itch, the relative stability of gaming GPU roadmaps also means that you can make the leap to a Blackwell card (or competing Radeon) with some confidence that you won’t be taken by a surprise next-generation GPU launch any time soon.

That said, if you’re still happy with your current graphics card and gaming monitor setup, don’t mind missing out on DLSS Multi-Frame Generation, and don’t feel the need to explore maxed-out ray-traced or path-traced effects in the latest titles, we wouldn’t blame you for squeezing every last bit of life out of your Ampere GPU. No matter which path you take, it’s hard to lose.

Getting into PC hardware and building can be daunting. There’s a ton of jargon, spools of technical details, and the thorny topic of compatibility to consider, just to avoid the simplest errors. However, Redditor JoeBiden-mytime (JB) has got off to a grand start thanks to a generous neighbor. They are now the owner of a potent PC packing an RTX 3090 FE and liquid-cooled i9-10850K, which said neighbor passed on to them – for free.

Being a newcomer to the PC DIY scene, JB posted details of their neighbor’s kind donation onto the PCBuild subreddit. Under the heading of ‘Got a free PC,’ they asked “are these components any good or are they outdated?” JB admitted they were a PC building greenhorn, but were attracted to the platform.

Of course, some wise crackers said the system was junk and offered to take it out of JB’s hands to recycle, for free. The first two pictures shared by JB showed that the system was a little dusty, which may have made the poster wonder how old the PC was. However, they’ve followed up with more specs, so we have a good overview of this great neighborly gift.

Out of interest, we have tabulated the key components in this free PC next to their prices at launch:

To get a modern GPU to replace or rival the RTX 3090 in 2026, you would probably be pondering an RTX 5070 Ti from the green team or an AMD RX 9070/XT. Both of those have less VRAM (16GB vs 24GB) but not dreadfully so, and are more modern and efficient. Replacing this single component would set you back around $1,000 right now, if you wanted to stay with Nvidia.

The freebie PC’s supporting components are suitably premium, and appear to include an NZXT Kraken X63 AiO and an NZXT H510i case. The Task Manager screenshot also shows that the system came with 16GB of RAM and a pair of SSDs. In the same screenshot, we see this new member of the PC community is busy browsing the Steam sales for the unbeatable breadth and depth of titles available on the platform.

Later on, we see some pictures of the system powered up by JB, with RGB lighting across the fans, cooler, RAM sticks, etc. It still looks a little dusty, though, and perhaps the new owner of this older-but-gold system should do a bit of dust busting to keep the system ticking over smoothly. Free doesn't mean junk.

Since the US imposition of sanctions prohibiting the sale of high-performance GPUs to China, indigenous manufacturers have taken it upon themselves to mitigate these limitations by modifying existing solutions. One of the results of these modding endeavors is an RTX 4090 with 48GB of GDDR6X memory, which is now becoming commonplace among local circles. Russian YouTuber Мой Компьютер (My Computer) obtained a blower-style edition RTX 4090 48GB, later taking it apart to examine the PCB among other design features.

Both the RTX 4090 and RTX 3090 Ti's PCBs carry 12 available memory slots, which equate to 24GB of VRAM with one 16Gb IC populating each slot, and that's not the configuration we're looking for. The base RTX 3090, on the other hand, offers 24 memory slots (12 on each side of the PCB), suggesting a modified version of it could be used as a foundation here. It's hard to be certain, however. Either way, the RTX 4090 48GB we're examining today features a board design that strongly resembles RTX 30 series PCBs.

Matching the cooler's design, the YouTuber links this card to Peladin, a Chinese mini-PC/GPU manufacturer. The reviewer asserts the GPU's PCB is well-designed, with robust VRMs, and the added memory on the backside is covered with high-quality thermal pads and a backplate to keep temperatures in check. Unlike workstation-grade RTX 6000/5880 Ada GPUs, which stick with GDDR6, this RTX 4090 is equipped with GDDR6X memory, which is faster but also runs hotter.

Surprisingly, the GPU was recognized in Windows, and you can even use traditional GeForce drivers instead of relying on potentially risky third-party options. The YouTuber mentions a breach at Nvidia where hackers obtained internal tools, including MATS and MODS. Modders can employ these utilities to modify VRAM, disable non-working channels, and even flash new volumes on the GPU's firmware without violating any checksums or digital signatures.

The test bench used for the handful of benchmarks conducted featured the Ryzen 5 7600X3D, and 48GB of DDR5-6000 memory, on the Maxsun Challenger B650M motherboard. Under load, the GPU is quite loud, defined as "sounding like an airplane" at 65dB with fan speeds exceeding 5,000 RPM. That's one of the key drawbacks of blower-style coolers, but it was a necessary tradeoff to enable higher compatibility in servers. Even so, during a half-hour test in Superposition at 8K, the card maintained a relatively stable 2.7 GHz clock speed at 70 degrees Celsius with the memory at 86 degrees Celsius.

After extensively testing the GPU in synthetic benchmarks, LLMs, and games, the YouTuber revealed they were provided the GPU by XON, a local store specializing in workstations and servers. At the time of recording, the GPU was listed at $4,400 (370,000 Rubles). While expensive, this might still be one of the most affordable 48GB options on the market, at least until we see Nvidia's Blackwell workstation in action.

Another hold has been blown through the hull of the dreaded Akira ransomware attack: Blogger Tinyhack has discovered a new exploit to brute-force the virus's encryption — and has reportedly already used it to restore the data of an attacked company.

Akira is a well-known ransomware cyberattack, used by hackers in November to demand a $125k ransom in baguettes from a French company. It may now be escapable by affected companies thanks to a GPU-based brute-force counterattack. With an RTX 4090, the Tinyhack found they could crack the encrypted ransomware files in seven days, and with 16 GPUs, the process would take just over ten hours.

Akira-based ransomware attacks aimed at high-profile targets; it was first discovered in 2023 and known for ludicrously high ransom requests (sometimes reaching tens of millions of dollars). In 2023, Avast's Threat Research Team found the method Akira used to encrypt victim files, and published a free encryption breaker tool to free computers from the dreaded attack. Akira then patched this high-profile crack, adding some bespoke details to its originally publicly available encryption methods.

At least one Akira variant uses an encryption method that can be decrypted via the new GPU-based brute-force method over a period of days or weeks. The Akira attack uses the chacha8 and Kcipher2 encryption methods to generate per-file encryption keys, using four distinct timestamps, in nanoseconds, as seeds.

These timestamps can be deduced to a tight range of on average 5 million nanoseconds (0.005 seconds), and then precisely found with brute-force, a process requiring top-end GPUs such as Nvidia's RTX 3090 or 4090.

Several things must go right for those hoping to execute the decryption method. Encrypted files must be untouched following the encryption so the timestamp that the file was last accessed can be found and used for the brute-force, for example.

Using an NFS (as opposed to files just living on the network's local disks) can also complicate decryption, as server lag will make it more difficult to determine the true timestamps used by the encryption.

Using an RTX 4090, decrypting a single file by running through every possible nanosecond in the average range of 4.5 million nanoseconds, finding the correct four timestamps, and generating the appropriate decryption keys takes around 7 days. Affected organizations are recommended to rent servers through services like runpod or vast.ai, using multiple GPU servers to bring the time down.

Tinyhack's client took around 3 weeks to successfully decrypt a full set of VM files.

Ransomware attacks are most often impossible to decrypt without paying ransom, so finding a method to circumvent the attack is a big win for cybersecurity research. While those behind Akira will likely quickly patch this method for future attacks as they did after the Avast decryption release, those already hit by Akira may be able to free infected systems with this method.

Tinyhack's blog post runs through the entire process of discovering the vulnerability and full instructions to decrypt with it, so please head there to get an exhaustive look at brute-forcing a way into Akira. Ransomware has come a long way since its beginnings on a floppy disk sent by mail, and today marks another victory against it.

North West Repair has ably demonstrated that "death by liquid metal" isn't necessarily permanent. After analyzing and diagnosing a deceased Nvidia GeForce RTX 3090, the repair tech demonstrated dexterous GPU and VRAM chip removal, cleaning, reballing, replacement, and testing. Except for one of the VRAM chips requiring its solder balls to be reflowed a second time, the repair went without a hitch, and his customer must have been happy to get back a fully working (and still potent in 2024) GeForce RTX 3090.

Liquid metal thermal interface material (TIM) is notoriously tricky to handle, with commercial applications using specially designed application tools and barriers to stop it escaping from where it's needed. Asus, for example, is known to use solder mask on nearby conductive surfaces and a tiny barrier sponge just 0.1mm tall around processor sockets. Unlike the long-established oxide thermal pastes, liquid metal conducts electricity, so if it seeps out it can cause all sorts of electrical havoc (usually short circuits). This seems to be exactly what has happened to the RTX 3090 in the North West Repair video.

In the video, repairer Tony explains that he received an Nvidia Founders Edition GeForce RTX 3090 "after an unsuccessful liquid metal application." He seems to be wary of such cases, recommending good-quality thermal paste instead. The first job is to check the graphics card isn't completely dead, and to assess the spread of the liquid metal. It is 'known' to be under the GPU – but did it get into any other nooks and crannies?

A multimeter confirmed that most circuits were OK, but the card wouldn't output any video signal. The next step, connecting the card to run tests on its memory banks, revealed that two GDDR6X chips were causing errors. Knowing liquid metal got under the core, it was suspected that it went under from the upper-left side, affecting the two marked memory chips.

Tony then got to work, using specialized equipment to evenly heat the GPU and memory chips to around 250 degrees Celsius. This allowed the spill-affected components to be removed. Next, the cleanup process began, and North West Repair noted an unusual coating on the freshly exposed PCB area that had to be cleaned with adhesive remover.

The next segment of the video shows the GPU and two VRAM chips getting cleaned and solder re-balled, which is a tricky specialized task. Tony makes this look easy though. After the successful component re-ball, the components were soldered back to the graphics card PCB for testing. As we mentioned in the intro, one of the memory chips precipitated an error in the text phase, so the problematic chip got a fresh squirt of flux and was reheated to the solder melting point, for an operation called a reflow.

After a last spell of cleaning, using an ultrasonic bath to remove old dried-up flux residue, the GeForce RTX 3090 card was ready for testing again and passed all Tony's heat, stress, and benchmark testing. It was a happy ending for a card that easily could have been tossed in the trash after the owner's very expensive mistake.

On paper, Nvidia's RTX 6000 Ada professional graphics card has more compute horsepower than the GeForce RTX 4090 flagship board for gamers. Yet, when tested in 3DMark TimeSpy Graphics benchmark, it could not actually beat it, according to a Reddit post noticed by @harukaze5719. 

Nvidia's RTX 6000 Ada 48GB graphics board is based around AD102 GPU with 18,176 CUDA cores enabled. While Nvidia and its partners do not specify clocks of the GPU, they disclose that the card has a peak compute performance of 91.1 FP32 TFLOPS (which suggests clocks of around 2,505 MHz). By contrast, Nvidia's GeForce RTX 4090 24GB uses AD102 GPU with 16,3840 CUDA cores clocked at up to 2,520 MHz and rated for up to 82.575 FP32 TFLOPS. 

There is one major difference between the two cards (although they seem to use the same PCB): the professional one is rated for up to 300W total board power, whereas the gaming one can consume up to 450W of power, which means that the latter can run at high clocks considerably longer than the former. On the one hand, this limits performance, but on the other hand this increases longevity of the RTX 6000 Ada versus GeForce RTX 4090. 

This is exactly the case with performance of the two cards in 3DMark Time Spy Graphics benchmark. The RTX 6000 Ada's Graphics score was 30,158 points with the latest Windows update (as seen in the screenshot), but it could score 36,844 points before the update, according to the individual who posted the score, Reddit user Healthy-Blood-54. Meanwhile, GeForce RTX 4090 can easily score 40,964 points. 

The reason for such relatively low performance are of course comparatively low GPU clocks that drop from a whopping 3+ GHz in the first couple of minutes of the test to around 1.5 GHz later on. Interestingly, the owner says the board does not overheat, so perhaps it starts to drop clocks at relatively low temperatures. 

"When running full usage, it is about 85°C on the core with the fans at around 70%, explained Healthy-Blood-54. "No louder than the [RTX] A6000 [based on GA102]. But again, remember these are battle cards. These can probably run at 95°C core 24/7 for 10 years before any issues. The cost is not just in the VRAM." 

While the RTX 6000 Ada 48GB graphics card cannot beat the GeForce RTX 4090 despite the higher number of enabled CUDA cores and higher peak FP32 compute throughput, the rumored GeForce RTX 4090 Ti with 18,176 stream processors enabled will likely beat the current flagship by a wide margin.

Baka, a Chinese PC hardware enthusiast and content creator, has attempted to use a modern gaming PC equipped with Intel's Core i9-13900K processor and Asus's GeForce RTX 4090 ROG Strix graphics card at subzero temperatures outside in the city of Mohe for a night, reports WCCFTech. To a great surprise, the hardware not only worked fine for some time, but actually survived the experiment. 

Modern high-end processors and graphics cards for desktops dissipate so much heat that usage of advanced liquid coolers is recommended for CPUs, whereas graphics boards come with rather monstrous three-wide triple-fan air cooling systems. But such sophisticated coolers apparently aren't necessary if the temperature falls between -30 degrees Celsius to -53 C (-22 degrees Fahrenheit to -63.4 F) outside and you can ensure that that the hardware does not short circuit. 

At first, Baka just installed her PC outside and removed its side panel, as it did not let her film the insides at such temperatures. But at -30 C and -40 C, liquid in the closed-loop 360-mm liquid CPU cooler froze and the device simply stopped working. To cool down the processor, she had to install Noctua's NH-P1 passive heatsink and build a monstrous air compressor out of dozens of high-pressure server fans that created sound that was as horrific as the compressor looked.

But it worked out just fine. The CPU temperature dropped to 10°C to 15°C and the 24-core Raptor Lake processor — the best gaming CPU you can get these days — had enough thermal headroom to hit 6.18 GHz. Temperature of Nvidia's GeForce RTX 4090 — the best graphics card available today — dropped to -40°C at idle and never went past 30°C under high loads. The machine survived at subzero temperatures for a couple of nights, although it is unclear how significantly hardware degraded when working in such mode. We would be particularly concerned about DRAM and 3D NAND memory chips as well as things like MACs, DACs, and capacitors. 

Baka used the system for some to play some Minesweeper, but did not attempt to overclock the CPU and the GPU. This is probably because it was pointless to compete against extreme overclockers who use liquid nitrogen(-90°C).

In any case, while it is hardly possible to use high-end PCs in a freezer for prolonged periods, the experiment looks like a lot of fun. 

Nvidia has quietly started selling its RTX 6000 Ada Generation graphics card. Based on the AD102 GPU with 18,176 CUDA cores enabled, this is the 'fattest' AD102 configuration available to date — 142 of the 144 Streaming Multiprocessors are enabled. Rated for 300W of power consumption, the board is designed for computer aided design, digital content creation, virtual desktop infrastructure, and other professional applications.

Pricing varies pretty widely, ranging from $6,800 to $8,600 depending on the retailer. Nvidia has the card listed at $6,800, with a limit of five per customer — just in case you're trying to put together a bunch of workstations. Other outlets like CDW are marking it up an additional 25% or so, with availability set for 4–6 weeks out.

Nvidia's RTX 6000 Ada based on the AD102 GPU has 18,176 CUDA cores enabled, and while Nvidia doesn't specify clocks anywhere that we can find, it's partners like PNY and Leadtek quote to 91.1 FP32 TFLOPS of compute performance. That equates to a GPU boost clock of 2505 MHz and is about 10% higher than the GeForce RTX 4090, which features 16,384 CUDA cores. However, there are clear differences between the RTX 6000 Ada and the consumer-focused RTX 4090.

The RTX 6000 Ada is designed for professional applications, and so it carries 48GB of GDDR6 memory with ECC enabled, featuring a peak memory bandwidth of 960 MB/s — slightly lower than 1,008 MB/s offered by the GeForce RTX 4090. That suggests Nvidia is using 20 Gbps GDDR6 memory rather than the slightly faster 21 Gbps GDDR6X, but power use may also be slightly lower on the GDDR6 chips.

The RTX 6000 comes equipped with four DisplayPort 1.4a connectors that can drive four 4K or 5K displays (4K at 240 Hz is supported via DSC), or two 8K displays (120 Hz with DSC).

One of the intriguing things about the RTX 6000 Ada is its power consumption. Despite its higher compute performance and 48GB of memory onboard, the RTX 6000 Ada is rated for just 300W of power, down from the 450W rating on the gamer-oriented GeForce RTX 4090. Both GPUs have similar official boost clocks of around 2.5 GHz, though the minimum guaranteed clocks on the RTX 6000 are lower and we suspect it won't clock quite as high in heavy computational workloads.

Because of its relatively limited power consumption, the RTX 6000 Ada Generation graphics card comes with a dual-wide cooling system and a blower fan to ensure that it fits into workstations and servers. It uses one 12VHPWR (16-pin CEM 5.0 PCIe) power connector for power delivery, so it will require an appropriate cable adapter to fit into existing machines that do not have a native 16-pin connector.

Another interesting thing about Nvidia's RTX 6000 Ada professional graphics card is its price. Nvidia sells the board for $6,800, slightly lower than the $6,999 launch price of its RTX A6000 predecessor. But value added reseller CDW lists the product for $8,615, while a Japanese retailer found by @momomo_us lists the unit for $8,524 without tax. We are not sure why resellers are charging so much more than Nvidia itself, but for now it looks like it makes the most sense to buy the RTX 6000 Ada board directly from the Nvidia.

Nvidia "unlaunched" its 12GB RTX 4080 Ti late last year, amidst backlash over its fundamentally different specs than the full-fat RTX 4080, and its $899 price. But after rebadging and repackaging, that card is returning to store shelves on January 5th as the RTX 4070 Ti, with a $799 starting price that's at least slightly easier to stomach.

Announced earlier today at Nvidia's CES keynote kickoff, the full specs of the RTX 4070 Ti have yet to be officially revealed. But just days ago, Nvidia accidentally listed the card on its own site, before pulling it. So we know the card will sport 7,680 CUDA cores, 12GB of GDDR6X memory and a boost clock of 2.61 GHz — just like the graphics card formerly known as the 4080 12GB.

For its part, Nvidia has now re-launched a product page for the RTX 4070 Ti, where it emphasizes its better-than RTX 3090 Ti performance at not much more than half the power consumption. Sorry to any of the poor souls who bought an RTX 3090 Ti when it was selling for $2,000 or more just 10 months ago! The company says the RTX 4070 Ti idles at 12W, and pulls an average of 226W while gaming, compared to 398W on the RTX 2090 Ti.

Of course, DLSS 3 support also comes along for the ride, which will become much more important as more games launch that support it. According to Nvidia's latest count, DLSS 3 support for existing and upcoming games sits at 50. Nvidia also says the 4070 Ti should deliver up to a 70% performance boost over the RTX 3070 Ti in apps like Blender, Unity, and Unreal Engine.

Nvidia has lined up its partners to carry this card's launch, and links to RTX 4070 Ti cards from Asus, Colorful, Galax, Gigabyte, MSI, PNY, Zotac and others. We'll  need to test the card to see if it has what it takes to make our Best Graphics Cards page. But even if it performs well at $799, the real question is going to be how many cards will be available at that price, or if the vast majority of models will sell at or above the original $899 asking price of the now re-branded RTX 4080 12GB.

There's also still the question of how the RTX 4070 Ti performs without DLSS 3, which is a big part of the "faster than 3090 Ti" story so far. While 50 released and upcoming games might feature support for the new frame doubling technology, it's not without critics and at times creates artifacts. We also tested Portal RTX and found that, as expected, you need a base performance well above 30 fps before Frame Generation kicks in if  you don't want a game to feel sluggish.

$799 might be more palatable than the $899 price originally announced for the RTX 4080 12GB, but it's also $200 more than the starting price for the outgoing RTX 3070 Ti. The Ada Lovelace architecture has plenty of new features, and we should see a big jump in performance relative to the old 3070 Ti — and 50% more memory will be beneficial as well, even if it sits on a 192-bit interface, as the larger L2 cache size comes into play. Still, it's sad to see such a massive increase in generational pricing.

An interesting new benchmark result, apparently involving a device with an Nvidia GeForce RTX 4090 Laptop GPU, has surfaced from Twitter’s Benchleaks. According to the accompanying sys-info captured by Geekbench, the RTX 4090 Laptop will feature 9,728 CUDA cores and 16GB of VRAM, matching the RTX 4080 desktop graphics card. As for performance, it is noted that the ‘Notebook X370SNx’ laptop under scrutiny scored 210,290 in Geekbench Open CL tests. Benchleaks notes that this score is nearly 50% faster than an RTX 3070, but it is also slightly better than some RTX 3090 desktop graphics cards in the online database. Remember to add a pinch of salt to this benchmark result news.

We have seen several indications of the impending release of powerful new laptops leveraging the potent twin charms of Intel’s Raptor Lake mobile CPUs and Nvidia’s RTX 40 (Ada Lovelace architecture) GPUs. The Notebook X370SNx of an unknown brand looks like it could be one of the top models due to be unveiled at CES 2023 next week. Geekbench data indicates it includes the unannounced Intel Core i9-13900HX CPU, with 24C / 32T and a maximum turbo of 5.2 GHz. On the GPU side of this offering, the aforementioned RTX 4090 Laptop GPU has 76 SMs (for 9,728 CUDA cores), runs at a max 2.04 GHz, and is accompanied by 16GB of VRAM. This Windows 11 toting laptop also featured a respectable 32GB of system memory in the configuration tested.

Geekbench isn’t among the best benchmarks for real-world gaming performance – an aspect of GPUs such as the RTX 4090 Laptop we would be most interested in. However, it is all we really have right now, so we have constructed a table with a handful of comparative OpenCL scores featuring new and familiar GPUs.

It is quite exciting to see the laptop under test provide a better OpenCL score than the powerful desktop RTX 3090. Another important comparison you will see in the table above is that the RTX 4090 Laptop offers over 50% better OpenCL performance than the previous mobile GPU champ, the RTX 3080 Ti Laptop. This is a big step for mobile graphics, yet the new RTX 4090 Laptop isn’t expected to have a TDP much higher than the RTX 3080 Ti laptop (between 150 and 175W).

Remember, laptops like the Notebook X370SNx are expected to launch next week after Intel and Nvidia announce the related mobile parts. During this big tech event, we will have coverage of all the latest and greatest developments in laptops, so please stay tuned.

The Nvidia GeForce RTX 4080 is the follow-up to last month's RTX 4090 launch, now one of the best graphics cards and the top listing in our GPU benchmarks hierarchy. Of course, a bit of the shine has come off thanks to the melting 16-pin connectors. The good news: RTX 4080 uses less power, which should mean it's also less likely to funnel enough power to melt the plastic connector… maybe. The bad news: At $1,199, it's still priced out of reach for most gamers and represents a big jump in generational pricing, inheriting the RTX 3080 Ti launch price that we also felt was too high.

We already know most of what to expect from Nvidia's Ada Lovelace architecture, so the only real question now is how performance scales down to fewer GPU shaders, less memory, less cache, a narrower memory interface, etc. Let's quickly look at the specifications for a few of the top Nvidia and AMD GPUs. 

There's a relatively large gap between the RTX 4080 and the larger RTX 4090. You get most of an AD103 GPU — 76 of the potential 80 Streaming Multiprocessors (SMs) — but that's still 40% fewer GPU shaders and other functional units than the RTX 4090. Clock speeds are similar, you get 33% fewer memory channels, VRAM, and bandwidth, and the rated TBP drops by 29%. On paper, the RTX 4090 could be up to 70% faster based on the theoretical compute performance, and that's a concern.

$1,199 is hardly affordable, so it feels like anyone even looking at the RTX 4080 should probably just save up the additional $400 for the RTX 4090 and go for broke — or melted. But then the RTX 4090 has been sold out at anywhere below $2,100 since launch, which means it could actually be a $900 upsell, and that's far more significant.

The pricing becomes even more of a concern when we factor in AMD's Radeon RX 7900 XTX/XT cards coming next month. We now have all the pertinent details for the first cards using AMD's RDNA 3 GPU architecture, and they certainly look promising. Prices are still high, but the specs comparisons suggest AMD might be able to beat the RTX 4080 while costing at least $200–$300 less. This means, unless you absolutely refuse to consider purchasing an AMD graphics card, you should at least wait until next month to see what the red team has to offer.

However, Nvidia does have some extras that AMD is unlikely to match in the near term. For example, the Deep learning and AI horsepower in RTX 4080 far surpass what AMD intends to offer. If we've got the figures right, AMD's FP16 and INT8 throughput will be less than a third of the RTX 4080.

Nvidia also offers DLSS 3 courtesy of the enhanced Optical Flow Accelerator (OFA). Ten games already support the technology: Bright Memory: Infinite, Destroy All Humans! 2 - Reprobed, F.I.S.T.: Forged in Shadow Torch, F1 22, Justice, Loopmancer, Marvel’s Spider-Man Remastered, Microsoft Flight Simulator, A Plague Tale: Requiem, and Super People. That's about half as many DLSS 3 games in less than a month as those with AMD's FSR2 technology. Of course, you need an RTX 40-series GPU for DLSS 3, while FSR2 works with pretty much everything.

Nvidia GPUs also tend to be heavily favored by professional users, or at least their employers. So while true workstations will likely opt for the RTX 6000 48GB card as opposed to a GeForce RTX 40-series, there's certainly potential in picking up one or more RTX 4080 cards for AI and deep learning use. Content creators may also find something to like, though again, if you're willing to pay for a 4080, it may not be a huge step up in pricing to nab a 4090 instead.

Another piece of good news (depending on which side of the aisle you fall, we suppose) is that GPU mining remains unprofitable. Gamers won't be able to offset the price of a new graphics card through cryptocurrency mining, but at least there should be more GPUs available for gamers. Now let's see exactly what Nvidia has to offer with its new RTX 4080. 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

Nvidia RTX 4080 Founders Edition Design

Nvidia's RTX 4080 Founders Edition looks identical to the RTX 4090 FE — that goes for the new packaging as well. Both also look mostly the same as the previous generation RTX 3090 FE. The new cards aren't quite as long, but they're a bit taller and thicker to make up for it. The RTX 4080 card measures 304 x 137 x 61mm and weighs 2132g, a bit lighter than the 4090's 2189g, thanks to having two fewer GDDR6X chips and a smaller GPU.

The industrial design of the RTX 40-series hasn't changed much at all, though the end results have improved substantially. The two fans are slightly larger than on the equivalent 30-series, and of course, where the 3080 and 3080 Ti were dual-slot cards that tended to run quite hot, the triple-slot design proves more than adequate for the RTX 4080.

That also goes for the video outputs. You get three DisplayPort 1.4a and a single HDMI 2.1 port, all with Display Stream Compression (DSC) support. Those can handle up to 4K at 240 Hz or 8K at 60 Hz, which should be sufficient for the time being — we're still waiting for monitors that exceed those values to actually show up. AMD's DisplayPort 2.1 54 Gbps is superior, but it may not matter too much in practical terms unless you upgrade to a 4K display that supports a refresh rate above 240.

Things have changed a lot more under the hood, both for the GPU and the GDDR6X memory. We've tested five RTX 4090 cards now, and while all of them run the memory at the official 21 Gbps speed, two of the cards (Asus and Colorful) appear to have used faster 24 Gbps chips that Micron down-binned and labeled as 21 Gbps chips. The result was that memory temperatures, even under our most strenuous workloads, were only in the 60–65C range, almost 20C lower than the 'normal' 21 Gbps chips.

With the RTX 4080, all models should come with 24 Gbps GDDR6X memory since it's the only option that officially supports the 22.4 Gbps speed. That means the memory should stay chill, and it's exactly what we saw in testing: The RTX 4080 Founders Edition showed VRAM temperatures that peaked at 52C, even when we pushed the overclock as far as we could (we'll cover overclocking on the next page).

Unfortunately for Nvidia, our biggest concern right now has to be the 16-pin power connector. We haven't had any problems, but over 20 reported cases suggest there's more going on than just a few incorrectly connected cards. Does the single 16-bit connector even help compared to dual 8-pin connectors? Nope. It isn't easy to imagine a scenario where the cost savings of a single 16-pin connector with an adapter significantly outweigh the cost of dual 8-pin connectors.

It's worth pointing out that the RTX 4080 doesn't get the same 'quadropus' adapter as the RTX 4090 cards we've tested. Instead, you get a triple 8-pin to single 16-pin adapter. Will that be less prone to failure? We don't know, but the 320W TBP ought to help.

Nvidia RTX 4080 Founders Edition Internals

We don't normally disassemble reference cards like the RTX 4080 Founders Edition, since we like to keep them in their original state for future testing purposes. Nvidia provided the above photos of the PCB, GPU, and cooler, showing clear differences between the 4080 and 4090 PCBs. That's to be expected.

The two cards use different GPUs (AD102 and AD103), with the 4080 supporting two fewer GDDR6X memory channels and chips. That results in a lot of changes with the power phases and other aspects, and yet the two PCBs are the same dimensions so they can use the same cooler. It's clearly overkill for the 4080, but at least you shouldn't need to worry about the GPU or memory ever overheating.

The RTX 4080 Founders Edition board shows two input power phases, ten more for the GPU and then six more that appear to be for the GDDR6X memory — a 2+10+6 power phase design. (I might have that wrong, but there are 18 large inductors in various places on the board.) In addition, the RTX 4090 Founders Edition uses a 2+20+3 phase layout, while partner cards may have as many as 2+24+4 phases. You can also see eleven spots where the 4080 PCB has the mounting points for additional inductors but doesn't bother to fill them.

It's a safe bet that some partner cards will push much higher power limits, with more phases. The 16-pin connector using a triple 8-pin input should allow for up to 450W, plus 75W from the PCIe x16 slot, so there's ample headroom if an AIB (add-in board) partner chooses to expose it.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

Nvidia RTX 4080 Founders Edition Overclocking

As always, overclocking involves trial and error, and results are not guaranteed. We attempt to dial in stable settings while running some stress tests, but what at first appears to work just fine may crash once we start running through our gaming suite.

We started by maxing out the power limit, which in this case was only 110%. Higher clocks were able to complete some tests, but we eventually settled in at a +200 MHz GPU core clock and +1500 MHz (25.4 Gbps effective) memory clock. We also set a custom fan speed that ramps from 30% at 30C up to 100% at 80C, though we never approached 80C in testing.

As with the RTX 4090, there's no way to increase the GPU voltage short of doing a voltage mod (not something we wanted to do), which seems to be a limiting factor. We did get above 3 GHz in some of our initial testing, and with our final overclock, we saw clocks in the high 2.9 and low 3.0 GHz range. We'll include our overclocking results for 1440p and 4K in our charts. 

Nvidia RTX 4080 Test Setup

We updated our GPU test PC and gaming suite in early 2022, and we continue to use the same hardware for now. Nvidia recommends using an AMD Ryzen 9 7950X or Intel Core i9-13900K to further remove CPU bottlenecks… but I don't have one of those yet. (We're working on getting one for an updated GPU testbed.) We do enable XMP for a modest performance boost, and if you're actually going to go out and buy an RTX 4080 or 4090, it's at least worth considering a full PC upgrade.

Our CPU sits in an MSI Pro Z690-A DDR4 WiFi motherboard with DDR4-3600 memory — a nod to sensibility rather than outright maximum performance. We also upgraded to Windows 11 and are now running the latest 22H2 version (with VBS and HVCI disabled, as well as the latest patches) to ensure we get the most out of Alder Lake. You can see the rest of the hardware in the boxout.

We used Nvidia-provided 526.72 pre-release drivers for the RTX 4080. Most of the other cards, including the RTX 4090, were tested with earlier drivers, though we did retest a few games with the same 526.72 drivers where we saw performance oddities (i.e., the 4080 shouldn't outperform the 4090).

Our gaming tests consist of a "standard" suite of eight games without ray tracing enabled (even if the game supports it) and a separate "ray tracing" suite of six games that all use multiple RT effects. We've tested the RTX 4080 at all of our normal settings, though the focus here will be squarely on 4K and 1440p performance.

We tested at 1080p, which might be slightly more meaningful on the 4080 than on the 4090, but CPU limits definitely come into play. We also enabled and tested the RTX 4080 with DLSS 2 Quality mode at 1440p and 4K in the games that support it — two of which (Forza Horizon 5 and Flight Simulator) currently show no benefit and can actually have reduced performance. That's because the games are already hitting CPU limits, something DLSS 3 sort of overcomes.

As we did in the RTX 4090 Founders Edition review, we also tested DLSS 3 and total system latency in several games. The DLSS 3 tests will be limited to RTX 40-series GPUs, and we'll have to eventually look at retesting once final builds become available.

Finally, besides the gaming tests, we have a collection of professional and content creation benchmarks that can leverage the GPU. We're using SPECviewperf 2020 v3, Blender 3.30, OTOY OctaneBenchmark, and V-Ray Benchmark. 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

Given the price, we'll start with the performance that matters most: 4K at maxed-out settings. Some might argue that the xx80-class of Nvidia cards fit better with 1440p gaming monitors, but that's only true when they don't cost $1,200 or more.

There are several ways of looking at performance. The RTX 4080 delivers 78% of the performance of the RTX 4090 for 75% of the price. Alternatively, it's 21% faster than the previous generation RTX 3090 Ti and 51% faster than the RTX 3080. Except it's more of a replacement for the RTX 3080 Ti, in which case it's only 38% faster than the card it replaces.

As for AMD's previous generation cards, the RTX 4080 beats the RX 6950 XT by just 29% in our standard gaming suite. However, if AMD's claims of 50% to 70% more performance from the RX 7900 XTX prove true, the RTX 4080 could be in a lot of trouble next month.

The gains are decent overall, but after the RTX 4090 review, they don't feel particularly amazing. Remember, the 4090 beat the previous generation 3090 Ti by 56% and the RTX 3090 by 72%, and the price was roughly in the same ballpark. RTX 4080 improves gen-on-gen performance, regardless of which Ampere card you care to compare it against, but the improvement is more like what we've seen with previous generations of hardware, and the price is generally much higher.

For example, the RTX 3080 10GB was 35% faster than the RTX 2080 Ti, and it was priced $500 lower. This time, the RTX 4080 is 51% faster than the 3080 10GB, but it costs $500 more. That's a 71% increase in generational pricing, unless you're willing to pretend the 3080 didn't exist and focus only on the 3080 Ti.

Some games are still bumping into CPU bottlenecks, so it's also useful to look at the best-case and worst-case results. The biggest improvement we measured relative to the 3090 Ti was 49% (Total War: Warhammer 3), but some of that likely comes from the newer drivers. Forza Horizon 5 also improved by over 40%. But several games show less than a 20% improvement, and Flight Simulator was in the single-digit percentage points at 8%.

As for overclocking, considering we're bumping into CPU bottlenecks with the Core i9-12900K, it's no surprise that the boosted GPU and memory clocks don't help very much. We measured an average performance improvement of 6.9% across our standard suite, ranging from 4.4% (Forza) to 9.6% (Borderlands).

DLSS 2 Quality mode also improved performance by 18% in Red Dead Redemption 2 and around 35% in Horizon Zero Dawn and Watch Dogs Legion. Flight Simulator and Forza Horizon 5 both have DLSS 2 support now as well… but performance remained unchanged in our testing. It's potentially a bug in the DLSS implementation of Forza, while Flight Simulator is still fully CPU limited — which is where DLSS 3 comes into play.

Our ray tracing test suite proves far more demanding, fully eliminating any CPU bottlenecks at 4K — and in fact, some of the results are barely playable. Overall, the RTX 4080 delivers 26% more performance than the 3090 Ti, is 48% faster than the 3080 Ti, 72% faster than the 3080, and just over twice as fast as AMD's RX 6950 XT.

All six games support DLSS, and that tacks on another 89% using Quality mode — that doesn't include Minecraft, which automatically applies Performance mode at 4K. Overclocking still doesn't do much, providing a 7.4% bump in framerates.

While most of the games are technically still playable, the RTX 4080 really needs some form of upscaling — or alternatively, DLSS 3's Frame Generation — to deliver smooth framerates with maxed-out ray tracing settings at 4K. These results also show how much the Ada Lovelace architecture has improved Nvidia's RT performance, yet again keeping a wide gap between the Nvidia and AMD GPUs, though we'll have to see how RDNA 3 fares next month.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

Even though the RTX 4080 is the second-fastest graphics card right now, it's still clearly hitting CPU bottlenecks, especially in our standard test suite. While the 4080 was 22% slower than the 4090 at 4K, it's only 10% slower at 1440p. This is definitely a card that wants a higher resolution monitor — 3440x1440 ultrawide would be a bit more demanding than 2560x1440, but 4K remains the better choice for non-ray tracing games.

The gap between the 4080 and previous generation cards shrinks as well. It's still 17% faster than a 3090 Ti, 29% faster than a 3080 Ti, and 40% faster than the 3080 10GB, but those aren't massive gen-on-gen gains. The lead over AMD's 6950 XT has also shrunk to 14%, which could mean a serious blowout from the RX 7900 series when those arrive. Overclocking doesn't help much either, providing just 3.8% more performance.

Since the 4080 inherits the 3080 Ti's price point, let's stick with that comparison for more detail. Looking at the individual game charts, the 4080 ranges from being 8% slower in Flight Simulator (the 40-series still underperforms there compared to the 30-series, possibly due to drivers) to being up to 70% faster (Total War: Warhammer 3), with Borderlands 3 also showing around a 40% improvement.

If you use a 1440p monitor, the 4080 might be better than the overkill 4090. It's basically tied with the 4090 in half of the games, and the biggest gap (Borderlands 3) is only 23%. With both GPUs easily clearing 144 fps, this is about as fast as you would reasonably need to go for 1440p gaming purposes… until we get to the ray tracing results.

Complex ray tracing games remain mostly GPU-limited still, with the 4090 leading the 4080 by 34%. Meanwhile, the 4080 beats the 3090 Ti by 27%, the 3080 Ti by 47%, the 3080 by 66%, and the RX 6950 XT by 96%. The margins are a bit smaller than at 4K, but of course both AMD and Nvidia will generally agree that ray tracing games are one of the best options for using some form of upscaling.

With DLSS Quality mode (Balanced in Minecraft), RTX 4080 performance improves by 62%. That's enough to go from clearing 60 fps in all six games to breaking 100 fps. You'd still want a monitor with G-Sync (alternatively a G-Sync Compatible FreeSync display) support for the best experience, but even the most demanding games are well within reach of the 4080.

The 7% GPU overclock and 13% GDDR6X overclock help a bit as well, increasing performance by 6.6%. That suggests the GPU clocks are more important than the memory, though both contribute to the gains. As we'll show later, the minor boost in performance from overclocking also requires more power, though in the case of the 4080 Founders Edition, it's not too bad.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

We're not going to spend a lot of effort dissecting the 1080p gaming results. Yes, the RTX 4080 can run 1080p ultra very fast — about as fast as anything currently available. Perhaps a CPU upgrade would create more of a separation, but with our 12900K, the 4090 is only 8% faster than the 4080 for our standard test suite, though it remains 26% faster with the ray tracing games.

Naturally, the high-end previous generation cards also manage 1080p just fine for the most part. Because of that, the RTX 4080 only beats the 3090 Ti by 10%, RTX 3080 Ti by 18%, and the 3080 10GB by 26%. AMD's RX 6950 XT continues to perform very well at 1080p, thanks to its large Infinity Cache, and the 4080 lead shrinks to just 5% — with half of the games in our standard suite favoring the AMD card.

Ray tracing at maxed-out settings still isn't "easy," however. Compared to the RTX 30-series cards in the charts, the 1080p ultra DXR margins of victory are basically the same as at 1440p. The 4080 is also still 90% faster than the RX 6950 XT. We didn't bother with DLSS testing at 1080p, but it should still provide a decent boost.

Even so, the RTX 4080 really isn't built for 1080p gaming. Lighter games might run at 480 fps on the latest and greatest 1080p monitors, but they'd probably run just as fast on a slightly cheaper GPU as well. Perhaps in the next couple of years, games will again get more demanding and the 4080 might start to become less CPU limited, but we recommend it mostly for higher-resolution displays.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

As we did on the RTX 4090, we're providing a section on some of the forward-looking technologies like DLSS 3 with the RTX 4080. DLSS 3 Frame Generation requires (for now) an RTX 40-series graphics card, though games with DLSS 3 support are also backward compatible with DLSS 2. Several games with DLSS 3 are now publicly available, like A Plague Tale: Requiem and Microsoft Flight Simulator, and there are at least ten games with DLSS 3 support here or coming soon.

DLSS 3 can be the difference between minor performance increases and huge gains, but there's still plenty of debate about how much it improves the overall experience. Creating an intermediate frame adds latency and doesn't increase the user sampling rate, but games can look more fluid. Some people have found situations where the Frame Generation exhibits graphical glitches, so Nvidia and the game devs may still need to iron out a few kinks.

But if you just focus on the performance gains, some of the results are quite impressive. Four of the six DLSS 3-enabled benchmarks allowed us to enable Frame Generation while running at native resolution and improve performance by 50 to 75 percent. It's similar to DLSS 2 Quality mode, so you could potentially choose between Frame Generation or upscaling if you want.

Only two of the six tests failed to run at reasonably smooth (>50 fps) framerates at native resolution, so you could certainly argue that Frame Generation isn't even that critical in a lot of games. One of those two was the 3DMark DLSS feature test, so we can toss that out as well. That leaves just one game — Cyberpunk 2077 — where DLSS 3 might be truly necessary. DLSS Quality mode managed 50 fps at 4K on its own, while Quality upscaling plus Frame Generation allowed the 4080 to reach 75 fps, which was slightly higher than Performance mode's 4x upscaling.

But there's a potential downside with Frame Generation: latency. Nvidia recommends comparing Frame Generation with Reflex enabled to DLSS Super Sampling without Reflex. That does make DLSS 3 look… not as bad. But since games that support DLSS 3 will allow users to turn on Reflex, we feel most would appreciate lower latency gaming, whether or not they use Frame Generation.

We only collected latency data for Cyberpunk, but you can see the pattern, which should also hold in other games. Of course, latency largely depends on framerate, so higher FPS means lower latency gaming. But Reflex helps to eliminate extra frames of latency, dropping the 4080 from 70ms at native to 43ms with DLSS Quality mode, and 32ms with Performance mode — that's without Frame Generation.

With Frame Generation (DLSS 3), latency increases to 88ms at native, 58ms with DLSS Quality, and 45ms with DLSS Performance. Ultimately, you can choose between higher fps and lower latency with a larger upscaling factor, higher fps with more latency but with better upscaling quality, or a blend between the two. Most people likely wouldn't notice the slightly increased latency in a game like Cyberpunk, but competitive shooters like Fortnite favor lower latency gaming — without ray tracing enabled.

And finally, here's a look at just pure ray tracing performance — or at least as close as we can reasonably get to isolating RT performance in a game engine. Here's another case where the gap between the 4080 and 4090 looks more like a massive gulf. The 4090 is 63% faster, and the paltry 6% improvement from overclocking the 4080 does little to mitigate that.

On the other hand, the RTX 4080 still easily outclasses any other graphics card currently available. It's 38% faster than the 3090 Ti, 57% faster than the 3080 Ti, and 79% faster than the 3080 it replaces in name if not price. The 4080 also demolishes AMD's current best card and delivers over three times the performance.

In our AMD RDNA 3 architecture deep dive, we note that AMD claims ray tracing performance on the RX 7900 XTX has increased by 80%. That's a big improvement, but that could still mean the 4080 is potentially 70% faster in ray tracing potential. We'll have to see how things stand when we get AMD's cards in December, but it looks like users will again need to choose between maximum ray tracing performance and DLSS support, or potentially better rasterization performance and universally supported FSR2 upscaling.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

GPUs are also used with professional applications, AI training and inferencing, and more. We're hoping to add some AI benchmarks soon… but we ran out of time. For now, we've got several 3D rendering applications that leverage ray tracing hardware along with the SPECviewperf 2020 v3 test suite.

SPECviewperf 2020 consists of eight different benchmarks, and we use the geometric mean from those tests to generate an aggregate "overall" score. Note that this is not an official score, but it gives equal weight to the individual tests and provides a nice high-level overview of performance. Few professionals use all of these programs, however, so it's generally more important to look at the results for the applications you plan to use.

Across the eight tests, Nvidia's RTX 4080 trails the costlier 4090 by 20% on average. Performance is nearly tied in creo-03. Energy-03 and medical-03 favor the 4090 a bit more, but the performance difference is relatively similar to what we saw in our 4K gaming results. In turn, the 4080 beats the 3090 Ti by 20%, leads the 3080 Ti by 33%, and comes out 48% faster than the RTX 3080. It's only tied with the Titan RTX, thanks to the significantly higher score that card gets in snx-04.

AMD tends to be a bit more generous with professional application optimizations on consumer hardware, and the latest drivers gave a sizeable boost to AMD's SPECviewperf scores. That results in AMD delivering better performance than the RTX 4080 in half of the tests, and the major Nvidia deficit in snx-04 gives AMD's GPUs the overall lead as well. However, if your professional application of choice is 3D Studio Max, Maya, or SolidWorks, the RTX 4080 takes the lead.

Blender is a popular open-source rendering application, and we're using the latest Blender Benchmark, which uses Blender 3.30 and three tests. Blender 3.30 also includes the new Cycles X engine that leverages ray tracing hardware on AMD, Nvidia, and even Intel Arc GPUs. It does so via AMD's HIP interface (Heterogeneous-computing Interface for Portability), Nvidia's CUDA or OptiX APIs, and Intel's OneAPI.

Being open-source has one major advantage, and it's the ability of the various companies to deliver their own rendering updates. This gives us the closest option for an apples-to-apples performance comparison of professional 3D rendering using the various GPUs. Blender and other 3D rendering applications tend to show the raw ray tracing hardware performance, though architectural elements can come into play.

In this case, Nvidia's latest Ada Lovelace GPUs thoroughly crush the competition. The RTX 4080 delivers about four times the performance of the RX 6950 XT, and even the old Titan RTX comes in ahead of AMD's "best" with about double the performance.

That said, we have to wonder how much of the gap comes down to better optimizations in the code for Nvidia's CUDA and OptiX APIs. Maybe AMD will release a driver at some point that massively improves its Blender performance. Barring that, even if the RX 7900 XTX is twice as fast as the 6950 XT, it will still fall behind Nvidia's RTX 40-series cards.

Our final two professional applications only have ray tracing hardware support for Nvidia's GPUs. We reached out to the companies to find out why that's the case, and it largely comes down to reliability and stability — for example, Optane dropped OpenCL support because the API was no longer seeing active development.

Without AMD GPUs in the mix, all we can really see is how the RTX 4080 fares against the 4090 and 30-series cards. Even though it has less memory and memory bandwidth, the 4080 still leads the 3090 Ti by nearly 30% on average in Octane and closer to 35% in V-Ray. It's also 60–70 percent faster than the RTX 3080. Basically, Nvidia's RTX GPUs are the gold standard in professional 3D rendering applications for a reason: It was first with hardware, and it provided and continues to provide active developer support.

[I've run out of time (again), but I'm going to try to add some machine learning benchmarks in the near future. Stay tuned…]

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

We measure real-world power consumption using Powenetics testing hardware and software. We capture in-line GPU power consumption by collecting data while looping Metro Exodus (the original, not the enhanced version) and while running the FurMark stress test. Our test PC remains the same old Core i9-9900K as we've used previously, to keep results consistent.

For the RTX 4080 Founders Edition, we ran Metro at 3840x2160 using the Extreme preset (no ray tracing or DLSS), and we ran FurMark at 2560x1440. The following charts are intended to represent something of a worst-case scenario for power consumption, temps, etc.

The RTX 4080 Founders Edition used quite a bit less power than the rated 320W TBP. We could only break that limit with overclocking and FurMark, where we hit 332W — still less than the previous generation RTX 3080 running at stock settings. That's a nice change of pace, and the result is a card that's far more efficient than the specifications might lead you to believe.

Clock speeds were also quite impressive, averaging 2.78 GHz in Metro, though the GPU did drop to 2.3 GHz with FurMark. Meanwhile, our manual overclock pushed the GPU beyond 3.0 GHz — the first time we've ever managed that using a card's stock cooler! FurMark clocks also increased to 2.62 GHz with our overclock.

Temperatures, fan speeds, and noise levels all go hand-in-hand. Given the rather massive cooler, it's not too surprising that the RTX 4080 Founders Edition stayed in the low 60C range at stock. Our overclock includes higher fan speeds, resulting in even lower temperatures of 55C–58C. The fans only hit around 1300 RPMs at stock, but our more aggressive fan curve pushed that closer to 2000 RPM when overclocked.

Finally, we have our noise levels, which we gather using an SPL (sound pressure level) meter placed 10cm from the card, with the mic aimed right at the GPU fan(s). This helps minimize the impact of other noise sources, like the fans on the CPU cooler. The noise floor of our test environment and equipment is around 32 dB(A).

After running a demanding game (Metro, at appropriate settings that tax the GPU), the RTX 4080 settled in at a fan speed of 38% and a noise level of 42.6 dB(A). While it's not shown in the chart, our overclocked settings resulted in a 70% fan speed and 55.5 dB(A). Finally, with a static fan speed of 75%, the 4080 Founders Edition generated 57.4 dB(A) of noise.

RTX 4080 Additional Power, Clock, and Temperature Testing

Besides our Powenetics testing, we collect all of our frametime data using Nvidia's FrameView, which also logs clock speeds, temperatures, power, and GPU utilization (and plenty of other data as well). We've verified with our Powenetics equipment that Nvidia's GPUs report power draw that's within a few percent of the actual power use, so the above results give a wider perspective on how the card runs.

As we've mentioned already, Nvidia's RTX 4090 and 4080 tend to come in below their maximum rated TBP in a lot of gaming workloads. That's especially true of our 1080p results, where the average power use for the card was just 221W. With a faster CPU like a Core i9-13900K or Ryzen 9 7950X, we'd likely see higher power use from the 4080, though it's likely even an overclocked 13900K wouldn't max out its power limit at 1080p.

In terms of efficiency, you can see the overall power, performance, etc., for thirteen of the games in our test suite. Overclocking increased performance at 4K by 7%, while power use went up by 7.7%. Normally, overclocking can really drop efficiency, but at least with the 4080 Founders Edition the card seems to handle a modest OC fairly well. 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

Nvidia's RTX 4080 Founders Edition has a lot going for it. The overall performance is great, and you can enable DLSS or some other upscaling algorithm for more demanding games. DLSS 3 holds plenty of future potential for when more games use it, and based on what we've seen from DLSS 2 over the past couple of years, we suspect plenty of games will at least give users the option to try Frame Generation.

The card is also a beast when it comes to cooling. In that sense, it's the polar opposite of the RTX 3080 Ti Founders Edition. Nvidia skimped on that card and saved the larger cooler for the 3090 and 3090 Ti, and it showed. With the 4080, Nvidia isn't making any excuses: The triple-slot cooler keeps the chip frosty, and the GDDR6X memory (rated for up to 24 Gbps, even though it runs at 22.4 Gbps) feels completely different from the first-generation GDDR6X chips used on the Ampere 3070 Ti and above.

There's so much to like about the RTX 4080 Founders Edition. There's just one problem: It's really expensive, arguably too expensive! That will be a big problem for a lot of gamers that might otherwise want the card. But all we have to do is look at what's happened so far with the RTX 4090 launch to see plenty of reasons for Nvidia to start prices on the 4080 well above $1,000.

Even though the price feels too high to us, especially in light of where the RTX 4090 sits and where AMD plans to launch the RX 7900 series, that doesn't mean everyone will refuse to pay the piper his exorbitant fees. RTX 4090 sells for over $2,000 right now, though admittedly, we lack data on how many people are actually paying that much for a graphics card. Hopefully, the early adopters have found what they needed while the rest of us wait for prices to adjust.

Put another way, the price feels like a holdover from the days when GPU mining was still viable. Ada Lovelace was created amid the worst and longest-lasting GPU shortage we've ever witnessed. The past six months have mostly erased those problems, but the 4080 is stuck with a price tag courtesy of the class of 2021.

Or maybe it's just that Nvidia and its partners have a lot of RTX 30-series GPUs they still need to sell, and they don't want to cause those prices to collapse. But if that's the case, why are RTX 3080 and 3090 cards going up in cost over the past month? I call shenanigans, and hopefully people are wise enough not to get sucked into the idea that a 3080 is worth anywhere near $800 in the current market.

Nvidia also pitches the 4080 as a replacement for the 3080 Ti, but that card was overpriced due to the aforementioned mining and supply 'perfect storm' of 2021 — and this is not a 4080 Ti. The result is a card with a name that suggests it's taking over the ostensibly $700 position occupied by the RTX 3080, only it costs 71% more. That sucks, and when Nvidia trots out an RTX 4070 Ti priced at $799 or whatever, that will be equally bad. GPUs over $1,000 used to be for the "more money than sense" Titan crowd; now they're being marketed as upper-high-end gaming cards.

Performance and power look good on their own, but the 4080 is a very poor value. On the other hand, if the RTX 4080 were an $800-$1000 card, we’d be a lot more excited … and then it would probably just sell out faster, and scalpers would end up charging $1200. So we get it, but it still stinks to see what used to be a $500–$700 model number turning into a $1200+ model. Note also that the AD102 chip is 61% larger, and the cost per die can scale quickly as size increases.

The market will ultimately decide how much people are willing to pay for the RTX 4080. Maybe Nvidia can't sell them to gamers at $1,199 or more, and short of some new cryptocoin popping up and repeating what Ethereum did, it's difficult to imagine any miners rushing out to buy 4080 cards. But Nvidia also has other aspirations, like the autonomous vehicles market and other deep learning use cases.

It still hurts to see the 4080 shooting for $1,199, though. If you can afford that, why not just go whole hog and buy an RTX 4090? Of course, we also need to see where AMD actually lands with the Radeon RX 7900 cards next month. If an RX 7900 XTX delivers better non-ray tracing performance than the 4080, and maybe the RX 7900 XT beats it as well, we'd like to think Nvidia and its partners will be forced to reconsider the pricing.

But we've been thinking that for almost two years, and it still hasn't happened. On the other hand, EVGA left the graphics card market, and looking at the RTX 4080 and RTX 4090 launches it's not difficult to find reasons to do so. Nvidia launched the RTX 3090 Ti at $1,999, an obscene price that was again influenced by cryptominers and which it couldn't maintain. When sales slumped, Nvidia slashed the price of its RTX 3090 Ti Founders Edition to $1,099 at Best Buy and its AIB partners were forced to respond. Somehow we doubt Nvidia was offering refunds for unsold 3080 and 3090 cards that could no longer command their formerly extreme prices.

The old cliche about no bad products, just bad prices, rings true once more. Even though the hardware and technology are good, $1,200 feels like too much. If the 4080 drops below $1,000, it would be much nicer; even better would be getting closer to the official RTX 30-series launch prices. It's difficult to imagine people lining up to spend this much for the "one big step down from the top" GPU, but we'll have to wait and see what happens.

And frankly, that's really what we think most people should do: Wait and see what happens next month. Few people need to go out and upgrade to a newer, faster GPU absolutely right this minute. Being first to market with an extremely powerful new architecture gives some advantages and some disadvantages. AMD has seen Nvidia's hand and can now decide how it wants to proceed.

But if you prefer Nvidia cards, or if you're interested in AI research or professional workloads like 3D rendering, $1,200 might not feel too bad at all. Admittedly, it's a lot for an awful lot for a gaming GPU, but it's downright affordable compared to professional GPUs that can cost over $5,000.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

As the availability date of Nvidia's GeForce RTX 4080 approaches, makers of add-in-boards have published specifications for their devices, revealing some interesting details. For example, this week Colorful revealed the final specs of its iGame GeForce RTX 4080 16GB Neptune OC graphics card with a high-end closed loop cooling system and total graphics power capped at 'just' 380W. 

Colorful's iGame GeForce RTX 4080 16GB Neptune OC carries Nvidia's AD103 graphics processor with 9728 CUDA cores paired with 16GB of 22.4 GT/s GDDR6X using a 256-bit interface. Once the out-of-box single-click overclocking feature is applied, the GPU can boost to 2640 MHz, about 5% faster than Nvidia's recommended boost clock of 2520 MHz. However, makers of add-in-boards have not yet enabled truly high clocks for Ada Lovelace architecture cards out-of-box. In any case, AD103 AIBs will likely join the ranks of the best graphics cards for gaming this year. 

Colorful equips its GeForce RTX 4080 Neptune OC graphics card with a 22+4-phase voltage regulating module (VRM), potentially more advanced than the 18+3-phase power circuitry Nvidia used for its GeForce RTX 3090 as well as GeForce RTX 3090 Ti Founders Edition boards rated for 350W and 450W, respectively. With a VRM like this, overclocking potential of Colorful's GeForce RTX 4080 Neptune promises to be significant. 

To ensure proper cooling of its top-of-the-line AD103-based offering, the manufacturer equips the iGame GeForce RTX 4080 Neptune OC card with its capable closed-loop cooling system that it also uses for its current range-topping iGame GeForce RTX 4090 Neptune OC card, which has a thermal graphics power of 630W, notes VideoCardz. But there is an interesting tidbit about the TGP of the iGame GeForce RTX 4080 16GB Neptune OC. 

TGP of the RTX 4080 Neptune OC is capped at 380W, which is a ~19% increase over the 320W recommended by Nvidia. 380W is still an extremely high TGP (higher than AMD's Radeon RX 7900 XTX), but Colorful limiting its AD103-based offering rather conservatively is noteworthy. Perhaps, the AD103 overclocks very well even without a significant increase in power consumption, or maybe Colorful just wants to play it relatively safe with this graphics chip. 

Colorful could start selling its iGame GeForce RTX 4080 16GB Neptune OC on November 16, which is when other AD103-powered boards will hit the market. The pricing of the board is unclear, but since it is a factory-overclocked product equipped with an advanced VRM and an ultra-capable cooling system, do not expect to buy it at Nvidia's MSRP of $1199.  

Nvidia's GeForce RTX 4080 graphics card with 16GB of memory is a little more than two weeks away, so it is about time for manufacturers of add-in-boards (AIBs) to start sharing images of their products. TechPowerUp and VideoCardz published the first photos of GeForce RTX 4080 cards from Gigabyte and MSI. For avid readers who are wondering about the melting issue, we can confirm your worst 16-pin fears.  

Nvidia's GeForce RTX 4080 graphics card uses the AD103 graphics processing unit based on the Ada Lovelace architecture with 9728 CUDA cores connected to 16GB of 22.4 GT/s GDDR6X memory on a 256-bit interface. The product, which could join the ranks of the best graphics cards for gaming, is expected to consume up to 320W of power. That figure is in line with the consumption of Nvidia's previous-generation GeForce RTX 3080 graphics board. So expect these units to come with huge coolers.  

Gigabyte's GeForce RTX 4080 Eagle graphics card comes with a monstrous 3.5-wide triple-fan cooler that is also rather 'tall.'. VideoCardz does not disclose the graphics card clocks, though it definitely will not be clocked lower than Nvidia's recommended 2210 MHz base and 2505 MHz boost clocks. The board will use a 12VHPWR (12+4-pin power connector), though we do not know whether Gigabyte will bundle Nvidia's 16-pin adapter with the product. 

MSI also published some images of its six GeForce RTX 4080 products itself. The company's RTX 4080 lineup will consist of six models, including the GeForce RTX 4080 16GB Ventus 3X, GeForce RTX 4080 16GB Ventus 3X OC, GeForce RTX 4080 16GB Suprim, GeForce RTX 4080 16GB Suprim X, GeForce RTX 4080 16GB Gaming Trio, and GeForce RTX 4080 16GB Gaming X Trio. All boards will feature gargantuan dimensions of up to 337 x 140 x 67 mm as they will feature a triple-wide, triple-fan cooler. Also, they will come with a 12VHPWR power connector. 

By today's standards, a 320W thermal graphics power (TGP) is not shocking, considering that flagship boards are designed for 350W – 450W these days. But 320W is still a lot of power to dissipate, so all GeForce RTX 4080 boards will be equipped with a large and sophisticated cooling system. To that end, most of these AIBs will not fit into Mini-ITX and Micro-ATX chassis. 

Some AIB vendors will likely come up with hybrid cooling systems that will be more compact than air cooling solutions, but to fit those boards into smaller chassis, one will probably need to use an AIB with a smaller PCB and a custom liquid cooling system.

Hardware leaker Harukaze5719 has shared the alleged Ada Lovelace dies that will power Nvidia's next-generation RTX 40-series Mobile GPUs and their associated power targets. Most interestingly, the leak claims that Nvidia could prepare a GeForce RTX 4090 Mobile, the first time this model number has appeared in the mobile segment. But, of course, treat the leaked information with a grain of salt.

The leak (via REHWK) reveals five GeForce RTX 40-series Mobile GPUs in total, including the GeForce RTX 4090 Mobile, GeForce RTX 4080 Ti Mobile, GeForce RTX 4070 Mobile, GeForce RTX 4060 Ti Mobile, and GeForce RTX 4060 Mobile. In addition, the mobile variant of the GeForce RTX 4090 will reportedly utilize the AD103 die, while the GeForce RTX 4080 Ti will rely on the AD104 die, whereas the AD106 silicon will serve the GeForce RTX 4070 Mobile. Finally, the GeForce RTX 4060 Mobile models could sport the AD107 die.

Power consumption for the GeForce RTX 4090 Mobile and GeForce RTX4080 Ti Mobile allegedly peak at 175W, and the GeForce RTX 4070 Ti Mobile tops out at 140W. Unfortunately, the leaker didn't provide the TDP for the GeForce RTX 4060 Mobile and GeForce RTX 4060 Ti Mobile SKUs. If these figures are accurate, the GeForce RTX 4090 should be a massive upgrade in the mobile space, providing desktop GeForce RTX 3090 or GeForce RTX 3080-like performance at just 175W. To keep performance up at these lower TDPs, Nvidia may have to unlock all cores on the AD103 for maximum performance and efficiency compared to the desktop GeForce RTX 4080 16GB, which will come with a slightly neutered AD103 die.

As for the GeForce RTX 4080 Ti Mobile, we should expect this GPU to be a significant downgrade in performance compared to the GeForce RTX 4090 Mobile due to the rumored die change from AD103 to AD104. Since AD104, with all cores enabled, features a 40% reduction in core count, essentially, the GeForce RTX 4080 Ti Mobile will be the mobile variant of the now deleted GeForce RTX 4080 12GB SKU. However, performance will be dictated heavily by clock speed in the mobile segment since clock speeds have the chance to fluctuate significantly due to the considerably lower TDPs required on mobile SKUs. With a 40% reduction in core count on AD103, there's a likely chance we'll see far higher clock speeds on the GeForce RTX 4080 Ti Mobile compared to the GeForce RTX 4090 Mobile.

GeForce RTX 40-Series Mobile Specifications*

*Specifications are unconfirmed.

Unfortunately, we don't have enough details on AD106 or AD107 to know how fast or slow the GeForce RTX 4060 (Ti) Mobile and GeForce RTX 4070 Mobile will be. But we expect all to pack a noticeable performance upgrade over their GeForce RTX 30-series counterparts. What is interesting, though, is the potential introduction of the GeForce RTX 4060 Ti Mobile so early in the mobile pipeline, with both the GeForce RTX 4060 Mobile and Ti variant GPUs using the entry-level AD107 die instead of AD106. It differs from 30 series models, where the xx60 series cards operated on the GA106 die.

It would suggest that AD107 will be an exceptionally powerful die on Ada Lovelace. Still, we suspect this die change is only due to the GeForce RTX 4090 Mobile being introduced to the product stack, forcing other models to go down a notch in core count and GPU dies.

Overall, Ada Lovelace mobile is looking very strong based on the power consumption targets and die models suggested in the leaks. GeForce RTX 4070 Mobile and higher SKUs will feature noticeably boosted power targets over GeForce RTX 30-series GPUs, with the GeForce RTX 4090 Mobile having the possibility to match the desktop GeForce RTX 3090 or GeForce RTX 3080 gaming performance. But we'll have to wait and see the spacing between these GPUs once Nvidia releases them. Nvidia's GeForce RTX 40-series Mobile GPUs still have a 50% effective nerf to power targets over RTX 30-series desktop GPUs, which can hugely impact performance. It's all down to Ada Lovelace's efficiency on the TSMC 4n node at these "very low" power targets.

Security researcher Sam Croley took to Twitter to share just how incredible Nvidia's new RTX 4090 really is... at cracking passwords. It turns out it's twice as fast as the previous leader, the RTX 3090, at breaking one of your passwords — even when faced off against Microsoft's New Technology LAN Manager (NTLM) authentication protocol and the Bcrypt password-hacking function. 

Essentially, this means that any wealthy gamer sporting the RTX 4090 can crack an average password in a matter of days — and that's if you follow good password-setting practices (and most of us definitely don't).  

The benchmark, HashCat V.6.2.6., is a renowned password-cracking tool that lays best in the hands of system administrators and cybersecurity professionals (of which Croley was a core programmer, by the way). It allows researchers to test or guess user passwords in the few situations that might require it. 

Unfortunately, this means that cybercriminals can do it, too. And with the evolution in graphical user interfaces (GUIs) and the ease of use of these programs in modern computers sporting a high-performance graphics card, it's become easier than ever to deploy these tools.

In testing, the RTX 4090 trumps the RTX 3090 in almost every algorithm with almost doubled performance — which isn't that shocking, even if that still represents a higher performance improvement than we see in the RTX 4090's graphics performance. This is likely the result of Nvidia still investing a lot of its graphics chip design development to increase its performance on the data-center side. The RTX 4090 shone across the several attack types provided in the HashCat software: dictionary attacks, combinator attacks, mask attacks, rule-based attacks, and brute force attacks.

The researchers estimate that a purpose-built password hashing rig (pairing eight RTX 4090 GPUs) could crack an eight-character password in 48 minutes. According to Statista and from 2017 data, 8-character passwords are the most common among leaked passwords, commanding a 32% share of them. This doesn't mean that they're the least safe; it just very likely means that it's the most common password character length. And they can now be taken out in under an hour by a "specialized" hashing rig.

Of course, that assumes that the password is as least eight characters long and that it follows the required conventions (at least one number and a special character included). When HashCat is driven to test the most commonly used passwords, however, it can bring a theoretical 48 minute cracking operation that attempted all 200 billion possible combinations down to the millisecond range. But then, that was to be expected: even a human would be extremely fast in cracking a password such as "123456" — apparently the most common password of 2021.

Another interesting element to note is that password cracking naturally has an associated cost; investing in a $1,600 RTX 4090 is costly, and each attempt at cracking a password will incur in power costs as well. So it's not just a matter of will. What the RTX 4090 does is bring down the cost to actually crack passwords — something that happens as long as more powerful GPUs come out while security algorithms remain relatively static. Jacob Egner has an extremely detailed and interesting analysis on his blogpost detailing his discoveries on the $/hash ratios. 

Of course, another chip on cybersecurity's shoulder is the amount of data that needs to be encrypted against the inexorable development of quantum computing — computers that will render almost all currently-used encryption schemes pedestrian. Looking at the cost decreases in password-cracking just with GPUs, however, it seems that current security should be upgraded to newer, post-quantum algorithms sooner rather than later.

Relax — not every RTX 4090 owner will turn their top-tier graphics card towards a password-cracking pastime. Additionally, the password-cracking ease of tools such as HashCat are usually deployed against offline assets, not online ones. This means that the chances of your PC being the target of a deranged RTX 4090-owner cracking passwords at will are slim — so slim they're almost nonexistent. 

Yet, in light of this, perhaps it's still a good idea to brush up on online security best practices, starting with storing lengthier passwords in one of the best password managers.

YouTube channel Benchmark Lab has decided to test Nvidia's new GeForce RTX 4090 in a unique fashion — by pitting it against two RTX 3090s in SLI. 

That's right, SLI — Nvidia's obsolete multi-GPU tech that Benchmark Labs managed to run using some unknown SLI-enablement magic. The channel was able to run SLI decently well in several DX11, DX12, and Vulkan titles, but ultimately the 3090s couldn't beat the RTX 4090 (though they got close). 

Before we jump to the results, it's worth noting that all of this should be taken with a grain (or several) of salt. SLI has been unsupported for years at this point, and getting it to work with most modern titles — especially with DX12 and Vulkan — would require...unorthodox methods. We're not sure how Benchmark Lab managed to enable SLI in most of the games tested — again, specifically the DX12 versions — or if the results were doctored or extrapolated. That said, the results are pretty interesting (if they are, indeed, accurate). 

Benchmark Lab tested 9 games: Spider-Man Remastered, Cyberpunk 2077, Watch Dogs Legion, Microsoft Flight Simulator 2020, Minecraft RTX, God of War, Witcher 3, Horizon Zero Dawn, and Read Dead Redemption 2. Testing was performed at 4k resolution using a variety of different quality settings and DLSS modes. 

In Spider-Man Remastered, with maxed out settings (Ray Tracing and DLSS Balanced mode enabled), the RTX 3090s in SLI achieved 80 - 85 frames per second (fps) on average, while the RTX 4090 managed a noticeably higher 95 fps average. 

In Cyberpunk 2077, with maxed out settings (Ray Tracing and DLSS Balanced mode enabled), the RTX 3090s in SLI managed 50 fps while the RTX 4090 averaged 70 fps.

In Watch Dogs: Legion, with maxed out settings (Ray Tracing and DLSS Quality mode enabled) the gap closed a little: the RTX 3090s in SLI averaged 73 fps, while the RTX 4090 averaged 80 fps.

The results were even closer in Microsoft Flight Simulator with maxed out settings (Ray Tracing and DLSS Quality mode enabled). The RTX 3090s in SLI averaged 80 fps, while the RTX 4090 averaged 83 - 85 fps. (However, note that Microsoft Flight Simulator is known to be CPU-bound, especially at higher frame rates.)

In Minecraft RTX, with max settings (Ray Tracing enabled), the RTX 3090s in SLI averaged 70 fps while the RTX 4090 hit a slightly higher average of 75 fps. And in God of War at max settings (DLSS Quality mode enabled), the RTX 3090s in SLI hit an average of 103 fps while the RTX 4090 did quite a bit better with an average of 120 fps.

For the rest of the titles, check out the video below: 

RTX 3090 SLI Is The Only Setup That Can Get Close To The RTX 4090

With the exception of the Cyberpunk 2077 results, the RTX 3090s in SLI were around 8 - 15% slower than the RTX 4090. For an SLI implementation, this isn't bad at all — and it's the only GPU configuration we've seen so far that can get close to the RTX 4090's performance bracket. For some perspective, the next-closest GPU is the RTX 3090 Ti — which trails the 4090 by over 50% based on our tests. 

Sadly, the 3090s in SLI setup wasn't being used to its full potential, as the secondary 3090 consistently maxed out at 45% utilization. This is one of the pitfalls of SLI setups, in which bad optimization leads to less-than-superb GPU utilization on the secondary card. In theory, if we could get anywhere close to 100% utilization on both GPUs, we would probably see the 3090s in SLI outperform the 4090 by a good margin. 

But the chances of that happening are extremely unlikely, since Nvidia officially killed off SLI in the form of physical bridges with the RTX 40 series. SLI support has dwindled so much over the past several years that it's basically only useful in synthetic benchmarks at this point. 

There are technically ways of enabling SLI (unofficially) in games that don't support it — like in the case of the benchmarks seen in this video. But these results are unpredictable at best, and usually result in system instability or severe micro-stuttering problems while gaming. 

There is some hope for SLI, however. Multi-GPU workloads are very common in the enterprise space, and Nvidia even has multi-GPU technologies that don't require an NVLink or SLI bridge. 

On the gaming side, modern APIs such as DX12 and Vulkan do have the capability to render frames to two completely different GPUs in tandem. So there's a chance multi-GPU tech might make its way back to the gaming space, eventually (but whether developers want to support multi-GPU technology for gaming is a different story). 

Galax-sponsored TecLab OC team from Brazil has managed to push Nvidia's GeForce RTX 4090 graphics processor to a record-breaking 3.45 GHz with the help of some modding and liquid nitrogen. 

This is the highest GPU clock speed ever achieved. 

But there is a catch. At such a high frequency the card started showing performance degradation — and the team had to reduce the clock speed before they were able to run benchmarks.

According to VideoCardz, the team of overclockers took Galax's upcoming GeForce RTX 4090 SG (Serious Gaming) graphics card — a likely contender for one of the best graphics cards — which uses a custom-printed circuit board that is designed to make graphics processors run at frequencies higher than those recommended by Nvidia. 

Using the card's quad-fan stock cooler, the Galax GeForce RTX 4090 SG can be overclocked to 3.075 GHz. But with higher voltages and extreme LN2 cooling, the GPU can hit speeds as high as 3.45 GHz, with memory data transfer rates hitting speeds of up to 24 GT/s (probably the default speed of Micron's latest 16GB GDDR6X chips). 

To prep the card for their overclocking session, the TecLab OC team had to remove the power limit controller, resolder some of the elements from the front side of the PCB to the backside of the PCB to install the liquid nitrogen pot, and add voltage probes to better understand the board's behavior. 

The TecLab OC team managed to get the Galax GeForce RTX 4090 SG to 3.45 GHz, which is the highest GPU clock speed ever achieved. However, at such a high frequency, and with liquid nitrogen, the GPU started to show degraded performance, and the team had to reduce the clock speed to 3.27 - 3.36 GHz to run benchmarks on the card. 

Even overclocked to ~3.3 GHz, the GeForce RTX 4090 still managed to break five benchmark records, achieving the world's highest performance scores in 3DMark Fire Strike Ultra, 3DMark Port Royal, 3DMark Time Spy Extreme Graphics, Superposition 1080p Extreme, and Unigine Superposition 8.

When Nvidia introduced its AD102 graphics processor as well as its GeForce RTX 4090 Founders Edition graphics card, it said the GPU was designed to hit high clock speeds and claimed to have already run the AD102 GPU at 3.0 GHz in its labs.  And now the TecLab OC team has proven the GPU can hit even higher clock speeds outside of Nvidia's labs — once it's fed with enough power and cooled with liquid nitrogen, of course. 

Nvidia has posted its first official performance numbers for its GeForce RTX 4080 12GB and GeForce RTX 4080 16GB graphics cards. The version with 16GB of memory appears to be 21% – 30% faster than its 12GB counterpart. Meanwhile, Nvidia's previous generation flagship GeForce RTX 3090 Ti is still faster than the GeForce RTX 4080 12GB — at least without any DLSS enabled. 

Nvidia compared its latest GeForce RTX 40-series graphics cards (including the flagship GeForce RTX 4090) against each other and against its previous-generation GeForce RTX 30-series products in Microsoft Flight Simulator, A Plague Tale: Requiem, and F1 22. 

The comparison focused primarily on performance advantages provided by the company's latest DLSS 3 technology (with optical multi frame generation, resolution upscaling, and Nvidia Reflex), but the Nvidia also included performance numbers without any enhancements. Furthermore, it included its first official GeForce RTX 4080 12GB and GeForce RTX 4080 16GB performance comparison. 

Traditionally, Microsoft's Flight Simulator games are tough on both CPUs and GPUs and the latest version is no exception. Both DLSS 2 and DLSS 3 bring significant performance uplift to the game, but so does Nvidia's new Ada Lovelace GPU architecture (perhaps courtesy of enhanced ray tracing performance). 

With DLSS 3 on, the GeForce RTX 4080 16GB is 21% faster than the GeForce RTX 4080 12GB; without DLSS 3 the 16GB version is 26% faster than the 12GB. The GeForce RTX 4080 12GB offers the same performance as the GeForce RTX 3090 Ti 24GB without DLSS 3 — not bad at all, and this is still one of the best gaming graphics cards today. 

For some reason Nvidia decided to test A Plague Tale: Requiem without enabling ray tracing, but with DLSS technologies enabled. DLSS 3 brings even more tangible performance enhancements than DLSS 2, so all GeForce RTX 40-series boards are substantially faster than their predecessors. 

In this test, the GeForce RTX 4080 16GB is 21% faster than the GeForce RTX 4080 12GB with DLSS 3 enabled, and is 30% faster with DLSS 3 disabled. Meanwhile, the GeForce RTX 3090 Ti is 19% faster than the GeForce RTX 4080 12GB without DLSS enabled. Also, the new GeForce RTX 4080 12GB is only 4 frames per second (fps), or about 12%, faster than the original GeForce RTX 3080 in this test. 

In F1 22 you want to get all the framerates you can possibly have, so both DLSS 2 and DLSS 3 are helpful. The new version is somewhat more efficient with its optical multi frame generation, and Ada Lovelace boards are clearly faster than Ampere boards in this title. 

With DLSS 3 enabled, the GeForce RTX 4080 16GB is 27% faster than its 12GB sibling, and is 24% faster without DLSS 3 enabled. Also, the GeForce RTX 3090 Ti turns out to be 7% faster than the GeForce RTX 4080 12GB without DLSS enabled. Still, the GeForce RTX 4080 12GB is consistently faster than the original GeForce RTX 3080. 

While Nvidia's GeForce RTX 4080 12GB and GeForce RTX 4080 16GB carry the same model number, they are based on completely different graphics processors (AD104 vs. AD103) and feature different CUDA core counts (7680 vs. 9728). When combined with distinctive memory bandwidth (504 GB/s vs. 717 GB/s), the two models offer very different performance levels than you would typically expect from graphics boards carrying the same model number. 

For an easier comparison, we've combined the relative performance differences between Nvidia's latest desktop GPUs into a table.

Performance of GeForce RTX 40 and GeForce 30-Series Graphics Cards (without DLSS)

The Nvidia GeForce RTX 4090 hype train has been building for most of 2022. After more than a year of extreme GPU prices and shortages, CEO Jensen Huang revealed key details at GTC 2022, with a price sure to make many cry out in despair. $1,599 for the top offering from Nvidia's Ada Lovelace architecture? Actually, that's only $100 more than the RTX 3090 at launch, and if the card can come anywhere near Nvidia's claims of 2x–4x the performance of an RTX 3090 Ti, there will undoubtedly be people willing to pay it. The RTX 4090 now sits atop the GPU benchmarks hierarchy throne, at least at 1440p and 4K. For anyone who's after the fastest possible GPU, never mind the price, it now ranks among the best graphics cards.

That's not to say the RTX 4090 represents a good value, though that can get a bit subjective. Looking just at the FPS delivered by the various GPUs per dollar spent, it ranks dead last out of 68 GPUs from the past decade. Except our standard ranking uses 1080p ultra performance, and the 4090 most decidedly is not a card designed to excel at 1080p. In fact, it's so fast that CPU bottlenecks are still a concern even when gaming at 1440p ultra. Look at 4K performance and factor in ray tracing, and you could argue it's possibly one of the best values — see what we mean about value being subjective?

Again, you'll pay dearly for the privilege of owning an RTX 4090 card, as the base model RTX 4090 Founders Edition costs $1,599 and partner cards can push the price up to $1,999. But for those who want the best, or anyone with deep enough pockets that $2,000 isn't a huge deal, this is the card you'll want to get right now, and we'd be surprised to see anything surpass it in this generation, short of a future RTX 4090 Ti. 

Here's a look at the who's who of the extreme performance graphics card world, with the fastest cards from Nvidia, AMD, and now Intel. Obviously, Intel's Arc A770 competes on a completely different playing field, but it's still interesting to show how it stacks up on paper.

We're going to simply refer you to our Nvidia Ada Lovelace Architectural deep dive if you want to learn about all the new technologies and changes made with the RTX 40-series. The above specs table tells a lot of what you need to know. Transistor counts have nearly tripled compared to Ampere; core counts on the RTX 4090 are 52% higher than the RTX 3090 Ti; GPU clock speeds are 35% faster, and the GDDR6X memory? It's still mostly unchanged, except there's now 12x more L2 cache to keep the GPU from having to request data from memory as often.

On paper, that gives the RTX 4090 just over double the compute performance of the RTX 3090 Ti, and there are definitely workloads where you'll see exactly those sorts of gains. But under the hood, there are other changes that can further widen the gap.

Ray tracing once again gets a big emphasis, and three new technologies — Shader Execution Reordering (SER), Opacity Micro-Maps (OMM) and Displaced Micro-Meshes (DMM) — all offer potential improvements. However, they also require developers to use them, which means existing games and engines won't benefit.

Deep learning and AI workloads also stand to see massive generational improvements. Ada includes the FP8 Transformer Engine from Hopper H100, along with FP8 number format support. That means double the compute per Tensor core, for algorithms that can use FP8 instead of FP16, and up to four times the number-crunching prowess of the 3090 Ti.

One algorithm that can utilize the new Tensor cores — along with an improved Optical Flow Accelerator (OFA) — is DLSS 3. In fact, DLSS 3 requires an RTX 40-series graphics card, so earlier RTX cards won't benefit. What does DLSS 3 do? It takes the current and previously rendered frames and generates an extra in-between frame to fill the gap. In some cases, it can nearly double the performance of DLSS 2. We'll take a closer look at DLSS 3 later in this review.

From a professional perspective, particularly for anyone interested in deep learning, you can easily justify the cost of the RTX 4090 — time is money, and doubling or quadrupling throughput will definitely save time. Content creators will find a lot to like and it's a quick and easy upgrade from a 3090 or 3090 Ti to the 4090. We'll look at ProViz performance as well.

But what about gamers? Unlike the RTX 3090 and 3090 Ti, Nvidia isn't going on about how the RTX 4090 is designed for professionals. Yes, it will work great for such people, but it's also part of the GeForce family, and Nvidia isn't holding back on its gaming performance claims and comparisons. Maybe the past two years of cryptocurrency mining are to blame, though GPU mining is now unprofitable so at least gamers won't have to fight miners for cards this round. 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

Nvidia's GeForce RTX 4090 Founders Edition superficially looks nearly the same as the previous RTX 3090 Founders Edition. However, there have been a few changes. For one, the new card is slightly thicker and not quite as long — it measures 304x137x61mm compared to the 3090's 313x138x57mm. Weight is virtually unchanged at 2186g, compared to 2189g on the previous generation.

Cosmetically, the new RTX 4090 has a slightly concave shape for the outer chassis frame. You can see the difference in the gallery below where we've put the 3090 and 4090 cards side by side. Otherwise, the changes are slightly more functional rather than purely aesthetic.

There are still two fans: one sucking air through the heatsink at the rear of the card and the other pushing air through the fins and out the exhaust port around the video connectors. The new fans are 115mm in diameter compared to 110mm on the previous generation, and the fans also have a slightly smaller central hub as well (38mm versus 42mm). Nvidia says the fans improve airflow by 20%, presumably at the same speed.

The power situation has also changed in the past two years. Nvidia pioneered the 12-pin power connector with the RTX 30-series. It's pin compatible with the new PCIe 5.0 12VHPWR 16-pin connector, just minus the extra four sense pins. The 4090 16-pin connector also lies parallel to the PCB, where the 3090 12-pin connector jutted out perpendicular to the board.

Finally, where the RTX 3090 had a dual 8-pin-to-12-pin adapter included in the box, the RTX 4090 has a quad 8-pin-to-16-pin adapter — and the same adapter seems to be a standard part of the kit for other RTX 4090 models from Nvidia's partners.

In terms of things that definitely haven't changed, the RTX 4090 still comes with three DisplayPort 1.4a outputs and a single HDMI 2.1 output. We still haven't seen any DisplayPort 2.0 monitors, which might be part of the reason for not upgrading the outputs, but it's still a bit odd as the DP 2.0 standard has been finalized since 2019.

Similarly, the PCIe x16 slot sticks with the PCIe 4.0 standard rather than upgrading to PCIe 5.0. That's probably not a big deal, especially since multi-GPU support in games has all but vanished. We'd argue that the frame generator and DLSS 3 also make multi-GPU basically unnecessary. To that end, Nvidia has also removed NVLink support from the AD102 GPU and the RTX 40-series cards. Yes, we can effectively declare that SLI is now dead, or at the very least it's lying dormant in a coma. 

  Test Setup for GeForce RTX 4090 

We updated our GPU test PC and gaming suite in early 2022, and we continue to use the same hardware for the time being. AMD's Ryzen 9 7950X might be slightly faster, though we're using XMP for a modest boost to performance and we're not too concerned (yet) about a few percent higher frame rates. But if you're actually going to go out and buy an RTX 4090, you'll definitely want at least a Core i9-12900K like we're using, if not the upcoming  Core i9-13900K or the aforementioned 7950X.

Our CPU sits in an MSI Pro Z690-A DDR4 WiFi motherboard, with DDR4-3600 memory — a nod to sensibility rather than outright maximum performance. We have a Crucial P5 Plus 2TB SSD, which ought to feel capacious but with the size of modern games it's beginning to seem just adequate. We also upgraded to Windows 11 and are now running the latest 22H2 version (with VBS and HVCI disabled) to ensure we get the most out of Alder Lake. You can see the rest of the hardware in the boxout.

Our gaming tests consist of a "standard" suite of eight games without ray tracing enabled (even if the game supports it), and a separate "ray tracing" suite of six games that all use multiple RT effects. We've tested the RTX 4090 at all of our normal settings, though the focus here will be squarely on 4K performance, with 1440p as a step down for those that might want one of the upcoming 1440p 360 Hz displays over a 4K 144 Hz panel.

Testing at 1080p was conducted, but it's almost entirely meaningless data unless you're running a game that supports ray tracing. Most games were already largely CPU limited at 1080p with the RTX 3090 Ti and RX 6950 XT. The RTX 4090 moves the needle so far to the right that there's less than a 5% difference in overall performance between 1080p ultra and 1440p ultra. We also enabled and tested the 4090 with DLSS 2 Quality mode in the ten games that support it.

Besides the gaming tests, we also have a collection of professional and content-creation benchmarks that can leverage the GPU. We're using SPECviewperf 2020 v3, Blender 3.30, OTOY OctaneBenchmark, and V-Ray Benchmark.

Finally, because the Ada Lovelace and RTX 40-series GPUs provide a lot of new features, we'll be looking at DLSS 3 in preview builds of several games provided by Nvidia. DLSS 3 only runs on the RTX 4090 for now, so we won't have direct comparisons to other GPUs. We'll also look at performance — and latency on Nvidia GPUs — across a collection of high-end graphics cards using the preview build of Cyberpunk 2077. (Nvidia GPUs are required for Reflex support, which in turn is required for Nvidia's FrameView utility to capture latency data.) 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

We'll start with the performance that matters most: 4K at maxed out settings. Again, if you're using a 1080p monitor — even one with an extreme refresh rate — the RTX 4090 will almost certainly be overkill. That's less true if you're playing games with lots of ray tracing effects, but we'll get to that in a moment.

Let's pause for a moment while you check through those charts — there are nine in total, so don't just look at the overall average! But if you do just check the average, you'll see that the RTX 4090 provides a massive 55% improvement over the RTX 3090 Ti that launched six months ago. If you just bought a 3090 Ti, this one's going to sting more than a little. Yeah, you might have saved $500, but if you're in the market for a GPU that costs $1,000 or more, we think this level of performance at least justifies the cost.

Looking at other GPUs, there's a 71% generational uplift compared to the RTX 3090 Founders Edition, and a slightly larger 77% improvement over the RTX 3080 Ti. Against AMD's best, the RX 6950 XT, in our standard gaming suite that doesn't tap into most of the Ada Lovelace architecture's extra features, you're still getting a 64% improvement. That's huge and might put Nvidia's raw performance out of reach for AMD's upcoming RDNA 3 GPUs. We'll have to wait a month or so to see where AMD lands, but this is Nvidia throwing down the gauntlet.

It's also important to note that the 55% average includes games that are still hitting CPU bottlenecks, like Flight Simulator. The RTX 4090 barely drops in performance going from 1080p ultra to 1440p ultra to 4K ultra — which is part of why Nvidia's DLSS 3 Frame Generation technology is so exciting, but we'll get to that in a bit.

In the eight individual test results, the RTX 4090 beats the 3090 Ti by anywhere from 11% (Flight Simulator) to 112% (Total War: Warhammer 3). Those are the two outliers, with the remaining six games falling in a tighter range of 46% (Far Cry 6, Red Dead Redemption 2) to 70% (Forza Horizon 5).

It's also worth noting what DLSS 2 in Quality mode does for performance in the four games that support it. Flight Simulator performance drops 4%, again due to the CPU limited nature. Horizon Zero Dawn only gains 10%, Watch Dogs Legion gets a 13% boost, and Red Dead Redemption performance improves by 14%. The RTX 3090 Ti saw up to a 35% increase in performance with DLSS 2 Quality mode, so again there are clear CPU bottlenecks coming into play, even at 4K.

But this isn't a card primarily designed for standard rasterization rendering techniques. It can easily beat the previous generation cards in such games, but where it's really destined to shine is with significant levels of ray tracing.

If the standard gaming performance didn't spark your interest, surely the ray tracing results will. Now we're looking at a 78% improvement over the previous gen RTX 3090 Ti, and this is in existing games. More demanding ray tracing titles are on the way, though there's a good chance you'll need an RTX 40-series with DLSS 3 to get decent performance out of those games.

Looking at other GPUs, the 4090 outpaces AMD's top RX 6950 XT by 190% — as in, it's nearly three times faster. It's just over twice as fast as the RTX 3090 and 3080 Ti as well. And if you turn on DLSS Quality mode upscaling, which could be done on any of the RTX cards, the 4090 performance improves by 78%. RTX 4090 running the highest quality mode of DLSS thus ends up with nearly five times the performance of the 6950 XT in demanding DXR games.

The individual gaming charts as usual show a range of performance, though it's not nearly as wide as in our standard rasterization benchmarks. That's because most ray tracing games will fully tax the GPU, particularly at 4K. Across our six games, the 4090 advantage ranged from 67% in Control Ultimate Edition — which also happens to be the oldest of the games in our DXR suite — up to a 106% advantage in Fortnite with all the RT options maxed out.

Even without DLSS, the RTX 4090 almost manages to deliver native 4K ultra ray tracing performance of 60 fps or more. Certainly the performance is playable, at 40 fps or higher in all of the games. DLSS 2 Quality upscaling pushes that to 100 fps on average, and all of the DXR tests are now well above 60 fps. Again, we'll look at what DLSS 3 can do in some preview builds of games in just a moment. 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

 GeForce RTX 4090: 1440p Gaming Performance

The RTX 4090 makes 1440p ultra the new 1080p ultra. Or in other words, CPU and other system bottlenecks now become much more of a factor even at 1440p. The 4090 still holds onto an average lead of 28% compared to the 3090 Ti, and 25% over the RX 6950 XT, but clearly it's being held back. The DLSS results also prove this, where performance at best matches the native result and at worst drops frame rates by 5%.

In the individual game charts, Flight Simulator clearly isn't performing as well as it could — it's 10% slower on the 4090 than on the 3090 Ti. Normally, CPU bottlenecks would apply equally to all GPUs from the same vendor, but since Ada is a new architecture it likely requires some changes to the drivers, and at least in this one instance the Ada pathway isn't as optimized as the older Ampere architecture pathway.

That's the only game where performance is worse on the 4090 than on the 3090 Ti, but Far Cry 6 and Horizon Zero Dawn are now down into single-digit percentage advantages, and Watch Dogs Legion only averages a 23% improvement.

Ray tracing games on the other hand don't really mind the drop to 1440p, at least if you're not using DLSS. The RTX 4090 still outpaces the 3090 Ti by 69%, so only a 9% drop compared to 4K. The gains from DLSS 2 Quality mode shrink a lot, however, adding just 43% to the overall performance — that's a 35% drop compared to 4K.

Other GPUs continue to lag significantly behind, with the 4090 beating the RX 6950 XT by 160%, the 3090 by 90%, and the 3080 Ti by 95%. Remember last year when the RTX 3080 Ti launched at $1,199 and immediately sold out? Now you're able to get double the performance for 33% more money, at least in heavy DXR games. 

GeForce RTX 4090: 1080p Gaming Performance

The standard rasterization results at 1080p ultra are mostly to show that, yes, some games still show modest scaling of performance while others are clearly CPU bottlenecked. The RTX 4090 now only leads the RTX 3090 Ti by 16% on average, and a couple of games (Far Cry 6 and Flight Simulator) run faster on the previous generation card.

AMD's RX 6950 XT gains some ground as well and now only trails the 4090 by 9%, which is sort of interesting. Both the 6950 XT and the 4090 have lots of cache, which previously helped AMD's performance a lot more at 1080p than at 4K. Nvidia's caching architecture is of course different, and both GPUs are hitting CPU bottlenecks. But if we further reduce the workload to 1080p medium (not shown here), the 6950 XT ends up with a slight 1% lead overall.

The point being: Don't buy an RTX 4090 if you only plan to game at 1080p. Seriously, just don't, even if you have a 360 Hz gaming monitor, you're likely running reduced quality settings to get framerates up, and at that point, the 4090 won't offer much more than a previous generation card.

Ray tracing again shows reduced gains at 1080p, though the 4090 is still 59% faster than the 3090 Ti overall. That's in the most demanding titles currently available, though some are clearly not quite as demanding as others (we're looking at you, Metro Exodus Enhanced Edition).

The RTX 4090 still more than doubles the performance of the RX 6950 XT, so AMD's soon-to-be previous-gen RDNA 2 architecture definitely isn't keeping up. It's going to be interesting to see what if anything AMD does with the RDNA 3 architecture to potentially close the gap. Intel's Arc A770 for example nearly matches the DXR performance of the RX 6800 and beats the RX 6750 XT, and it only has 32 RTUs compared to AMD's 60 Ray Accelerators (40 on the 6750/6700).

DLSS only improves performance by 21% on average at 1080p. It's also worth noting that if you're using DLSS in Performance mode (4x upscaling), native 1080p is what the Nvidia GPUs have to render, which then gets upscaled to 4K. There's a slight performance hit from DLSS (smaller on Ada Lovelace than on Ampere), so the native 1080p results also provide a good hint of what to expect from Nvidia's DLSS Performance mode. 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

The Nvidia Ada Lovelace and RTX 40-series GPUs feature a lot of new tech, so there are some things we can't even test against on previous generation graphics cards — like DLSS 3. The short recap is that DLSS 3's main new feature is Frame Generation, which uses a more powerful Optical Flow Accelerator fixed function unit that's only found in Ada GPUs. Ampere and even Turing GPUs also have OFAs, but they are much slower and apparently can't handle the computational requirements of Frame Generation — or at least that's Nvidia's claim and we have no way to prove or disprove it.

DLSS 3 is otherwise backward compatible with DLSS 2, meaning if you have an RTX 20- or 30-series graphics card, it will just do regular DLSS 2 Super Sampling. All of the preview games that Nvidia provided access to include a toggle for Frame Generation, and sometimes you can even use it without enabling DLSS — which could be useful in fully CPU-limited games like Flight Simulator. As an added bonus (to Nvidia GPU owners at least), DLSS 3 requires that a game also implement Nvidia Reflex, which can substantially lower total system latency in games that support it.

Nvidia has announced that over 35 games and apps are already working to implement the technology, and given the benefits we experienced, we expect many more will join that list in the coming years.

Frame Generation works by taking the previous frame and the current frame and then generating an interstitial frame. Every other frame thus ends up being created via Frame Generation, with the drivers balancing frame outputs to keep frames coming at an even pace. You'll note that this means Frame Generation basically adds one frame of additional latency, plus additional overhead, though Reflex basically overcomes the latency deficit relative using DLSS 2 without Reflex. In other words, Nvidia is spending the latency gains from Reflex on Frame Generation.

In a best-case scenario, Frame Generation can theoretically double the framerate of a game. However, it does not increase the number of frames generated by the CPU and there's no additional input sampling. Still, games should look smoother and subjectively they also feel smoother. For competitive esports gaming, however, the lower latency of Reflex without Frame Generation may still provide an advantage.

With that preface out of the way, here's a preview of what DLSS 3 can do in seven different games and/or demos.

Nvidia claims that with DLSS 3, performance with the RTX 4090 can be anywhere from 2x to 4x faster than on an RTX 3090 Ti — with the latter running DLSS 2. Most of the games and applications we tested don't support cards other than the RTX 4090 right now, and timing constraints limited how much we could test, but generally speaking DLSS 2 Performance upscaling (4x upscale, or rendering 1080p and upscaling to 4K) roughly doubles your FPS, at least in situations where CPU bottlenecks aren't a factor.

With DLSS 3, there are situations where the RTX 4090 is up to 5x faster than native, like in the Chinese MMO Justice, though every preview we looked at shows different results. 3DMark's DLSS Feature Test shows 3x the performance. A Plague Tale: Requiem and F1 2022 are up to 2.4x faster. Cyberpunk 2077 improves up to 3.6x over native, and the Unreal Engine 5 demo Lyra is up to 2.2x faster. Finally, Flight Simulator only improves by 2x, but it's also a game that's almost entirely CPU limited on the RTX 4090. Doubling framerates in that case represents a nice improvement.

How does DLSS 3 image quality with Frame Generation compare to normal rendering? We're still working on looking at the various games, which requires some changes to our normal testing as we need to capture games at higher than 60 fps rates to get meaningful results. Subjectively, though, we can say that other than looking smoother, we wouldn't be able to immediately spot the difference between the rendered and generated frames while gaming.

We did want to look at latency with and without Frame Generation, so we turned to the preview version of Cyberpunk 2077. Thanks to Reflex, we could run the game on RTX 30-series GPUs as well, so we captured performance using Nvidia's FrameView utility. We maxed out the settings, so 4K with RT-Ultra and then turned Ray Traced Lighting to the Psycho setting. We also tested with Reflex enabled on all the GPUs, because that's how most gamers will play.

Latency has a direct correlation with frame rate, so faster GPUs inherently get lower latency. We see that in the native results, where the RTX 4090 has 52ms of latency compared to 81ms on the 3090 Ti, and it just gets worse from there with the Titan RTX (we picked it as the fastest 20-series GPU, and also because it has 24GB of VRAM) sitting at 279ms. The RTX 4090 with Frame Generation ends up at 67ms latency, basically about one frame's delay if the game were running at 60 fps — which is pretty close, as we'll see in a moment.

Enabling DLSS Quality upscaling nearly doubles the performance of the RTX cards, and latency drops quite a bit. The 4090 now sits at 34ms, or 46ms with Frame Generation. That's a 12ms difference, with the DLSS 3 running at 106 fps. The 4090 with DLSS 3 ties the latency of the 3090 Ti with DLSS 2, with more than double the performance.

Finally, with Performance mode upscaling, the 4090 with Frame Generation enabled actually ends up with slightly more latency than a 3090 Ti, but still with more than double the frame rate. If you have a 144 Hz 4K display, which is what we used for testing, DLSS 3 looks better and feels a bit smoother. It's not a huge difference, but it's nice to see triple-digit frame rates at 4K in such a demanding game.

The last thing we want to show is how the various fastest GPUs from Nvidia, AMD, and even Intel stack up in pure ray tracing performance — or at least as close as we can get to "pure" RT as possible. We used the same Cyberpunk 2077 RT-Ultra plus Psycho Lighting settings as our latency testing, but now we have several AMD GPUs along with the Intel Arc A770. We also used the 3DMark DXR Feature Test, set to 20 rays, which apparently does do full ray tracing rather than hybrid rendering.

Starting with 3DMark's DXR Feature Test, it's no surprise that the RTX 4090 ranks way above any other GPU. Also note that this is still without the test implementing Ada's SER, OMM, and DMM features, so just the use of faster RT cores, with more of them, accounts for the difference between the 4090 and the 3090 Ti.

Moving down the chart, if you're like us, the placement of Intel's Arc A770 should raise an eyebrow. Yes, with just 32 RTUs running at somewhere around 2.35 GHz, Intel with its first generation of ray tracing hardware manages to match 84 of Nvidia's first generation RT cores. That probably won't happen in every game that supports ray tracing, but you'll also note that all of AMD's fastest RX 6000-series GPUs rank below the Arc A770 and Titan RTX. When people say AMD's first-generation RT hardware is weak, this is why.

Things don't look quite so dire in Cyberpunk 2077, though nothing below the RTX 3090 Ti even comes remotely close to hitting playable frame rates at maxed out settings and 4K. The RTX 4090 at native gets 39 fps, and Frame Generation nearly doubles that to 67 fps.

We also did some apples-and-oranges comparisons of DLSS modes versus FSR modes. We're calling DLSS Quality upscaling more or less equal to FSR Ultra Quality, and DLSS Performance as somewhat equal to FSR Quality mode. Neither of those are entirely accurate, but we also just wanted to see if we could get any of the non-RTX cards up to playable levels of performance. The answer: Nope.

The RX 6950 XT tops out at 23 fps with FSR Quality mode; we'd have to drop to Balanced or Performance mode to break 30 fps, and we can definitely tell the difference in visual fidelity when those are in use. Also, note that the Arc A770 can't keep up with AMD's faster GPUs, but it does rank ahead of the RX 6800 in all three tests.

Bottom line: Nvidia has seriously upped the ante for ray tracing hardware performance with Ada Lovelace. We already knew that from the architectural overview, but this provides some hard evidence showing just how much things have improved. 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

GPUs are also used with professional applications, AI training and inferencing, and more. We're looking to expand some of our GPU testing, particularly for extreme GPUs like the RTX 4090 and would love your input on a good machine learning benchmark (that doesn't take forever to run). For now, we have a few 3D rendering applications that leverage ray tracing hardware, along with the SPECviewperf 2020 v3 test suite.

SPECviewperf 2020 consists of eight different benchmarks. We've also included an "overall" chart that uses the geometric mean of the eight results to generate an aggregate score. Note that this is not an official score, but it gives equal weight to the individual tests and provides a high-level overview of performance. Few professionals use all of these programs, however, so it's generally more important to look at the results for the applications you plan to use.

AMD recently discussed improvements to its OpenGL and professional drivers. Some or all of these improvements have been in the consumer Radeon Adrenalin 22.7.1 and later drivers, but we retested all of the GPUs with the latest available drivers for these charts. AMD's RX 6000-series cards all show an overall 65–75 percent performance increase compared to when we tested for the RX 6950 XT review. The new drives roughly a doubled catia-06 performance and quadrupled the performance in snx-04.

Those new AMD drivers help push most of AMD's GPUs ahead of even the RTX 4090 in our overall SPECviewperf 2020 v3 Geoman chart. In the individual charts, 3dsmax-07, energy-03, maya-06, and solidworks-07 still have the RTX 4090 in the top spot, so if you use any of these applications, consider these results — and also note that Nvidia's professional RTX cards like the RTX 6000 48GB card can offer major improvements thanks to their fully accelerated professional drivers (look at the Titan RTX in snx-04 for some idea of what Nvidia's pro drivers can do).

Intel's SPECviewperf results with the Arc A770 are decidedly mediocre. It places last on most of the charts, sometimes by a wide margin (snx-04 and solidworks-07). Driver updates could significantly improve its score, though some of the differences may be architectural. Plus the A770 costs about half as much as cards like the RX 6900 XT, and about one-fifth as much as the RTX 4090.

Next up, Blender is a popular open-source rendering application that has been used to make full-length films. We're using the latest Blender Benchmark, which uses Blender 3.30 and three tests. Blender 3.30 also includes the new Cycles X engine that leverages ray tracing hardware on AMD, Nvidia, and even Intel Arc GPUs. It does so via AMD's HIP interface (Heterogeneous-computing Interface for Portability), Nvidia's CUDA or OptiX APIs, and Intel's OneAPI.

Being open-source has one major advantage: the ability of the various companies to deliver their own rendering updates. This gives us the closest option for an apples-to-apples performance comparison of professional 3D rendering using the various GPUs.

Rendering applications like Blender will tend to show the raw theoretical ray tracing hardware performance, and again we see a major generational jump with the RTX 4090 over the RTX 3090 Ti — 83% higher to be precise in our overall score. In the three test scenes, the improvements are 101% with Monster, 61% in Junkshop, and 91% in Classroom.

And while Blender 3.30 does work on AMD and Intel GPUs, the performance gap is massive. The fastest AMD card is still less than half as fast as an RTX 3080 10GB and 45% slower than the Titan RTX. Intel's Arc A770 lands just above the RX 6800 this time. Put another way, the five non-Nvidia GPUs combined only manage about 73% of the performance of a single RTX 4090. Ouch.

Our final two professional tests only support Nvidia's RTX cards for hardware ray tracing. That's largely because Nvidia's CUDA and OptiX APIs have been around and supported for far longer than any alternative, and while Optane formerly supported OpenCL, stability and difficulty combined with a lack of development efforts caused them to drop support with more recent versions. We've asked both OTOY and Chaos Group if they intend to add support for AMD and Intel GPUs, but these charts are Nvidia only. Note that OctaneBench hasn't been updated since 2020, while the latest V-Ray Benchmark is from June 13, 2022.

As with Blender, the performance improvement delivered by the RTX 4090 in Octane and V-Ray varies with the rendered scene. The overall OctaneBench result is 78% higher than on the RTX 3090 Ti, while the two V-Ray renders are 83% and 86% faster. Again, if you do this sort of work for a living, the RTX 4090 could make for an excellent upgrade.

[Note: I'm still looking for a good AI / machine learning benchmark, "good" meaning it's easy to run, preferably on Windows systems, and that the results are relevant. We don't want something that only works on Nvidia GPUs, or AMD GPUs, or that requires tensor cores. Ideally, it will use tensor cores if available (Nvidia RTX and Intel Arc), or GPU cores if not (GTX GPUs and AMD's current consumer lineup). If you have any suggestions, please contact me — DM me in the forums, or send me an email. Thanks!] 

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

In our Intel Arc A380 review, we introduced video encoding performance and quality testing using FFmpeg. We wanted to do the same sort of testing — and more — on the RTX 4090, though time has conspired against us.

To quickly recap, Nvidia's Ada Lovelace architecture includes a new 8th generation NVENC block, which adds support for AV1 encoding. On RTX 40-series cards with 12GB or more VRAM, there are also dual NVENC blocks, which can either work on separate streams or double the encoding performance of a single stream.

The new NVENC allows for encoding at up to 8K 60 Hz, for the 0.001% of people that have such a display. No, I'm not jealous. Why do you ask? (Yes I am, and I'd love to have an 8K display for testing gaming performance as well while I'm here.) Maybe we'll even get 8K 120 Hz support in the future, using both encoders.

We'll update this page with our testing results once we've finished doing the testing, so stay tuned.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

We measure real-world power consumption using Powenetics testing hardware and software. We capture in-line GPU power consumption by collecting data while looping Metro Exodus (the original, not the enhanced version) and while running the FurMark stress test. Our test PC remains the same old Core i9-9900K as we've used previously, to keep results consistent.

For the RTX 4090 Founders Edition, our previous settings of 2560x1440 ultra in Metro and 1600x900 in FurMark clearly weren't pushing the GPU hard enough. Power draw was well under the rated 450W TBP, so we increased the resolution and settings to 4K Extreme for Metro and 1440p for FurMark. The following charts are intended to represent worst-case power consumption, temps, etc. and so we do check other settings to ensure we're pushing the GPUs as much as reasonably possible.

Compared to the Asus RTX 3090 Ti, the power results are very similar. That makes sense as both cards have a 450W TBP. Considering you're getting at least 50% higher performance with the 4090 (at 4K), that proves the core architecture is quite a bit more efficient than with Ampere. Both the 4090 and 3090 Ti average just under 440W in Metro Exodus, and 460W–470W in FurMark. Interestingly, the 4090 power draw does take a dip after about six minutes running FurMark, but we don't see a corresponding drop in clock or fan speeds.

GPU clocks are a massive improvement over any prior Nvidia GPU architecture. The RTX 4090 has an official boost clock of 2520 MHz, though Nvidia often exceeds that for gaming workloads. In Metro Exodus, as well as many of the other games we've checked, typical GPU clocks are in the 2750 MHz range. That's higher than even AMD's RX 6750 XT, and there are indications we'd be able to push those clocks even higher if we're willing to crank up the power limit. FurMark does drop to just under 2.5 GHz, but then the other Nvidia GPUs show even larger throttling, with the 3090 Ti as an example running at just 1.44 GHz.

We were impressed with the RTX 3090 Founders Edition at launch, as it had incredibly low noise levels and temperatures — at least while gaming. Cryptocurrency mining was another matter, and the Founders Edition cards routinely hit 110C on their GDDR6X memory before throttling kicked in. But for gaming, the 3090 was usually very cool and quiet. The 3090 Ti from Asus had to ramp up fan speed quite a bit in order to remain at its target temperature of around 65C. The 4090 Founders Edition ends up splitting the difference: slightly higher temps of 67C but lower fan speed. FurMark is also more favorable toward the 4090 than the 3090 Ti, though it's not representative of very many real-world workloads.

We measure noise levels at 10cm using an SPL (sound pressure level) meter, aimed right at the GPU fans in order to minimize the impact of other noise sources like the fans on the CPU cooler. The noise floor of our test environment and equipment is around 32–33 dB(A). The RTX 4090 while gaming plateaued at 45.0 dB(A) and a fan speed of around 40%. The Asus 3090 Ti in comparison ran at 49.1 dB(A) and a 74% fan speed, while a Sapphire RX 6950 XT only measured 37.3 dB(A).

The fans on the RTX 4090 can get louder if needed, and ramping up fan speed to 75% results in 57.2 dB(A). Hopefully the card won't ever need to run the fans that high — and with the death of GPU mining, that's far more likely to be the case during the lifetime of the card.

Overall, the RTX 4090 Founders Edition isn't worse than the RTX 3090 Ti from a pure power consumption point of view, and it offers far superior performance. We're also very interested in doing additional testing with overclocking and third-party cards, which we'll be looking at in the near future.

One thing to note is that, while the power draw of the RTX 4090 might be 450W and that seems like a lot (it is), for chips it's really more about thermal density. Cooling 450W in a 608mm^2 chip isn't that difficult. It's actually quite a bit easier than cooling 250W in a 215mm^2 chip, which is what Alder Lake i9-12900K has to deal with. Zen 4 is actually potentially worse, with a 70mm^2 CCD (core complex die) that could potentially pull well over 140W.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

RTX 4090 Third Party Cards and Overclocking

In our initial review, all of the testing was confined to just the RTX 4090 Founders Edition. Now that a few days have passed, we have additional test results from some of the third party cards. We'll be updating this page with more cards over time, so if you want to see how all the RTX 4090 cards we've tested stack up, this is the place to come.

We tested the third party cards using the public 522.25 Nvidia launch drivers. For several games (where we saw larger changes in performance), we retested the Founders Edition as well. Thus the 4090 FE results on this page may be slightly different than on the earlier pages.

We're limiting our testing of third party cards to just 1440p ultra and 4K ultra, for what should hopefully be obvious reasons — but if you need us to restate the obvious, it's because 1080p performance is severely GPU limited in non-DXR/RT games. It also saves time if we eliminate half of the tests we have to run.

Overclocking results are also included, at 4K only. We tested the RTX 4090 Founders Edition with a +200 MHz core clock and +1200 MHz on the GDDR6X memory (23.4 Gbps effective speed), with the power limit maxed out at 133%. Other cards managed similar results, give or take, though the Asus ROG Strix OC was able to reach +1750 MHz on its memory (24.5 Gbps effective), which suggested it had "down-labeled" 24Gbps Micron chips.

For third party cards, we're only running OC testing at 4K with ray tracing enabled, which will show the largest potential gains. As always with manual overclocking, the silicon lottery is in effect. We've only tested on sample of each card, so our results are not guaranteed. Other cards of the same model may hit higher or lower "stable" results.

Even at 4K ultra settings, the differences between the various cards are extremely limited. At present, the Asus ROG Strix OC ranks as the fastest 4090 we've tested, barely, beating the MSI Suprim Liquid X by about 1% and the Founders Edition by 2% in our DXR test suite. In the standard rasterization test suite, the difference shrinks to just 1%, give or take.

You should make your purchasing decision primarily on price, aesthetics, and other factors beyond raw performance. The Asus card for example had better thermal characteristics as well, and the GDDR6X memory only reached 64C even when overclocked. The Founders Edition meanwhile hit 78C on its memory, while the MSI card was "worst" at 82C.

Those are interesting results, as with the previous generation cards the GDDR6X temperatures tended to be far, far higher. Nvidia says Micron has deployed a new process node for its latest 16Gb GDDR6X modules, with lower power requirements, and these results bear that out.

As for overclocking, the Asus card takes first place again, beating the Founders Edition by 2%. The MSI card actually had the worst overclocking results, as its memory only managed +800 MHz (22.6 Gbps effective), so it was 1% slower than the Founders Edition. Compared to the reference card running stock clocks, a manual overclock on the Asus card improved overall performance by 9–13 percent in the individual games, and 10% on average.

The 1440p testing results echo what we showed above at 4K, so there's not much more to add. Even in ray tracing games, the gaps in performance between the various cards are very limited.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

Two years between major GPU architectural updates can feel like a long time, and the past two years have been incredibly painful for all gamers looking to upgrade their graphics card. Thankfully, the long dark night of GPU cryptocurrency mining is over (for now at least), and we can only hope that supply and availability of the RTX 40-series cards is vastly improved over the Ampere generation.

The RTX 4090 and Ada Lovelace are, frankly, impressive as hell. From a performance and technology perspective, Nvidia has pushed things further than we've likely ever seen between GPU architectures. In our testing, we saw performance improvements of over 50% at 4K ultra, and a 78% increase in ray tracing heavy games. Toss in DLSS and DLSS 3 Frame Generation and the potential gains are even more impressive.

We're finally at the point where game developers can really go nuts with ray tracing effects, at least for the highest tier of gamers who own RTX 40-series hardware. That's obviously going to be an extremely small market right now, but over time we can expect it to grow. Hopefully, we'll get some next-generation games in the coming months that can truly show off what the RTX 4090 can do.

It's not just games that benefit, though. Professional applications see equally large improvements in performance. Video editing can be more than twice as fast as on previous generation RTX 30-series hardware, and AV1 is fully supported as well. Updated applications should arrive this month with native AV1 support, including Discord and OBS.

There is of course a price to pay, and the piper isn't holding back. $1,599 (or more) for an RTX 4090 that's only used for gaming is far more than most people would be willing to pay. But we also don't expect Nvidia to ship huge quantities of 4090 cards. Looking at the RTX 30-series in the latest Steam Hardware Survey, the 3080 models combined account for over five times as many installations as the 3090, the 3070 models nearly double the 3080 models, and the RTX 3060 variants double that again. RTX 3050 as the low man on the totem pole drops back to slightly lower than RTX 3070 levels of adoption, likely in large part due to that card's history of an abnormally high price to performance.

Considering the price of the RTX 3090 and RTX 3090 Ti, there are clearly plenty of gamers willing to spend well over $1,000 on a graphics card, and that's the market Nvidia is targeting. In fact, if Steam is to be trusted (sprinkle liberally with salt), there are more RTX 3090 series cards in the hands of gamers than any single RX 6000-series card, with the exception of the RX 66xx models — those account for 0.74% of the total compared to 0.49% for the RTX 3090.

Should you buy an RTX 4090? That depends on how much disposable income you have and what you plan to do with it. Some people have expensive hobbies — sports cars, attending professional sports games, model airplanes, stylish clothing, fancy restaurants, whatever — and dropping a couple of grand over a two-year period wouldn't even make them flinch. Others struggle to make ends meet and are content with a complete gaming PC for under $500.

If you can't afford an RTX 4090, that's fine. Many of us will simply have to look on in awe or grumble like someone getting passed by an exotic sports car on the freeway. Lower-priced RTX 4080 models will arrive next month, and AMD will launch its RX 7000-series and RDNA 3 GPUs before the end of the year. Hopefully sometime in the not-too-distant future we'll see substantially faster GPUs enter the high-end, mainstream, and budget sectors.

But for now, Ada Lovelace reigns supreme. Maybe AMD will contend for the throne in the near future, but with a graphics die size (not including cache or memory controllers) of just 308mm^2 according to the latest rumors, it's difficult to imagine AMD packing as many significant upgrades in the package as Nvidia has here. If you can't or won't wait to see what AMD has to offer, the RTX 4090 will remain as the fastest GPU from Team Green for a good while, or at least until the inevitable RTX 4090 Ti arrives.

The GeForce RTX 4090 Founders Edition and other RTX 4090 models officially go on sale tomorrow, October 12, with a starting price of $1,599. We'll be reviewing quite a few AIB partner cards over the coming weeks, so stay tuned for that as we'll also add their performance results to this article.

• MORE: Best Graphics Cards
• MORE: GPU Benchmarks and Hierarchy
• MORE: All Graphics Content

Modern graphics processors have so many features and capabilities that even hardcore gamers and tweakers sometimes have difficulties finding the proper settings to achieve desirable performance. That's where one enthusiast claims that his professionally tuned performance profiles come in.

Yuri Bubliy claims that his upcoming Radeon Monster Profile (RMP), a third-party utility he is creating, will boost the performance of AMD's RDNA2 GPUs through voltage/frequency curve modifications. The creator even says the app can boost a Radeon RX 6800 XT up to the level of Nvidia's GeForce RTX 3090 Ti — but in one specific synthetic benchmark that doesn't tell us a lot about the broader picture. 

AMD's Adrenaline software for Radeon graphics processors already offers plenty of performance features, which even overvolt the GPU for overclocking. Furthermore, AMD's pack of FidelityFX features like FSR 2.0 also boost the performance of Radeon graphics cards. Yet the Radeon Monster Profile developed by Yuri Bubliy claims to deliver even more tweaked performance.

Based on the tests done by the developer, the profile purportedly increased the Radeon RX 6800 XT's performance in 3DMark: Time Spy tests by 13% over stock, bringing it up the GeForce RTX 3090 Ti level. It's an excellent way to improve one of the best graphics cards, if it works as claimed. 

Bubliy claims that the Radeon Monster Profile combines unique settings and a new volt-frequency curve for Radeon RX 6000-series graphics chips.

"Particular attention was paid to the memory subsystem, the operating voltage was reduced by 6%, which ultimately reduced the heat of GDDR6 modules," Bubliy said. "The main mechanism for increasing performance is to increase the requested core frequency relative to the operating voltage."

According to the developer, the result is that RMP increases GPU frequency by 300 MHz without affecting factory settings for thermal throttling and fans. Of course, a 300 MHz overclocking requires a decent cooling system, so while the maker does not urge the owner to buy an aftermarket cooler for AMD's reference graphics boards, he stresses that non-reference boards should be good with the profile.

"The profiles are universal and adaptive," Bubliy said. "The RMP does not increase the operating voltage or change the current limit."

Naturally, we would all like to boost the performance of our gear through simple software utilities, but you should keep your expectations in check, particularly when claims are made based on a single benchmark. Seeing is believing, especially when it comes to big claims like these. The utility is due later this fall, and we'll give it a spin to see how it fares. 

Nvidia on Wednesday revealed the printed circuit board (PCB) design of its GeForce RTX 4090 graphics card and disclosed some of the peculiarities of its new product. As it turns out, the company completely redesigned the card's voltage regulating module (VRM) by adding a PID controller with a feedback loop to ensure proper power supply to the GPU and maximize its overclocking potential.

When fully assembled, Nvidia's GeForce RTX 4090 Founders Edition graphics card resembles its predecessor as it continues to use an extensive dual-fan cooling system featuring a unibody design. But the new product has several distinctive differences outside and, more importantly, inside.

Just like previous generation Founders Edition graphics cards, the new GeForce RTX 4090 FE uses a very sophisticated yet compact PCB with a triangular cut on its back (or its right side, depending on how you look at it) that is meant to maximize the efficiency of Nvidia's flow through thermal design (which is further boosted by bigger fans).

The PCB has a single 12+4-pin 12VHPWR additional PCIe Gen5 power connector that can deliver up to 600W of power to the device, yet default BIOS settings do not allow Nvidia's GeForce RTX 4090 FE to draw more than 450W. The board also comes with a 20+3-phase VRM (20 phases for the GPU, 3 phases for memory), which is a slight improvement from the 18+3-phase power circuitry of the GeForce RTX 3090/3090 Ti FE.

In addition to more power phases, Nvidia implemented a PID controller with a feedback loop to reduce current spikes and drops, effectively stabilizing power delivery to its GeForce RTX 4090 Founders Edition graphics card, reports TechPowerUp. Nvidia says that its new VRM enables a substantially lower power management response time without quantifying it in absolute figures. A stable and clean power supply is crucial for stability under high loads and overclocked conditions. The green company calls this feature 'power transient management,' which might be one of the ways it manages to push its AD102 graphics processor to 3 GHz without overvolting it and using extreme cooling methods.

In addition to enhancing the cooling system with bigger fans that provide up to 20% more airflow and redesigning power delivery, Nvidia also uses new GDDR6X memory chips with its GeForce RTX 4090 FE that consume less power. While we can only speculate how the power consumption of DRAMs featuring PAM-4 signalings was reduced, we suggest that Micron uses a thinner fabrication process to make these chips, which automatically lowers their power consumption. Meanwhile, with colder memory, the cooling system will better cool down the GPU, which means extra overclocking headroom.

Nvidia's GeForce RTX 4090 Founders Edition graphics cards will go on October 12 and will likely join the best graphics cards, at least for those willing to pay $1,599 per unit.

Hot on the heels of Nvidia's GeForce RTX 4090 announcement, Alphacool and EKWB today introduced waterblocks for Nvidia's GeForce RTX 4090 Founders Edition and reference graphics cards. The waterblocks will enable owners of custom-built liquid cooling systems to push these shiny new graphics boards to their limits. In fact, EKWB already says that a 3 GHz GPU frequency is possible for Nvidia's AD102 graphics chip with proper cooling. 

Alphacool's Eisblock Aurora Acryl GPX-N RTX 4090 is aimed at Nvidia's GeForce RTX 4090 reference design, though the company's datasheet vaguely mentions GeForce RTX 4XXX graphics boards. The waterblock consists of a chrome-plated copper coldplate that cools down the GPU, memory, and the voltage regulating module with transparent acrylic top. To increase performance of the waterblock, Alphacool adapted its fin structure to maximize both waterflow and cooling surface area — fins are now 0.4 mm thick and are located on a 0.4 mm distance from each other. To ensure that the card does not bend, Alphacool's Eisblock Aurora Acryl GPX-N RTX 4090 comes with an aluminum backplate. 

EKWB's EK-Quantum Vector² water blocks are designed for Nvidia's GeForce RTX 4090 FE boards and closely follow the company's Quantum Vector concept of covering the whole board — thereby cooling down the GPU, memory, and power delivery circuitry. The new waterblock uses an 11mm nickel-plated copper coldplate with an optimized fin design and comes with either a transparent plexiglass or all-black acetal top. Both versions include a black-anodized aluminum backplate. 

Waterblocks from Alphacool and EKWB are compatible with standard G1/4-inch connections and also feature addressable LEDs that can be controlled using software from leading makers of motherboards. 

One of the things to note about waterblocks from EKWB is that they are designed only for Nvidia's own Founders Edition GeForce RTX 4090 graphics cards and not for reference boards or custom boards from the company's partners. While the green company equipped its GeForce RTX 4090 FE with an extremely sophisticated triple-slot air cooler that suggests overclocking, power consumption of the card is capped at 450W and so far Nvidia has not announced any plans to supply overclocking BIOS for the product to unlock higher power delivery. Thus, while Nvidia's GeForce RTX 4090 Founders Edition promises to be a good overclocker (at least from Nvidia's own words), some of its partners' offerings — which rumors suggest may have overclocking BIOSes — may offer even better potential. 

Waterblocks for custom liquid cooling systems are rarely cheap. EKWB will sell its EK-Quantum Vector² FE RTX 4090 D-RGB (nickel + plexi or nickel + acetal) for €240 ($237), while Alphacool's Eisblock Aurora Acrylic GPX-N RTX 4090 will sell for €160 ($158). The waterblocks for Nvidia's top-of-the-range Ada Lovelace GPU are no available for order from both companies, with Alphacool setting delivery at three to four weeks from now.

One of the things that caught our attention during Nvidia's GeForce RTX 40-series announcement was the company's claim that it has already run its AD102 GPU at 3.0 GHz in its labs. This suggests the new GPU could be an overclocking champ with plenty of headroom that AIB partners could tap on factory-overclocked boards, as well as manual overclocking. The RTX 4090 is set to launch on October 12 and will likely join the best graphics cards — at least for those with deep pockets.

Nvidia's AD102 chip packs 76 billion transistors and is made using TSMC's 4N fabrication technology. Running a GPU this complex at 3.0 GHz is a big deal, as typically chips that run at high clocks are considerably less complex. Meanwhile, Nvidia's GeForce RTX 4090 has a boost clock of 2520 MHz, so overclocking an AD102 GPU to ~3.0 GHz shouldn't be too hard for a good air or closed loop liquid cooler. Naturally, liquid nitrogen should enable the GPU to be pushed even further.

The TSMC 4N fabrication technology that Nvidia Ada Lovelace GPUs use has been customized for the company, possibly to help with yields, process window, power, and clock speeds. Nvidia does not disclose the peculiarities of TSMC's 4N or how it customized the company's 'default' N4/4P node to achieve its goals in terms of complexity, performance, and power. Considering the clock speed potential of Nvidia's AD102, Nvidia and TSMC likely optimized the production node both for complexity and frequency. However, we do not know whether TSMC implemented libraries set to be used for upcoming N4P or N4X fabrication process technologies.

With its GeForce RTX 30-series graphics cards, Nvidia made thick three-slot cooling systems relatively common on custom graphics card, but with its GeForce RTX 40-series boards it looks like the company will make such ultra-large coolers truly widespread. Based on the custom GeForce RTX 40-series boards we have seen, they will all feature sophisticated cooling systems that will probably enhance their overclocking potential. How many of these boards will be able to hit 3 GHz remains to be seen, but it's certainly a possibility for at least some of the models.

One of the things we don't know with AD102-based graphics boards is how much power a 3 GHz GPU frequency will require. Over the past few months, we've heard rumors of flagship products with 600W or even higher power envelopes, but the reference GeForce RTX 4090 is only rated for up to 450W thermal graphics power. Some rumors suggest that Nvidia's partners might enable an up to 600W power envelope for their premium offerings. If this is true, we may soon see some heavily overclocked GeForce RTX 4090 boards.

Hitting a 3+ GHz frequency for a GPU is hardly a breakthrough, as AMD's Navi 21 XTXH was overclocked to 3.20 GHz in mid-2021 and the world's first GPU to break the 3 GHz barrier was an Nvidia Maxwell GPU, circa 2015. But AMD's Navi 21 is a chip that packs 26.8 billion transistors, made using TSMC's N7 technology (which was presumably optimized for AMD's needs), whereas Nvidia's AD102 is almost three times more complex. And Maxwell required liquid nitrogen to hit that speed, while it seems like Ada may reach that on air cooling.

We're definitely interested in seeing just how far custom RTX 4090 cards push clock speeds, and how much power they require to do so. We should have full reviews up on or slightly before the October 12 launch date.

Although Nvidia lowered recommended price of its flagship graphics card by $400 with the launch of the GeForce RTX 4090, it increased the prices of other high-end offerings significantly with its GeForce RTX 4080 models. It is self-evident in Europe that the new GeForce RTX 4080 16GB cost almost the same amount of money as the GeForce RTX 3090 24GB several months ago.

Prices of graphics cards have been increasing gradually since the early 2000s, but in recent years prices of the best gaming graphics cards rose quite dramatically due to higher prices of graphics processing units (GPU) production at contract manufacturers like TSMC and Samsung Foundry, higher costs of actual boards production, chip shortages/complicated logistics as well as very strong demand from gamers, miners, and professionals.

With its Ada Lovelace family of GPUs, Nvidia faces higher costs at TSMC as it is significantly more expensive to make chips using TSMC's 4N technology than on Samsung Foundry's 8LPP production node at the same die size and yields. Hence, price hikes were always on the table for GeForce RTX 40-series graphics boards. Yet, Nvidia's price hikes look to be more significant than expected.

U.S. MSRPs do not include taxes, whereas German MSRPs include a VAT of 19%. Furthermore, when Nvidia launched its GeForce RTX 30-series two years ago, the euro was considerably stronger against the U.S. dollar, so those boards were more expensive in Germany than in the U.S.

Nvidia's GeForce RTX 3080 10GB used to cost $699 at launch two years ago, and it was not cheap at all because it was not the company's flagship offering. Yet, Nvidia's GeForce RTX 4080 12GB now costs $899, and it's not even the second most potent product in the lineup. Meanwhile, this 12GB board costs even more in Europe (€1m099, according to Andreas Schilling), which makes it somewhat less appealing.

It gets worse with the GeForce RTX 4080 16GB which costs $400 more than the company's GeForce RTX 3080 12GB released a few months ago and carried the same MSRP as the GeForce RTX 3080 Ti based on a flagship GPU. Of course, Nvidia's AD103 GPU is considerably more complex and hard to produce than Nvidia's GA102, but GeForce RTX 4080 16GB is not a top-of-the-range device to cost well above $1,000.

Regarding flagship offerings, Nvidia's GeForce RTX 4090 will cost $1,599 at launch, which is $100 more expensive than the GeForce RTX 3090 at launch and $400 cheaper than the GeForce RTX 3090 Ti at launch. We have no idea why Nvidia decided to make its flagship offerings more affordable (perhaps because it is prepping GeForce RTX 4090 Ti that will sit above the RTX 4090). Still, well-known hardware blogger Kopite7kimi says that selling the GeForce RTX 4090 for $1,599 was a 'last minute decision' by Nvidia's chief executive.

Overall, while there are objective reasons why Nvidia's new graphics processors and graphics cards are more expensive than their predecessors, and they promise to be considerably faster than the GeForce RTX 30-series parts, their recommended prices seem to be too high. Meanwhile, if Nvidia makes last-minute decisions about a price drop for its GeForce RTX 4090, the company probably has abilities to sell its latest products a bit cheaper.

In a rather unexpected move, Nvidia introduced its new flagship graphics card for the professional visualization market and its new GeForce RTX 40-series graphics boards for gamers. The new Nvidia RTX 6000 48GB carries almost the name of its predecessor, yet this is an all-new Proviz solution based on the Ada Lovelace architecture.

Nvidia says that its RTX 6000 graphics card with 48GB of GDDR6 ECC memory onboard will offer up to 2X – 4X the performance of the previous-generation RTX A6000 due to the massively increased number of CUDA and RT cores as well as the brand-new Ada Lovelace architecture.

Nvidia's new flagship RTX 6000 48GB graphics card leverages the company's AD102 graphics processing unit, yet the company does not disclose its exact configuration for now. Typically, the company tends to lower the power consumption of its workstation-grade graphics cards compared to top-of-the-range gaming offerings in a bid to prolong their lifespan, which is why the RTX 6000 48GB has a TDP rating of up to 300W power (vs. 450W in case of the GeForce RTX 4090). Therefore, do not expect the RTX 6000 48GB to feature the same HPU configuration as the GeForce RTX 4090 or rumored GeForce RTX 4090 Ti.

"The RTX 6000 GPU's larger L2 cache, a significant increase in number and performance of next-gen cores, and increased memory bandwidth will result in impressive performance gains for the broad Ansys application portfolio," said Dipankar Choudhury, Ansys Fellow and HPC Center of Excellence lead.

Due to the reduced power consumption of the RTX 6000 48GB, the card comes equipped with a relatively compact dual-slot cooling system (4.4-inches x 10.5 inches) with a blower fan designed to ensure that the new board fits into workstations as well as servers (since the product can also work for remote rendering and virtual desktop applications). As for display outputs, the adapter has four DisplayPort 1.4 connectors to drive four monitors with up to 5Kp60 or 8Kp30 resolution or two monitors with up to 8Kp60 resolutions.

Nvidia said its new flagship graphics board for workstations would be available this December. Unfortunately, the company did not reveal the price of its new RTX 6000 48GB card. But based on the fact that Nvidia sells its previous-generation RTX A6000 48GB graphics card for an MSRP of $6,999, we can make some guesses about the company's recommended price for the new one that implies higher performance.

AMD is on the cusp of refreshing its top-end parts with the Zen 4 Ryzen 7000 chips that arrive later this month. But the company employs its Zen 2 architecture for its latest low-end chips, the Ryzen 3 4100 and Ryzen 5 4500, and all signs point to these chips being the company's budget offerings well into next year. The $99 quad-core Ryzen 3 4100 and the $129 six-core Ryzen 5 4500 are designed to challenge Intel's potent $122 Core i3-12100 and the $99 graphics-less 12100F that have become the go-to budget chips for gamers. The new AMD chips came as part of the company's broader launch of seven new Ryzen 5000 models that aim to shore up its rankings in CPU benchmarks and retake its position on the Best CPUs for gaming list. 

The Ryzen 5 4500 and Ryzen 3 4100 are part of AMD's new strategy of using its monolithic (single-chip) Zen 2-powered 'Renoir' APU silicon to tackle the sub-$150 market that Intel has dominated with its Alder Lake processors. AMD's low-end lineup leverages one full-fledged APU model with active integrated graphics, the Ryzen 4600G, and two models with disabled Vega integrated graphics, the 4500 and 4100. 

The Ryzen 5 4500 and Ryzen 3 4100 still come with the efficient 7nm TSMC process, and their monolithic design allows AMD to hit lower price points than possible with the more expensive chiplet-based Ryzen processors. However, that results in some tradeoffs — the chips only support 20 lanes of the PCIe 3.0 interface and DDR4 memory. In contrast, Intel's competing Core chips support up to PCIe 5.0 and either DDR4 or DDR5 memory, giving Intel the connectivity edge. That's not to mention that AMD's Renoir APUs have the aging Zen 2 architecture that debuted back in 2019. 

AMD's chips come with bundled Wraith Stealth coolers and benefit from the robust AM4 motherboard ecosystem that's teeming with plenty of affordable options. AMD also allows overclocking on these lower-end SKUs, an advantage over Intel's Core i3. 

AMD's lowest-end chips are long overdue, arriving a year and a half after the first wave of Ryzen 5000 chips, but they're badly needed. Intel's Alder Lake launch caught AMD uncharacteristically flat-footed, wresting away AMD's performance and pricing lead. Intel's aggressive pricing also brought superior value in every price range while exploiting AMD's glaring lack of any sub-$250 Zen 3 chips. AMD's tactic of bringing single-chip Renoir silicon to market serves as the counter, even if it leverages older tech to lower price points. Here's how the low-end Ryzen chips stack up. 

AMD Ryzen 3 4100 and Ryzen 5 4500 Specifications and Pricing

The Ryzen 3 4100 and Ryzen 5 4500 are nearly identical to the graphics-armed Ryzen 4000G models that AMD brought to market for OEMs and SIs back in 2020. However, these models come with a disabled iGPU, so you'll need a discrete graphics card. 

The $99 Ryzen 3 4100 is a Renoir model that slots in with four Zen 2 cores and eight threads that operate at 3.8 / 4.0 GHz base/boost clocks. The 65W chip comes with 4MB of L3 cache and has a disabled Vega graphics engine, positioning it to compete with Intel's graphics-less Core i3-12100F. In many respects, the quad-core 4100 essentially replaces the Ryzen 3 3100 that was never really available at retail due to shortages during the pandemic.

The $129 Ryzen 5 4500 has six Zen 2 cores and 12 threads that operate at a 3.6 / 4.1 GHz base/boost frequency. Like the 4100, this chip has a 65W TDP but comes with 8MB of L3 cache. Even though it doesn't have an active iGPU of its own, pricing dictates that the 4500 competes with the full-fledged $125 Core i3-12100 that comes with an iGPU. The 4500 effectively replaces the Ryzen 3 3300X that wasn't ever widely available at retail.

Aside from the disabled iGPU, these chips have the same Renoir design as the OEM-only Ryzen 7 4750G (hit the link for the full architectural details), including support for PCIe 3.0 instead of PCIe 4.0. As a result, this chip will make a great pairing for older, lower-end AM4 motherboards (you definitely don't want to pay for functionality you don't need by pairing it with a PCIe 4.0-supporting motherboard).

The Ryzen 5 4500 and Ryzen 3 4100 come with a bundled Wraith Stealth cooler sufficient for stock operation and supports up to DDR4-3200 memory. In addition, these chips are fully unlocked for overclocking the CPU cores, memory, and fabric. In contrast, Intel's non-K models only support memory overclocking, though Intel's nonsensical decision to keep certain voltages locked still restricts memory overclocking headroom.

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: Zen 4 Ryzen 7000 All We Know
• MORE: Raptor Lake All We Know

AMD Ryzen 3 4100 and Ryzen 5 4500 Power Consumption and Efficiency

The Intel Alder Lake chips still suck more power than AMD's Zen 3-powered Ryzen 5000 series chips, but pairing the Intel 7 process with the hybrid architecture brings big improvements, particularly in threaded work. However, Intel's decision to dial up the power limits as it pushes for performance supremacy results in higher peak power consumption and less efficiency. 

The quad-core Ryzen 3 4100 draws a stunningly-low amount of power during the heavy HandBrake and y-cruncher power measurements, but that results in the slowest performance of our test pool. However, that also results in superior overall efficiency, as we can see in the renders-per-day-per-watt metrics. 

Meanwhile, the six-core Ryzen 5 4500 unsurprisingly pulls more peak power than the quad-core Ryzen 3 4100, but its faster performance results in better efficiency profile than the Core i3-12100 in some scenarios. 

Here we take a slightly different look at power consumption by calculating the cumulative energy required to perform x264 and x265 HandBrake workloads, respectively. We plot this 'task energy' value in Kilojoules on the left side of the chart. 

These workloads are comprised of a fixed amount of work, so we can plot the task energy against the time required to finish the job (bottom axis), thus generating a really useful power chart. 

Remember that faster compute times, and lower task energy requirements, are ideal. That means processors that fall the closest to the bottom left corner of the chart are best. 

Test Setup

We tested with Windows 11 on an X570 motherboard to maintain a comparable test environment with the rest of the processors in the test pool. Of course, you wouldn't pair this chip with this class of motherboard, but even lower-end 300-series motherboards should provide enough juice for full operation. We also tested with secure boot, virtualization support, and fTPM/PTT active to reflect a properly configured Windows 11 install. We have a breakdown of the test system configurations at the end of the article.

Our overclocks were rather straightforward — we enabled the auto-overclocking Precision Boost Overdrive (PBO) feature with 'advanced motherboard' settings and adjusted the scalar setting to 10X. Additionally, as outlined in the table below, we matched our memory overclocks with a 1:1 FCLK/memory clock ratio to keep latency low, which games love. We tested the Ryzen 5 4500 and Ryzen 3 4100 in two different configurations each:

• Ryzen 5 4500: Corsair H115i 280mm water cooler, PBO Disabled, DDR4-3200
• Ryzen 5 4500 PBO: Corsair H115i 280mm water cooler, PBO Enabled, Scalar 10X, DDR4-3800, FCLK 1900 MHz (1:1 Ratio)
• Ryzen 3 4100: Corsair H115i 280mm water cooler, PBO Disabled, DDR4-3200
• Ryzen 3 4100 PBO: Corsair H115i 280mm water cooler, PBO Enabled, Scalar 10X, DDR4-3800, FCLK 1900 MHz (1:1 Ratio)

• MORE: How to Overclock a CPU
• MORE: How to check CPU Temperature
• MORE: Zen 4 Ryzen 7000 All We Know
• MORE: Raptor Lake All We Know

AMD Ryzen 3 4100 and Ryzen 5 4500 Gaming Benchmarks — The TLDR

As usual, we're testing with an Nvidia GeForce RTX 3090 to reduce GPU-imposed bottlenecks as much as possible, and differences between test subjects will shrink with lesser cards or higher resolutions. You would never see these low-end chips paired with an RTX 3090, but this allows us to highlight unrestrained chip performance. 

The quad-core Zen 2-powered Ryzen 3 4100 is easily the slowest chip in the test pool. But that isn't too surprising given its tame 4.0 GHz boost clock and scant 4MB of L3 cache. The six-core Ryzen 5 4500 only boosts to 4.1 GHz, but the doubled L3 cache and extra cores weigh in as it delivers 13% more performance at stock settings than the 4100.

The Core i3-12100 costs $122 for the full-fledged version with an iGPU, competing with the Ryzen 5 4500, and also comes as the graphics-less Core i3-12100F for $99, competing with the Ryzen 5 4100. Both of these models provide identical performance. The Intel Core i3-12100's Golden Cove P-cores give even the Zen 3-powered Ryzen 5000G chips a stiff challenge, so it's no surprise that the Zen 2 Renoir struggles mightily in comparison.

For similar or slightly lower pricing, the Core i3-12100/F is a whopping 49% faster than the Ryzen 3 4100 and 21.7% faster than the Ryzen 5 4500 in 1080p gaming, making it clear that budget gamers should opt for Intel's chip over the Ryzen competition.

Intel still restricts Core i3 overclocking to the memory, limiting gains to a few percentage points. In contrast, AMD allows full overclocking of both Ryzen 4000 models, but it isn't too useful. The overclocked Ryzen 3 4100 is a mere 3% faster than the stock config, while the tuned 4500 is about 6% faster than the stock setup.

As an interesting comparison point, we also included the Ryzen 5 4600G in the test pool. However, this is a full-fledged Renoir APU with a working RX Vega 7 integrated graphics engine, so it isn't meant to be used with a discrete GPU. Look to these pages for a full review soon with iGPU performance benchmarks.

Overall the Core i3-12100/F underlines its status as the best budget gaming CPU on the market — the Zen 2-powered Ryzen 4 4500 and Ryzen 3 4100 can't compete. 

Moving over to 1440p brings a GPU bottleneck into the equation, so the performance deltas between the chips shrink tremendously. Be aware that large performance deltas in a few of the game titles can heavily impact these types of overall measurements. It's always best to make an informed decision based on the types of titles you frequently play, so be sure to check out the individual game benchmarks below. That said, we don't see much variance in the results below — the Ryzen 5 4500 and Ryzen 3 4100 get beaten by the Core i3-12100 across the board. As such, we'll skip the commentary in the individual titles below. 

3DMark, VRMark, Chess Engines on AMD Ryzen 3 4100 and Ryzen 5 4500

Synthetic benchmarks don't tend to translate well to real-world gaming, but they do show us the raw amount of compute power exposed to game engines. It's too bad most games don't fully exploit it. That tendency is evident here as we see the Ryzen 5 4500 beat the Core i3-12100 by significant margins in both the DX11 and DX12 synthetic CPU tests, but it doesn't win in a single real-world gaming benchmark below.

Far Cry 6 on AMD Ryzen 3 4100 and Ryzen 5 4500

F1 2021 on AMD Ryzen 3 4100 and Ryzen 5 4500

Hitman 3 on AMD Ryzen 3 4100 and Ryzen 5 4500

Horizon Zero Dawn on AMD Ryzen 3 4100 and Ryzen 5 4500

Red Dead Redemption 2 on AMD Ryzen 3 4100 and Ryzen 5 4500

Watch Dogs Legion on AMD Ryzen 3 4100 and Ryzen 5 4500

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel

AMD Ryzen 3 4100 and Ryzen 5 4500 Application Benchmarks — The TLDR

We can boil down productivity application performance into two broad categories: single- and multi-threaded. These slides show the geometric mean of performance in several of our most important tests in each category, but be sure to look at the expanded benchmark results further below. 

The Core i3-12100 easily dispatched the Ryzen 4000 duo in our gaming tests, and while the Ryzen 5 4500 is a bit more competitive in the application testing, the Ryzen 3 4100 does little to rectify its poor showing.

The Core i3-12100 is 44% faster than the Ryzen 3 4100 and 34% faster than the Ryzen 5 4500 in our cumulative measure of single-threaded performance. In other words, the single-threaded contest isn't a contest at all.

Moving over to threaded work, the quad-core 12100 is 40% faster than the quad-core Ryzen 3 4100, leaving no reason to purchase it over Intel's Core i3.  

However, the six-core Ryzen 5 4500 is 3% faster than the 12100 in threaded work. That rather slim delta is a win for the Zen 2 chip against the Alder Lake Core i3, and we see several larger wins for the Ryzen 5 4500 in our more expansive multi-threaded benchmarks below. As you'd imagine given its advantage of having two more cores, the 4500 is faster in some threaded rendering and encoding applications than the Core i3-12100. 

However, given the large disparities in single-threaded performance and gaming, not to mention connectivity options, the Core i3-12100's better blend of performance and features is more attractive even for the productivity-minded (particularly if you plan to use a PCIe 4.0 SSD).  

Rendering Benchmarks on AMD Ryzen 3 4100 and Ryzen 5 4500

The Core i3-12100 is impressive in single-threaded rendering work, leading all competing Ryzen chips in both Cinebench and POV-Ray benchmarks. However, its quad-core design isn't as well suited as the Ryzen 5 4500 for some of the heavily-parallelized rendering workloads you'll see in the real world. For example, the Ryzen 5 4500 is 11%, 5%, and 25% faster in the threaded Cinebench, Corona, and C-Ray renderers, respectively. 

Encoding Benchmarks on AMD Ryzen 3 4100 and Ryzen 5 4500

Encoders tend to be either heavily threaded or almost exclusively single-threaded. The Ryzen 3 4100 isn't competitive in any of these benchmarks, but while the Ryzen 5 4500 isn't a factor in the lightly-threaded tests, it is more competent in heavily-threaded tasks. 

The Core i3-12100 dominates the lightly-threaded LAME, FLAC, and WebP encoding tests. Flipping over to threaded applications, the Ryzen 5 4500 excels in HandBrake, SVT-HEVC, and the SVT-AV1 encoding tasks. 

Adobe Premiere Pro, Photoshop, and Lightroom on AMD Ryzen 3 4100 and Ryzen 5 4500

We've integrated the UL Benchmarks Procyon tests into our suite to replace the aging PCMark 10. This new benchmark runs complex Adobe Premiere Pro, Photoshop, and Lightroom workflows with the actual software, making for a great real-world test suite.

These multi-phase benchmarks have mixed workload profiles, with some sections of each benchmark relying more upon single-threaded prowess while other portions of the workflow rely upon sheer threaded horsepower. This mixed-use type of application is common, and here we can see that the Core i3-12100's superior mix of overwhelmingly faster single-threaded performance and more than sufficient threaded heft given its price point combine to take the lead in every benchmark. 

Web Browsing, Office and Productivity on AMD Ryzen 3 4100 and Ryzen 5 4500

The ubiquitous web browser is one of the most frequently used applications. These tests tend to be lightly threaded, so a snappy response time is critical. Here we see the Core i3-12100 take a commanding lead in both the web browsing and light office benchmarks. 

Compilation, Compression, AVX Benchmarks on AMD Ryzen 3 4100 and Ryzen 5 4500

This section includes a diverse set of workloads, including exceedingly branchy code in the LLVM compilation workload and the massively parallel molecular dynamics simulation code in NAMD. Frankly, most of these types of workloads in this section aren't well-suited for this class of chip, but we include them as a reference.

The Core i3-12100 isn't as competitive as the Ryzen 5 4500 in these heavily-threaded applications. AMD also benefits in the SHA3, AES, and HASH benchmarks from its cryptographic optimizations.

• MORE: Zen 4 Ryzen 7000 All We Know
• MORE: Raptor Lake All We Know

AMD's new tactic of repurposing APU silicon for the lowest-end budget chips increased the company's ability to ship low-priced silicon during the shortages. However, the Zen 2 architecture is showing its age, leaving Intel with a clear lead in the sub-$125 chip segment.    

Below, we have the geometric mean of our gaming test suite at 1080p and 1440p and a cumulative measure of performance in single- and multi-threaded applications. We conducted our gaming tests with an RTX 3090, so performance deltas will shrink with lesser cards and higher resolution and fidelity settings.

As a reminder, the $122 Intel Core i3-12100 and $99 12100F offer identical performance, but the latter comes with the iGPU disabled. The Core i3-12100 is an insurmountable 49% faster than the $99 Ryzen 3 4100 and 21.7% faster than the $129 Ryzen 5 4500 in 1080p gaming, making the Core i3 the clear choice for budget gamers. 

The Core i3-12100 also dominates in single-threaded applications, being 44% faster than the Ryzen 3 4100 and 34% faster than the Ryzen 5 4500 in our overall measurement. In threaded work, the Core i3-12100 is 40% faster than the quad-core Ryzen 3 4100, leaving no justification for the 4100 for new builds. 

The six-core Ryzen 5 4500 is 3% faster than the 12100 in our overall measurement of performance in threaded workloads, but be aware that we see somewhat larger advantages in the 4500's favor in some rendering workloads. That might make the 4500 attractive if you're interested solely in productivity work or looking for an upgrade for a Zen 1 system. 

However, the Ryzen 5 4500's few advantages in threaded workloads come with severe tradeoffs in gaming and single-threaded applications. As we often see, mixed-use types of applications are common. Our Adobe tests are a good example — the 12100's superior mix of overwhelmingly faster single-threaded performance and more than sufficient threaded heft often takes the lead in those types of workloads. 

The Ryzen 5 4500 is also limited to PCIe 3.0 connectivity, a disadvantage compared to Intel's support for PCIe 5.0 — particularly for the productivity-minded, who might appreciate the faster throughput of a PCIe 4.0 SSD. The Core i3-12100 also supports DDR4 or DDR5 memory, while the Ryzen 5 4500 is limited to DDR4. That doesn't factor in much because DDR5's premium pricing isn't a good fit for this class of chips. 

The Core i3-12100 and the Ryzen models come with bundled coolers, but AMD offers fully unrestrained overclocking, while Intel limits you to memory overclocking only. That isn't too much of a factor, either, as we saw limited gains with all of these budget chips. 

The Core i3-12100 and 12100F remain the hands-down winners for budget gaming builds. Most budget builders are looking for the best all-rounder chip they can buy, and here the Core i3-12100 and 12100F unquestionably offer the best blend of performance in gaming, single- and multi-threaded work, along with more modern connectivity options.

The Ryzen 3 4100 doesn't make much sense at its $99 price point, even for upgrades. The Ryzen 5 4500 would make a decent upgrade for an existing budget AM4 build if you can find it at below MSRP, but be aware that the Zen 3-powered Ryzen 5 5500 is a better upgrade, and you can often find it on sale for only ~$15 more than the 4500. For new builds, the Intel Core i3-12100 remains the champion of the budget arena. 

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: Zen 4 Ryzen 7000 All We Know
• MORE: Raptor Lake All We Know

Update 09/14/2022 1:20 pm PT

Later in the day, wxnod published images of Lenovo Legion GeForce RTX 4090's cooling system. Indeed, the cooler is as sophisticated as it looks. It features two massive heatsinks with dozens of fins made of anodized aluminum that dissipate heat from 13 heat pipes. To put the number into context, GeForce RTX 3090/3090 Ti graphics boards featured 8 – 10 heat pipes, depending on the model. That said, we can speculate that the upcoming GeForce RTX 4090 indeed has higher power consumption and heat dissipation than the existing Nvidia flagship.

Original Article

As the expected launch date for Nvidia's first GeForce RTX 40-series graphics cards draws nearer — our best graphics cards list is about to get some new entrants — new leaks reveal more details about the latest GPUs. Meanwhile, alleged images of Lenovo's Legion GeForce RTX 4090 have appeared online, with subtle pixilation hiding the model number, but we can just make out a familiar font. The RTX 40-series news is bolstered by alleged specifications also surfacing. All this information comes from unofficial sources, so apply your usual does of skepticism and pull out a salt shaker. Regardless, it looks like Nvidia's GeForce RTX 4090 is really getting close. We should learn more at the GeForce Beyond event on September 20.

Lenovo's Legion graphics boards are designed primarily with the company's own gaming PCs in mind, but these cards sometimes find their way to retail as well. As Lenovo looks to be getting ready to roll out its Legion gaming rigs with Nvidia's upcoming flagship, one of those systems with a purported GeForce RTX 4090 inside was pictured and those images were published by nga.178.com and then republished by @wxnod.

Based on the most recent leaks of Nvidia GeForce RTX 4090 specifications by @kopite7kimi and VideoCardz, the board will feature the AD102 graphics processor with 16,384 CUDA cores and operating at up to 2520 MHz, 24GB of 24Gbps GDDR6X memory, and a 450W~660W power consumption. Such extreme power consumption requires a formidable cooling system, something that we see in the images of alleged GeForce RTX 4090 cards.

Elsewhere, there are rumblings of Nvidia releasing two versions of GeForce RTX 4080, one with 12GB and the other with 16GB, but we still have our doubts about the 12GB model. Wouldn't a card with fewer cores and less memory and memory bandwidth make more sense as an RTX 4070?

Rumored Nvidia GeForce RTX 40-Series Specifications

Lenovo's supposed Legion GeForce RTX 4090 board features the signature Legion design with straight lines and a grey aluminum shroud, coupled to a triple fan cooling system. The alleged GeForce RTX 4090 marking on board has been pixelized, though we still see that we are dealing with a '90' card, so we cannot guarantee that this is indeed Nvidia's top-of-the-range board. However, there are other signs that we are dealing with something new and not just a rehash of RTX 3090.

The 12+4-pin 12VHPWR auxiliary PCIe Gen5 power connector implies this is a next-generation graphics card, as so far just a few graphics cards with such power plug have been released (e.g., EVGA's GeForce RTX 3090 Ti Kingpin, Gigabyte's GeForce RTX 3090 Ti Gaming OC 24G, etc.). Meanwhile, some of Lenovo's previous-generation graphics boards used Nvidia's 12-pin power connector, whereas others relied on the proven two 8-pin power plug approach.

Another thing to note about Legion's alleged GeForce RTX 4090 board is the lack of NVLink connectors. Keeping in mind that this card is designed primarily for pre-built PCs, if Lenovo does not plan to offer multi-GPU systems to gamers, there is no need for NVLink connectors. Or perhaps Nvidia is finally taking the last step and ended SLI support completely with Ada? That would be an interesting change, and we've long since stopped recommending dual GPUs for gamers.

Other images published by @wxnod depict a box of Gigabyte's GeForce RTX 4090 graphics card with a picture of the board itself on the back of the box. The card looks to feature a massive quad-slot cooling system that is identical to one used by the company's GeForce RTX 3090 Ti Gaming OC 24G. While the package resembles Gigabyte's current style, it has brand-new artwork, which might be an indication that this is indeed intended for Nvidia's GeForce RTX 4090.

Just a few months ago, the best graphics cards remained difficult to find at "reasonable" prices. Nvidia launched the RTX 3090 Ti at a then-obscene $1,999 MSRP at the end of March, for example. Today, retail prices have dropped as much as 43% — close to half off! Other cards have become much more affordable as well, and only a handful of GPUs remain above their official MSRPs.

Where 2020 and 2021 were a perfect storm of converging factors, including the pandemic, supply chain issues, and skyrocketing cryptocurrency prices, we're now seeing an about-face in nearly all areas. Crypto prices and mining profitability are tanking, inventory of current generation cards is "too high," and most of the supply chain kinks are being worked out. With Nvidia RTX 40-series Ada GPUs expected to debut in the next month or so, and AMD's RX 7000-series RDNA 3 parts not far behind, manufacturers are doing their best to clear the way for the upcoming launches, especially at the high-end and extreme ends of the market.

Unlike our GPU price index, which focuses on eBay prices in monthly chunks, here we want to look primarily at retail prices on new graphics cards. We're specifically looking at US prices, and we mostly won't bother with any previous generation GPUs, though there are a few Nvidia Turing models (RTX 2060 and GTX 16-series) that remain in production and are thus included.

Here's the full rundown of current retail prices, as of September 6 — do note that all of these are prone to sometimes large fluctuations. For example, short-term sales (like Labor Day) can hack off another 5–10% from the prices shown here.

Compared to the August 2022 retail GPU prices, graphics cards have dropped an average of 7%. We haven't tracked retails prices for Nvidia's Turing cards, and the Intel Arc A380 is new to the list (along with the RX 6700 non-XT), but the more expensive cards show the largest deltas.

Nvidia's highest-end RTX 3090 Ti GPU now sells for 45% less than its official launch MSRP — and Nvidia itself leads the way, as the cheapest card is the RTX 3090 Ti Founders Edition from Best Buy. That's another 21% decrease since last month! The RTX 3090 costs basically just as much, while Newegg has the RTX 3080 Ti for $809.99 (after a $40 instant rebate) — another large 15% drop. The RTX 3070 through 3080 still sell for slightly more than MSRP, but they're getting closer and should soon fall below that mark, assuming they're not simply discontinued to make way for new models. Meanwhile, the RTX 3050 through 3060 Ti are still going for 12–20% more than MSRP. That's probably because Nvidia doesn't plan to replace the budget and midrange offerings with new RTX 40-series GPUs until next year.

Flipping over to the AMD side of things, AMD's GPUs have mostly been selling below MSRP for a couple of months now, but now even the RX 6800 XT and RX 6800 are joining the fun. The Radeon RX 6900 XT for $699.99 looks particularly tasty — there's also an MSI RX 6900 XT with a rebate for that same price, though we don't particularly like mail-in rebates. If you're hoping to pay less than $500, or maybe even closer to $250, the Radeon RX 6600 for $249.99 is almost impossible to beat right now. Nvidia's closest alternative is the previous generation RTX 2060 for $229.99 — it's actually faster for ray tracing, particularly with DLSS, but up to 20% slower in standard gaming performance.

However you slice it, the RTX 2060 and RX 6600 make the prospects for anything that costs less than $200 rather unappetizing. Yes, they cost more than cards like the RX 6400 and RX 6500 XT, but they're also up to twice the performance. If you want a cheap card that might actually improve in performance over the coming months, Intel's Arc A380 at $139.99 has some excellent video codec hardware, though it's currently backordered — and gaming performance can be a bit iffy at times.

GPU Prices: What to Expect Next Month

Again, there are multiple GPUs all slated to launch within the next few months. I think there's a good chance we'll see Nvidia release RTX 4090 and RTX 4080 cards by October, with the GTC 2022 keynote likely digging into some of the architecture. However, it's also possible Nvidia will start with the rumored 48GB Ada card, which could be a new Titan — with a suitably high (obscene) price. We'll find out on September 20.

AMD's plans aren't quite as clear. It demonstrated working RDNA 3 silicon during it's Ryzen 7000 announcement. While the CPUs will go on sale September 27, I think RDNA 3 graphics cards won't land at retail until late November, give or take a couple of weeks — basically a repeat of the RX 6000-series launch. I'd love to be wrong, though, and it would be far better for AMD to get RX 7000 cards out the door in October, so that they're available for the holiday shopping spree that takes place throughout November and December.

Intel's not going to be left out of the fun this year either, if the latest indications are anything to go by. The drivers continue to get regular updates, and the worst bugs and issues are being squashed. Rumors indicate an "imminent" launch for the Arc A770 and A750, and possibly A580 as well, which could happen before the end of the month — certainly by October. If pricing looks anything like the A380, Intel will be aggressive and try to get cards off the shelves rather than stagnating any further. Could we see better than RTX 3060 Ti performance from the Arc A770 for under $400? That would make the most sense, as Intel's still the newcomer to the graphics scene.

All of those upcoming GPUs mean the existing cards will, if anything, have to continue to drop in price. For high-end cards that should be eclipsed by RDNA 3 and Ada Lovelace offerings, we definitely recommend a "wait and see" approach. We may even see midrange cards drop further in pricing to clear inventory and perhaps keep people from buying Intel's new GPUs, but there's not much room to go lower than the $250 range.

It's definitely a halo product; it's large, expensive, and over the top in many use-case scenarios. Nvidia's flagship GPU, the RTX 3090 Ti, is $1,099 at Best Buy. You can theoretically get it even cheaper, but that involves either taking out a Best Buy Credit Card or already possessing one to add a 10% discount. You can find out where the RTX 3090 Ti ranks on our GPU Hierarchy to get an idea of its performance.

If you're looking for gaming laptops under $1,500, we have a couple of laptop deals for you today. How about Lenovo's Ideapad Gaming 3 (2022) gaming laptop for only $779, with an AMD Ryzen 5 6600H CPU and Nvidia RTX 3050 GPU inside?

Also, the Asus TUF Dash 15 2021 gaming laptop is $819 at the moment. The TUF Dash 15 in this model uses an intel 11th Gen CPU but the same RTX 3050 GPU. These are some of the cheapest prices we've seen for these laptops on Amazon, and a quick check on the Camelizer confirms these prices. 

For more of today's deals, check below. 

TL;DR — Today’s Best Deals

• Nvidia GeForce RTX 3090 Ti: was $1,999, now $1,099 at Best Buy
• Lenovo Ideapad Gaming 3 (2022) Gaming Laptop: was $899, now $779 at Amazon
• Asus TUF Dash 15 (11th Gen, RTX 3050) 2021: was $949, now $819 at Amazon
• Corsair K55 RGB Pro: was $59, now $34 at Amazon
• Asus TUF Gaming (model: ‎VG279Q1R) 27-inch FHD Gaming Monitor: was $249, now $166 at Amazon

Today’s best deals in detail

Looking for more deals?

• Best deals on tech & PC hardware
• Best gaming laptop deals
• Best monitor deals
• Best PC and laptop deals
• Best SSD deals
• Best CPU deals
• Best Dell and Alienware deals
• Best 3D printer deals
• Best Raspberry Pi deals

An alleged photograph of the Nvidia GeForce RTX 3090 Super has surfaced on the Chinese NGA forums (via MegasizeGPU). However, the individual who found this GPU checked it in GPU-Z, which was reported as a GeForce RTX 3090 Ti.

One of the curious things about the image of the Nvidia GeForce RTX 3090 Super is that its hourglass trim features a black color scheme on a black vented cooling shroud. The Nvidia Founders Edition design sports a highly polished chrome if you check the official product pages. The RTX 3090 Ti text is embossed into this chrome detail.

Another design difference between this slightly out-of-focus GeForce RTX 3090 Super picture and the RTX 3090 Ti concerns the fan design. Both these models feature a 9-blade fan, but the purported Super model has inset blades instead of the flush design on the Founders Edition with the center spindle.

From the GPU-Z screenshot, the RTX 3090 Super appears to have the correct number of CUDA cores, memory type/quota, and many other hallmarks of being an RTX 3090 Ti.

This Super release would be an odd one, as if Nvidia were going to begin to roll out an RTX 30 Super series, it would not naturally start with a top-tier model which many believed to be the last in the Ampere family. Moreover, the RTX 30-series (Ampere) had several Ti variants. However, this last hurrah of the Ampere GeForce line wasn’t released as the RTX 3090 Super, but instead as the RTX 3090 Ti.

As we begin September 2022, the RTX 3090 Ti has already fallen out of favor with consumers. However, it is a lot cheaper than it was nearer its launch date. The higher-end GPUs from Nvidia, and AMD, have seen massive price drops in recent weeks/months and the relatively new RTX 3090 Ti, which launched near $2,000 a few months ago, is now purchasable for $1,100 from our best graphics cards of 2022 feature page. Many eyes are now focused on the upcoming RTX 40-series (Ada Lovelace) graphics cards which should be unveiled in the future.

A well-known leaker with a good track record has published what the user claims to be the first image of Nvidia's upcoming GeForce RTX 4080 Founders Edition graphics card. The board (or rather its cooling system) seems brand-new, but since the devices are packaged into an antistatic bag, not many details can be seen.

Elysian Realm (@KittyYYuko) was relatively accurate with preliminary information about Nvidia's Ampere back in March 2020 (1, 2); now, the leaker is back with an alleged image of Nvidia's upcoming GeForce RTX 4080 Founders Edition board. Since the image comes from an unofficial source, albeit with a good track record, one should take it with a grain of salt. The photograph raises more questions than it provides answers.

The first thing that strikes the eye in the picture is that the board continues to use the design language of the GeForce RTX 30-series products (which are among the best gaming graphics cards today, so to speak). Yet, another thing that immediately attracts attention is the strange font used for the '4080' model number. It may not indicate that this is a fake image (and may imply that the board is authentic) as the font corresponds to the font Nvidia uses on its website, as noted by VideoCardz. Another oddity is that we do not see mounts of the bracket.

The card does not look like anything we have seen before. The board is dual-wide (like the RTX 3080), but it seems as tall and as long as the GeForce RTX 3090/3090 Ti. It also has a different fan with a refined blade design, which implies that this is a new board, not a GeForce RTX 30-series with a 4080 sticker. Meanwhile, the fan seems to be as big as the one on the GeForce RTX 3090 (which is logical) and is thicker than the one used by the dual-slot GeForce RTX 3080, which is why it sticks out more than it does on the RTX 3080.

From a previous leak covering Nvidia's upcoming GeForce RTX 40-series products, we know that the model RTX 4080 is supposed to be based on the AD103 graphics processing unit, which is not the biggest and the most power-hungry GPU in the lineup. Therefore, boards based on it do not require a substantial cooling system flagship boards like those powered by Nvidia's 'big ferocious' GA102 GPUs (aka BFGPUs). Meanwhile, AD103 is projected to be more power hungry than Nvidia's GA103/104 (which is used for GeForce RTX 3080 products); hence the usage of a 'taller' dual-wide heatsink seems to be justified.

In any case, while the card does not seem to be a Frankenstein created out of GeForce RTX 30-series Founders Edition parts, but is something new. Yet, we cannot state for sure that this is the upcoming GeForce RTX 4080 board.

Nvidia has yet to formally announce its GeForce RTX 40-series graphics cards based on the Ada Lovelace architecture. Still, it looks like the next-generation flagship — the GeForce RTX 4090 with 24GB of memory — might be already in production if leaked documents from one of the Chinese manufacturers are authentic and correct.

Well-known hardware blogger @harukaze5719 has discovered excerpts from documents that allegedly originate from a graphics card production facility. The documents supposedly describe production schedules, volumes, and other peculiarities of graphics board production at an unknown factory. Among the products listed, Nvidia's GeForce RTX 4090 card is mentioned next to the August 16 date and said to be 'expected in August.'

The excerpts of the documents claim that the card leverages the codenamed AD102 graphics processing unit (though without model/configuration suffix), carries 24GB of memory, and is equipped with four display outputs — three DisplayPorts and one HDMI.

We have no idea whether the documents are genuine, but at times, sources from China leak actual documents. Moreover, considering that we do not have access to these papers and are not sure what they describe the document, we would refrain from stating that a Chinese factory started to make Nvidia's GeForce RTX 4090-based graphics cards on August 16. However, the lack of any remarks about the number of boards to be made may indicate that it did not.

However, it is about time for manufacturers to start producing Nvidia's next-generation flagship graphics cards, or at least their samples based on final designs, if they want to get them to the market in September or October. Indeed, the absence of references about the amount of the GeForce RTX 4090 production might mean that the factory planned to build samples and expected to get AD102 chips and board components on September 16.

Like other companies, Nvidia never pre-announces its products or discloses its production schedules for competitive reasons. Yet, keeping in mind the number of GeForce RTX 40-related leaks that have emerged in the last couple of months, we would suggest that the new family could arrive rather sooner than later unless Nvidia decides to delay it by a few weeks in a bid to allow its partners to sell off their existing GeForce RTX 30-series stock, which is still among the best graphics cards around.

Nvidia’s next-generation GeForce RTX 40-series ‘Ada Lovelace’ graphics cards are rumored to be mere weeks away, going up against the best graphics cards, but they are not going to be cheap initially. If you want something cheaper, it may be time to consider buying a GeForce RTX 30-series ‘Ampere’ board, as they are reportedly going to get even cheaper in the coming weeks.

Nvidia’s add-in-board (AIB) partners are planning to cut prices of GeForce RTX 30-series products already later this month, according to reports from China-based MyDrivers and Zol that cite their own sources (via VideoCardz). At present, it is unknown how significant price cuts are going to be and which cards will get cheaper, but the reports indicate that they will get more affordable than they are today. In fact, promotional campaigns for higher-end GeForce RTX 3080 Ti/3090/3090 Ti that will put them hundreds of dollars below MSRP for a limited time are also planned.

Both AMD and Nvidia are expected to introduce their next-generation Radeon RX 7000-series and GeForce RTX 40-series GPUs in the coming weeks or months. Before doing so, they need to sell off existing inventory, which is why both companies are going to slash prices on existing graphics cards. Since Nvidia appears to have more GPUs in stock, price cuts on GeForces are expected to be more aggressive when compared to price reductions of AMD's Radeon products, though AMD's GPUs are already selling well below MSRP in most cases.

It should be noted that Nvidia slashed prices of its higher-end GeForce RTX 30-series boards in mid-July. In fact, many of those boards are sold below their new MSRPs right now. To add some context, here are the existing EVGA.com prices when compared to recommended prices from mid-July as well as original MSRPs.

Keeping in mind that Nvidia will start rolling-out its GeForce RTX 40-series with the higher-end models, it's logical for the company's partners to sell off GeForce RTX 3080 Ti/3090/3090 Ti first, which is why these products are now below MSRPs. It also helps that those were all severely overpriced to begin with. If Nvidia's AIB partners still have excessive boards in stock before the green company launches its Ada Lovelace lineup, we expect these products to get even cheaper, which is what two Chinese websites are reporting.

It is noteworthy that some of EVGA's midrange GeForce RTX 30-series boards are sold at prices that are higher when compared to Nvidia's recommended prices. Considering the fact that we are months away from mid-range GeForce RTX 40-series, there is no current need to sell them below MSRP. Several weeks down the road, Nvidia's AIB partners may reconsider dropping their prices on mid-range GeForce RTX 30-series devices as well.

Nvidia and its allies will not confirm planned prices cuts before they start to sell at those prices, so we do not expect any official information to emerge right now. Meanwhile, we do see rational to cut prices of GeForce RTX 30-series ahead of next-generation launch, so it is reasonable to expect existing boards to get cheaper in the coming weeks.

Now that the Chips and Science act is signed into law, companies that are supposed to get subsidies on chip development and production in the U.S. can start proposing plans for their expansion. For example, Micron and Western Digital, two leading makers of DRAM and NAND memory, propose setting up a memory research and development coalition in the U.S. and then producing innovative types of memory in the country.

Bringing Memory Development and Production Back to the U.S 

While the U.S. produces only about 12% of the global semiconductor output, companies like Intel, GlobalFoundries, and Samsung Foundry produce relatively advanced chips in the country. They have been used worldwide since many of them are unique. But regarding computer memory production, the U.S. is far behind South Korea, Japan, and Taiwan mainly because of North America's high fab and labor costs. It is a bit odd as many innovative memory technologies and fabrication processes are developed either in the U.S. or in the U.S. and Japan. With funds provided by the Chips and Science act, Micron and Western Digital hope to correct this wrong.

"In order to secure U.S. leadership in the critical area of semiconductor memory and storage technology, the NSTC should develop and articulate a long-term (>5 years) vision and roadmap for the enablement of the next generation of these technologies," a joint statement between the two companies reads.

One of the public-private organizations that will be set up with the help of the Chips and Science funding as well as investments from interested parties will be the National Semiconductor Technology Center (NSTC), which will bring together industry, government, national labs, and academia to conduct advanced semiconductor research and prototyping, according to the Semiconductor Industry Association. For example, Micron and Western Digital propose to form the Memory Coalition of Excellence (MCOE), which will be a part of the NSTC and will focus entirely on new memory technologies. In addition, both companies expect to get subsidies to build new manufacturing capacity in the USA.

"A Memory Coalition of Excellence (MCOE) will support this era of transformation and new technological innovations required," the statement by Micron and Western Digital reads. "This MCOE should be a focused effort across industry, academia, and government with clearly defined objectives related to overcoming the challenges outlined in this paper and should be aligned with other key coalitions of excellence (COEs) to support the overall objectives of the NSTC."

New Types of Memory Required, Let's Work Together 

An avid reader of Tom's Hardware would justly note that 3D NAND and DRAM are commodities whose prices fluctuate and significantly affect the profitability of manufacturers, which is why it is crucial to make such types of memory in regions with the lowest costs and, obviously the U.S. hardly belongs there. Furthermore, both Micron and Western Digital already have DRAM and 3D NAND R&D operations in Japan, which is why MCOE in the USA may seem excessive. But there are several factors to be considered.

Firstly, next-generation compute devices, which Micron and Western Digital call domain-specific architectures (DSAs), will require all-new types of memory. In particular, the two memory manufacturers mention general purpose compute architectures that use different kinds of memory, accelerator-aware designs that use high-speed memory, and memory-centric architectures that tightly wed compute (logic) and memory. While traditional architectures will continue to use things like 3D NAND, DRAM, and HBM, emerging architectures will need new types of memory, which will have to provide across-the-board benefits in a variety of device metrics, including performance, power, area, functionality, cost, and complexity, according to the companies.

Secondly, those new memory types do not exist today, which means that companies like Micron and Western Digital have to invest in fundamental memory research in a bid to design them eventually. For example, Western Digital has been investing in ReRAM — a perspective storage class memory (SCM) type — for years, but without success. Meanwhile, Micron has not even adequately commercialized 3D XPoint that it co-developed with Intel. The industry has identified several promising technologies for SCM applications (PCM, MRAM, FeRAM, etc.), but none of these technologies has become widespread. That said, it makes a lot of sense to develop fundamental technologies behind innovative types of memory by R&D consortiums/coalitions in collaboration with academia to share costs and speed up time-to-market.

Thirdly, in many cases, innovative types of memory will require brand-new materials, new device structures, manufacturing technologies, and several other things that need heavy investments in fundamental research. Again, coalitions between commercial companies and academia generally represent a more comprehensive approach to basic science than in-house R&D operations. For example, IBM and SUNY Poly jointly conduct next-generation semiconductor R&D.

Fourthly, bringing memory close to compute logic is a significant challenge, so 2.5D and 3D packaging technologies will gain importance in the future. Developing them collaboratively will be beneficial for everyone.

Finally, there are fundamental challenges in developing and producing chips using leading-edge fabrication technologies. Therefore, Micron and Western Digital assert that it is essential to jointly develop simulation (TCAD) and electronic design automation (EDA) tools to speed up the development of next-generation solutions based on new materials, structures, and design. Furthermore, it is vital to accelerate the growth of innovative production equipment technologies, such as EUV mask and wafer patterning solutions, to increase the productivity of EUV tools and new metrology and materials analysis/characterization tools.

Micron and Western Digital expect MCOE to focus on the following activities:

• Advanced 'pre-competitive' R&D for materials, manufacturing technologies, and analysis techniques.
• Memory technologies for memory-centric computing (including in-memory compute, near-memory compute, and analog compute).
• Innovative 3D memory technologies.
• Advanced packaging and stacked memory solutions.
• Heterogeneous integration (functional and/or physical) at wafer and chip level.
• X-point array integrated with advanced CMOS for new concept validation. 
• Modeling methodologies and tools for rapid development and co-optimization of complex technologies and systems.

In general, while companies like Micron and Western Digital can come up with proprietary types of memory addressing specific applications (or rather DSAs, as they put it), they want to conduct fundamental R&D as well as production enablement together with other American companies. Therefore, if the NSTC and the MCOE organizations are successful, they will improve the competitive positions of participants and, consequently, the U.S. semiconductor industry in general.

Leading-Edge Memory Fabs Coming to the U.S. 

Discovering innovative materials and developing advanced memory technologies in the U.S. is one thing. But bringing memory production to the USA is a different challenge. On Tuesday, Micron announced plans to invest $40 billion in leading-edge memory manufacturing operations in the U.S. by the end of the decade.

"This legislation will enable Micron to grow domestic production of memory from less than 2% to up to 10% of the global market in the next decade, making the U.S. home to the most advanced memory manufacturing and R&D in the world."

Micron did not specify what kind of memory it plans to produce in the USA (e.g., 3D NAND, DRAM, SCM, etc.), but we can speculate that the company will attempt to create premium types of memory in the States.

Meanwhile, over the next seven or eight years, $40 billion invested in fabs could buy Micron one EUV-enabled fab that could produce virtually all types of memory or even a couple of moderate-sized EUV and High-NA EUV-enabled fabrication facilities (though we are speculating). In any case, it looks like Micron is looking forward to bringing leading-edge process technologies to the U.S. in a bid to make advanced and expensive memory devices.

Nowadays, manufacturers in South Korea, Japan, and Singapore use the most advanced memory fabrication processes. Micron plans to reverse this trend.

Micron has quietly started volume production of GDDR6X memory chips featuring a 24 GT/s data transfer rate. The new memory devices are expected to be used by some of Nvidia's upcoming high-end GeForce RTX 40-series graphics cards based on the Ada Lovelace architecture. 

Micron's new 24 GT/s GDDR6X chips have a capacity of 16Gb (2GB) and can be used to build graphics cards with loads of memory onboard (via @Harukaze5719). It is expected that these ICs will be used for Nvidia's GeForce RTX 4090 graphics card, carrying 24GB of GDDR6X memory connected to the GPU using a 384-bit memory bus (and therefore offering a peak memory bandwidth of 1.152 TBps, assuming that Nvidia will use memory at its highest rated speed). Other products from the lineup are rumored to use slower GDDR6X devices with an up to 21 GT/s data transfer rate. 

In addition, the 24 GT/s GDDR6X memory chips are projected to be used by Nvidia's next-generation flagship graphics board currently known as The Beast (sometimes referred to as the GeForce RTX 4090 Ti), which will carry 48GB of memory on both sides of the card (and will likely be the best graphics card available in terms of raw performance in the coming months).  

Nvidia is rumored to formally introduce its GeForce RTX 40-series 'Ada Lovelace' products this fall. Therefore, it is about time for Micron to kick off volume production of its next-generation GDDR6X memory chips with an up to 24 GT/s data transfer rate. 

Neither Nvidia nor Micron have officially commented about usage scenarios of 24 GT/s GDDR6X memory chips, so take unofficial information about Nvidia's GeForce RTX 40-series lineup with a grain of salt. Meanwhile, Micron is certainly starting production on this memory that will be used by one of its clients. At present, the only company to support GDDR6X is Nvidia. 

It is noteworthy that Samsung has been sampling its 24 GT/s GDDR6 (not GDDR6X) memory chips since December, 2021. So Micron has some speedy competition. 

In an unpredecented move, EVGA (via FrameChasers) has slashed $1,000 of the brand's custom GeForce RTX 3090 Ti models. The GA102-powered graphics cards is one of the best graphics cards right now, which debuted early this year at $1,999.

The GeForce RTX 3090 Ti FTW3 Gaming, which is the least expensive SKU on EVGA's lineup, now sells for $1,149, 47% cheaper than what the company used to sell it for. At $1,149, the GeForce RTX 3090 Ti FTW3 Ultra Gaming is also one of the most affordable custom GeForce RTX 3090 Ti models on the market. It's no push-over on the performance side, either. The graphics card flaunts a 1,890 MHz boost clock.

If you need something with a bit more firepower, the GeForce RTX 3090 Ti FTW3 Ultra Gaming retails for $1,199 on EVGA's store. It carries a $50 premium over the non-ultra model but offers a 30 MHz higher boost clock speed. The GeForce RTX 3090 Ti FTW3 Ultra Gaming is the second fastest model in the lineup, with a factory overclock that pushes the boost lock to 1,920 MHz.

EVGA GeForce RTX 3090 Ti

The GeForce RTX 3090 Ti FTW3 Ultra Gaming is also available with a hybrid cooling design. The GeForce RTX 3090 Ti FTW3 Ultra Hybrid Gaming goes for $1,499, $300 more expensive than the air-cooled counterpart. In addition, the graphics card arrives with a beefy 360mm AIO cooler for maximum heat dissipation.

Even the exclusive GeForce RTX 3090 Ti K|NGP|N Hybrid Gaming is one sale, but unfortunately, it's out of stock. The over-engineered graphics card, which caters to extreme overclockers, used to carry an eye-watering price tag of $2,499. EVGA has lowered the price to $1,999, which still costs more than your average high-performance gaming PC. For what it's worth, EVGA includes a "free" SuperNOVA 1600W P2 power supply.

EVGA's GeForce RTX 3090 Ti graphics cards come with Nvidia's latest "Face Your Demons" bundle that consists of Ghostwire: Tokyo, Doom Eternal, and Doom Eternal Year One Pass (Includes Doom Eternal: The Ancient Gods Part One & Part Two and Battlemode).

Graphics card pricing started to plummet a few months ago as Ethereum is nearing its transition to a proof-of-stake (PoS) blockchain. Cryptocurrency miners have started to offload their hardware on the second-hand market to cut losses, whereas scalpers have also lowered pricing on platforms like eBay. EVGA is probably attempting to get rid of Ampere inventory while the company still can. Pricing will most likely continue to improve in the upcoming months as Nvidia readies its GeForce RTX 40-series (Ada Lovelace) graphics cards and AMD answering the call with its RDNA 3 graphics cards. Let's not forget that Intel also has Arc Alchemist, which the chipmaker expects to bring to the market in the summer.

More details about what will probably be marketed as Nvidia’s GeForce RTX 40 series has seeped through the cracks. Serial leaker Kopite7kimi today shared some insight into the full-fat AD104 GPU. This GPU is probably destined for an upcoming RTX 4070 Ti graphics card. According to the leaker, with a good track record with regard to these things, this next gen 70 Ti product will “easily match RTX 3090 Ti.” As ever with leaks, take the news with a pinch of salt.

To see if Kopite7kimi’s assertions hold water, it is worth constructing a comparison table of the specs we know about the current gen champ, vs the purported specs of the upcoming RTX 4070 Ti. Who knows, the RTX 4070 Ti may just make be a future contender for our best GPUs.

Most of the above AD104 specs have come from our ‘everything we know’ article about the Nvidia Ada Lovelace and GeForce RTX 40-Series. Of course the RTX 3090 Ti specs are common knowledge.

If you consider the table above, the upcoming Ada Lovelace GPU clearly has advantages in transistor count, and a big change in process node. It is also expected to be pushed to significantly higher clock speeds compared to the RTX 3090 Ti, which is why the 4070 Ti presumably gets such a high TDP. High TDP averse folk might look at the regular RTX 4070 with a purported TDP nearer to 300W. However, the purported GeForce RTX 4070 Ti should exceed or draw parity to the RTX 3090 Ti with a much lower bill of materials (e.g. smaller silicon die) and help bring 2022 flagship performance to much more affordable levels.

One area where the previous gen graphics card is much stronger, is in the memory department. They are both seemingly use GDDR6X at 21Gbps, but the expected configuration of the RTX 4070 Ti has half the memory and half the bus width, which again will help bring down costs. There will be a certain resolution and quality setting in every game where the greater memory bandwidth of the RTX 3090 Ti will make itself known, but we need thorough comparative testing in a host of games to find it.

In summary, it is exciting to think of Ampere flagship+ performance coming to the almost ‘midrange’, but we have some worries about power and cooling. AMD with its Radeon 7000 series and 5nm RDNA 3 GPUs, seems to be keeping a lid on power consumption – so the upcoming graphics card battle is going to be pretty interesting.

According to a Tweet posted by leaker @kopite7kimi earlier this morning, the "estimated" performance of Nvidia’s upcoming RTX 4080 and RTX 4070 GPUs in 3DMark’s TimeSpy Extreme benchmark looks incredibly promising. Projections say the RTX 4080 will be faster than the RTX 3090 Ti, and the RTX 4070 will be just as fast as the RTX 3080 10GB — one of the Best Graphics Cards on the market.

As usual, take these results with a grain of salt. In fact, Kopite says they aren’t fully confident in the data points shared, but the leaker has a stellar record for accuracy. We've been making similar estimates for a while now, based on rumored specs, but it's not clear if Kopite actually saw some real performance data and merely obfuscated it to protect the (not so) innocent, or if these are purely guesses.

For reference, Kopite already shared RTX 4090 Time Spy Extreme data a week ago, which indicates GPU performance to be around 30% quicker than the fastest performing RTX 3090 Ti’s — overclocked and with liquid cooling. With that data as a starting point, Kopite’s RTX 4080 and 4070 figures aren’t all that shocking.

Kopite believes the RTX 4080 will hit a very impressive Time Spy Extreme graphics score of 15,000 points, a touch faster than the RTX 3090 Ti at normal frequencies. In Time Spy Extreme specifically, this data would make the RTX 4080 40% to 50% faster than its predecessor, the RTX 3080 10GB, which would be a massive generational improvement. If GPU prices for the 4080 are remotely similar to the 3080, customers will be getting a very powerful graphics card for the money.

The RTX 4070 estimates aren’t as impressive, but it is still a healthy gen-on-gen improvement nonetheless (if true). Kopite believes the RTX 4070 will hit 10,000 points in Time Spy Extreme, which coincides perfectly with most of the RTX 3080 10GB results in 3DMark’s results browser. Compared to the 3070, this translates into a 25% performance improvement and performance equal to that of the RTX 3080.

We've long suspected that Nvidia's RTX 40-series architecture will be a monster — for the RTX 4090 and 4080 in particular, it will almost certainly bring record-shattering generational performance improvements. Still, it will come at the cost of power consumption. The biggest unknown now is GPU pricing, which could swing in any direction depending on shortage issues and the 40-series release date. Performance only goes so far, unless you can get it at a good value.

For a deeper dive into all that we know about the RTX 40-series 'Ada' architecture, check our existing article. Also note that 3DMark only represents one synthetic graphics benchmark, and what we really want to see are real-world gaming benchmarks. All indications are that we should have hardware in hand by October, possibly sooner.

Nvidia is projected to release three high-end GeForce RTX 40-series graphics cards based on the Ada Lovelace architecture later this year targeting demanding gamers. But apparently Team Green is also prepping a rather monstrous graphics board based on its top-of-the-range AD102 GPU that will carry 48GB of memory with a typical board power of 800W, clearly gunning for the top of the GPU benchmarks hierarchy and the fastest of the best graphics cards.

Hardware leaker @Kopite7kimi dubbed Nvidia's flagship GeForce RTX 40-series graphics card 'The Beast' without attributing a model name to it, but normally Nvidia would call it GeForce RTX 4090 Ti or Titan. This board is said to carry a GPU with 18,176 CUDA cores (up from 16,384 CUDA cores in case of the GeForce RTX 4090) as well as 48GB of GDDR6X memory featuring a 24 GT/s data transfer rate (up from 24GB of memory at 21 GT/s in case of the RTX 4090).

Considering the higher number of CUDA cores, additional memory, and substantially higher TBP, the alleged GeForce RTX 4090 Ti should be significantly faster than the alleged non-Ti variant, particularly in high resolutions with high detail settings. Meanwhile, the beasty card is said to feature a TBP of 800W and will need an extremely capable power supply unit as well as a sophisticated cooling system.

The rather extreme thermals of the monstrous Ada Lovelace card might cause consternation among ProViz professionals that work with highly detailed 3D models that can actually benefit from 48GB of GDDR6X memory onboard. Cooling an 800W card would inevitably mean more fans and higher fan speeds, neither of which are pleasant in a professional environment (outside of the data center).

What remains to be seen is how much will Nvidia charge for its 'Beast' graphics board with 48GB of memory. A fully-fledged AD102 chip is projected to cost a lot, and 48GB of GDDR6X memory should carry an extreme price tag too. Given the rumored specs, we expect Nvidia's alleged GeForce RTX 4090 Ti board will feature an MSRP above $2000, the price of the outgoing GeForce RTX 3090 Ti.

By way of comparison, even if Nvidia keeps clock speeds relatively consistent between generations, RTX 4090 Ti could have over 70% more computational power than the RTX 3090 Ti. Memory bandwidth meanwhile may only be 14% higher, so Nvidia would need improved caching or memory compression technologies to scale performance. Regardless, it looks like a massive bump in performance — and power use.

Nvidia has not yet formally announced its GeForce RTX 40-series graphics boards, so consider everything detailed here as rumors and speculation. This is certainly not the first nor the last time we'll hear some juicy rumors about Nvidia's upcoming Ada Lovelace solutions. The real question: What will it punish more, your wallet or your home circuits?

One of the most powerful graphics cards on the market at the moment has a huge chunk of money knocked off its current price for a limited period, or until stocks last. The Zotac Gaming GeForce RTX 3090 Trinty OC is only $999 on Woot! right now, making it almost $300 cheaper than the nearest priced RTX 3090 model. This card has a whopping 24GBs of GDDR6X VRAM — that's more than enough for any task, be it gaming or video editing.

Gigabyte's Z690 Ultra Durable AX is $189 on Amazon, this is the DDR4 version of the Z690 and is a much more affordable option because the price of DDR5 is still being rather costly. 

Another great spot today is this LG UltraGear 27GP850-B 27-inch gaming monitor for $296. With a 165Hz refresh rate, 1ms GtG response time, and a QHD nano IPS display, this monitor has some very good specs for this reduced price.  

Keep scrolling for more great deals.

TL;DR — Today’s Best Deals

• Zotac Gaming GeForce RTX 3090 Trinty OC GPU: was $1,399, now $999 at Woot!
• Gigabyte Z690 UD AX (DDR4) Motherboard: was $219, now $189 at Amazon
• LG UltraGear 27GP850-B 27-Inch 165Hz QHD Nano IPS Gaming Monitor: was $446, now $296 at Adorama
• Corsair iCUE 220T RGB Airflow ATX Case: was $124, now $79 at Newegg
• MSI Gaming Radeon RX 6400 GPU: was $179, now $149 at Amazon

Today’s best deals in detail

Looking for more deals?

• Best deals on tech & PC hardware
• Best gaming laptop deals
• Best monitor deals
• Best PC and laptop deals
• Best SSD deals
• Best CPU deals
• Best Dell and Alienware deals
• Best 3D printer deals
• Best Raspberry Pi deals

According to our resident Twitter leaker @kopite7kimi, Nvidia's upcoming GeForce RTX 4090 cracked the 15,000 point barrier in 3DMark Time Spy Extreme by a long shot, with a record-breaking graphics score of 19,000 points in the famous benchmark. Furthermore, if Kopite's news is accurate, it puts the RTX 4090's performance barrier well ahead of anything available today, including RTX 3090 Ti GPUs chilled on liquid nitrogen.

For some perspective, the current reigning champion of the 3DMark TimeSpy Extreme benchmark is user "biso biso," with an LN2 cooled EVGA RTX 3090 Ti Kingpin Edition graphics card punching out a world recorded graphics result of 14,611 points.

Again, if Kopites data is real, engineering samples of Nvidia's 4090's are already hitting TimeSpy Extreme scores 30% higher than the highest-overclocked RTX 3090 Ti's running today (not to mention normal RTX 3090 Ti's). As a result, we could see even higher scores once the RTX 4090 hits the market.

This data seems to confirm what we've already heard in the past: that Nvidia's RTX 40 series lineup will feature one of the most significant performance jumps we've seen in a single GPU generation from Nvidia.

Current rumors speculate that the top die for the 40 series, AD102, will pack 71% more CUDA cores and SMs than Ampere's top die GA102. In addition, it will reportedly feature similar or higher clock speeds compared to RTX 30 series, thanks to a more efficient TSMC 5nm process. We don't expect the rest of Nvidia's 40 series dies to pack the same core count increase, but they all are expected to pack a lot more cores if 71% is the ceiling for the 40 series.

Power consumption is also rumored to go up extensively, with flagship 40 series cards rumored to hit as much as 500 to 600 watts of power consumption.

As a result, high-performance gains should be expected to compensate for the crazy high power requirements for the 40 series. However, take this news with a grain of salt since we don't have full confirmation on this data. But it does have merit based on the current information that we have on Nvidia's next-generation GPUs.

Without much fanfare, Nvidia slashed the prices of its GeForce RTX 30-series graphics cards this week. Depending on the exact model, Ampere graphics boards are now $100 to $500 cheaper than they were just days ago. It is believed that Nvidia is lowering prices to reduce the inventory of its current-generation products ahead of the imminent launch of next generation GeForce RTX 40-series 'Ada Lovelace' GPUs. 

New Prices

Benchlife (via VideoCardz) was first to report the alleged price cut without any formal confirmation from Nvidia or its add-in-board (AIB) partners early on Thursday. Based on the information from Benchlife, Nvidia's GeForce RTX 3090 Ti will now cost $1,499 (-$500), GeForce RTX 3090 will be priced at $1,299 (-$200), and the GeForce RTX 3080 Ti will carry a $1,099 (-$100) price tag. Furthermore, the GeForce RTX 3080 12GB finally received its official MSRP — $799. Meanwhile, other boards in the Ampere lineup for consumers have not seen any price cuts so far.

Alleged New GeForce RTX 30-Series MSRPs

By now, multiple retailers in the U.S., as well as EVGA, have begun offering high-end Nvidia GeForce RTX 3090 Ti/3090/3080 Ti/3080 12GB graphics cards at new prices that are $100 to $500 lower than they used to be a week ago (remember that we are talking about boards that are still among the best graphics cards around). Meanwhile, all the boards are sold at a discounted price or with an instant rebate, so there is no formal confirmation of Nvidia's price cut.

For example, over at EVGA.com, all GeForce RTX 3090 Ti (except the Kingpin version) and GeForce RTX 3080 Ti graphics boards are priced at $1,499 and $1,099 (respectively) no matter how high they are clocked and what cooling they use. For instance, EVGA's GeForce RTX 3090 Ti FTW3 Ultra Hybrid Gaming received a $700 instant rebate, whereas the GeForce RTX 3080 Ti FTW3 Ultra Hydro Copper Gaming comes with a $430 instant rebate. Other high-end boards from EVGA's GeForce RTX 30-series lineup are also sold at their new MSRPs at EVGA.com. Meanwhile, the remaining SKUs from EVGA's Ampere family are sold below their official prices, but for the sake of truth, it has to be said that their price tags are higher than those recommended by Nvidia.

Retailers like BestBuy and Newegg also slashed prices of their higher-end Nvidia GeForce RTX 30-series boards. Meanwhile, Newegg offers not only EVGA's boards at their new MSRP but also GeForce RTX 3090 Ti graphics cards from Gigabyte and Zotac at $1,499, while Gigabyte's GeForce RTX 3080 Ti is available for $1,099. 

BestBuy even calls its new prices a 'clearance' sale, so perhaps it does not expect to receive new batches of certain boards. Meanwhile, entry-level, mainstream, and performance mainstream boards are still sold at prices that exceed their MSRPs, but at least they are significantly cheaper than they used to be weeks ago.

Ada Lovelace Looming

The new MSRPs of higher-end GeForce RTX 3090 Ti/3090/3080 Ti/3080 12GB graphics cards indicate that Nvidia wants to sell them off first before it launches its new GeForce RTX 40-series products later this year. This is logical as enthusiast buyers do not tend to buy previous-generation products.  

The fact that Nvidia has not 'semi-officially' cut prices on products that sit below the RTX 3080 12GB model may show that the first Ada Lovelace boards will address premium market segments with graphics boards that cost $799 and above (but this is, of course, our speculation). 

But even though it does not look like Nvidia is formally cutting down prices of GeForce RTX 3080 10GB and slower products, they are still sold with discounts both by EVGA (which is a very close partner of the green company and which market tactics usually reflects Nvidia's plans) and retailers. So we can conclude that Nvidia and its AIB allies want to get rid of those products sooner rather than later.

What remains to be seen is whether aggressive price cuts on the high-end of the GeForce RTX 30-series lineup signal that Ada Lovelace may be closer than we might think. 

EVGA's Kingpin (stylized as K|NGP|N) series graphics cards have always been a little special. They use a custom printed circuit board (PCB) with an enhanced voltage regulating module (VRM) along with a hybrid or liquid cooling system to ensure maximum performance. But with its GeForce RTX 3090 Ti, EVGA has gone one step further: it bundled a rather formidable 1600W power supply with the board. And now we see a special converter board for 12VHPWR.

EVGA's 1.6kW PSU lacks the 12VHPWR PCIe Gen5 auxiliary power connectors that EVGA's GeForce RTX 3090 Ti Kingpin graphics board uses, which is why EVGA had to create its special PowerLink 52U five 8-pin to two 16-pin (12VHPWR) power adapter exclusively for its flagship card (which will be among the best gaming graphics cards ever produced until the RTX refresh hits), Jacob Freeman, a spokesman for EVGA, has revealed in his Twitter. 

The EVGA GeForce RTX 3090 Ti Kingpin uses two 12VHPWR PCIe Gen5 auxiliary power connectors that can deliver up to 1200W of power, an amount that the board can barely devour even when heavily overclocked. Nonetheless, to make its overkill VRM work properly, EVGA wants to feed both connectors with enough power. Apparently the five 8-pin PCIe auxiliary power plugs (that can deliver 750W) can do the job, albeit using a special adapter.

EVGA's PowerLink 52U will certainly simplify cable management for existing GeForce RTX 3090 Ti Kingpin owners since messing with cabled adapters is not particularly user friendly. What remains to be seen is whether the same adapter will be supplied with future GeForce RTX 40-series graphics cards (or sold separately) since they will use 12VHPWR PCIe Gen5 power connectors and will not be compatible with existing PSUs out-of-box. 

Common logic says that since next-generation graphics cards will still be power hungry and will need new power connectors, there will be adapters for these boards to make them work with existing high-wattage power supply that will still be good for upcoming graphics adapters.

Today at Amazon, the EVGA GeForce RTX 3090 FTW3 Ultra Gaming Graphics Card has dropped down to its lowest price ever. Its recommended price is $1919 but it’s been going for around $1750, as of late. Right now, users can take it home for just $1349 with free shipping available for Amazon Prime members.

As of writing, no expiration has been specified for the discount. This is one of the best prices you can find right now for this particular RTX 3090 card but if you’re looking for something else, you should check out our list of Amazon Prime Day tech deals for more discounts throughout the weekend as we get ready for Prime Day.

Users have three DisplayPort outputs and one HDMI port to take advantage of for video output. It’s cooled using iCX3 technology and has a total of three fans with hydro dynamic bearings. There are addressable RGB LEDs inside for added visual flare.

It offers real-time ray tracing for top-of-the-line graphics performance while gaming. The unit measures in at 5.4 x 11.8 x 2.8in and features an all-metal backplate. Overall, this is a beefy graphics card and at $400 off, it’s hard not to take a closer look at the offer.

Visit the EVGA GeForce RTX 3090 FTW3 Ultra Gaming Graphics Card product page at Amazon for more details and checkout options.

If you’re on the fence about locking in this new graphics card, take a look at our list of best deals on tech as we gear up for Amazon Prime Day. There are plenty of vendors taking part in lowering prices in response to the sale, so you can expect monitor deals, gaming PC deals and even more specific components like CPU deals and SSD deals from more than just Amazon. Check back throughout the weekend and Prime Day as we continue to share the best offers we can find.

After months of rumors, EVGA has finally released its top-tier overclocking-focused variant of the GeForce RTX 3090 Ti under the Kingpin Hybrid Gaming branding. This card caters to extreme GPU overclockers but competes with the best graphics cards on the market. The Kingpin GeForce RTX 3090 Ti sells for an eye-watering $2,499.99, but to sweeten the deal EVGA has bundled a free 1,600W SuperNova P2 power supply to the 3090 Ti, which is an exceptional deal considering it costs $399.99 all by itself.

Coincidentally, EVGA is also advertising the same deal with the GeForce RTX 3090 Ti FTW3 Ultra Gaming for a limited time with even more stuff to celebrate EVGA's 23rd anniversary. With the FTW3 deal, you get a free EVGA SuperNova 1600 P+ PSU, a 23rd-anniversary shirt, and three months of Xbox Game Pass for PC. It is even better, considering this deal with the GeForce RTX 3090 Ti is on an attractive discount of $1,899.

The RTX 3090 Ti Kingpin is the most advanced model from EVGA's lineup. The card features a 360mm AIO liquid cooler for the GPU and a secondary heatsink and fan for cooling PCB components on the card, such as the power delivery system. Aesthetically the card is decked out in a matte black finish with a large silver accent, with all three fans on the AIO featuring RGB lighting. For the select few customers who won't be putting an LN2 pot or custom water block on this card, you will be greeted by EVGA's K|NGP|N logos everywhere to make it stand out from the rest of EVGA's hybrids. An OLED display is also mounted on the side of the card to show off unique animations or pictures.

The most significant selling point of the Kingpin is its incredibly overbuilt power delivery system and very high-quality componentry. In addition, the Kingpin is one of the only GeForce RTX 3090 Ti models to feature dual 16-pin power connectors, giving the card a whopping 1,200W power for overclocking.

It is highly doubtful that the GeForce RTX 3090 Ti core will ever take advantage of 1,200W of power or anything close to that. However, the twin 16-pin connectors ensure the card will never be power throttled under any scenarios, including liquid nitrogen overclocking.

To nobody's surprise, the GeForce RTX 3090 Ti Kingpin also comes with the highest factory overclock of any EVGA GeForce RTX 3090 Ti, with a 1,950 MHz boost clock by default. However, with its monstrous cooling solution, you should be able to overclock the card well past that limit for regular gaming use.

Intel's competitive Alder Lake chips have upset AMD's dominance with its Ryzen processors, particularly in the lower price ranges, but the new $199 six-core 12-thread Ryzen 5 5600 and $159 Ryzen 5 5500 are designed to plug key gaps in the company's portfolio. These new chips come as part of AMD's broader launch of seven new Ryzen 5000 models that aim to shore up the company's rankings in CPU benchmarks and retake its position on the Best CPUs for gaming list.

AMD's Ryzen 5000 chips fully eclipsed Intel's performance lead in desktop PCs when they launched back in 2020, but the company has long neglected to launch any sub-$250 chips with the potent Zen 3 architecture. That's kept the bar for entry unattainably high for value seekers.

AMD's new chips are long overdue, arriving a year and a half after the first wave of Ryzen 5000 chips, but they’re badly needed. Intel’s Alder Lake launch caught AMD uncharacteristically flat-footed, wresting away AMD’s performance lead. Intel’s aggressive pricing also brought superior value in every price range while exploiting AMD’s glaring lack of any sub-$250 Zen 3 chips. 

AMD's solution is pretty simple: The company is bringing back its non-X models, but with a twist. As a reminder, AMD's non-X models are the lower-cost and lower-performing versions of the 'X' models (like the Ryzen 5 5600X), but they have historically offered nearly the same performance as their counterparts, particularly after overclocking. AMD is also putting a new spin on the practice: Some of these new models, like the Ryzen 5 5500, are repurposed APUs with a disabled integrated GPU.

The four new non-X models are designed to respond to Intel's commanding lead on the low end, and AMD has also slashed pricing on its existing Ryzen 5000 models. AMD is even enabling support for Ryzen 5000 chips on older 300-series motherboards, opening up a value option that’s a good fit for the new low-end Ryzen lineup. Here's how AMD's new mainstream contenders stack up. 

AMD Ryzen 5 5600 and 5500 Specifications and Pricing

The Ryzen 5 5600 and 5500 have drastically different designs, but there are a few commonalities. Like their more full-fledged counterparts, both chips come with a bundled Wraith Stealth cooler that is sufficient for stock operation. They also support up to DDR4-3200 memory. In addition, these chips are fully unlocked for overclocking the CPU cores, memory, and fabric. In contrast, Intel's non-K models only support memory overclocking, and the company's nonsensical decision to keep certain voltages locked restricts DDR4 overclocking headroom.

The $199 Ryzen 5 5600 is a new Vermeer model, meaning it comes with the standard chiplet-based Zen 3 architecture. This chip is the long-awaited and badly-needed ‘non-X’ version of the Ryzen 5 5600X, so it shares the same feature set, albeit with reduced clocks. The 5600 has a 3.5 GHz base and 4.4 GHz boost, so you're only losing 200 MHz for the base and boost clock rate compared to the 5600X. This chip will grapple with Intel’s popular Core i5-12400.

For its lowest-end Zen 3 chip, the $159 Ryzen 5 5500 finds AMD employing a new tactic of repurposing its monolithic (single-chip) Cezanne silicon that it typically uses for APUs, but the company has disabled the chip’s integrated Radeon Vega graphics engine. This six-core 12-thread chip slots in to compete with Intel's graphics-less $167 Core i5-12400F.

Aside from the disabled iGPU, the 5500 has the same design as the Ryzen 5 5600G, including support for PCIe 3.0 instead of PCIe 4.0. As a result, this chip will make a great pairing for older, lower-end AM4 motherboards (you definitely don’t want to pay for functionality you don’t need by pairing it with a PCIe 4.0-supporting motherboard).

The 5500 is very similar to the 5600G that has the same architecture, but an active iGPU — there’s only a 200 MHz difference in CPU base/boost clock rates between the two chips. Like its counterpart, the Ryzen 5 5500 also comes with 16MB of cache, half that of the chiplet-based Ryzen 5 5600 that has the same number of cores and threads. This will result in reduced performance in several workloads.

The Ryzen 5 5600 and Ryzen 5 5500 would make great pairings for inexpensive 300- and 400-series motherboards. These chips are already supported on 400-series boards, and as of AGESA version 1207, most 300-series motherboards will support Zen 3 Ryzen 5000 processors after a BIOS update (make sure the BIOS has AGESA 1207 or newer). AMD says that Ryzen 5000 support will vary by vendor, as will the timeline for new BIOS revisions. However, we should see the updates arrive in the April-May timeframe. Notably, these BIOS revisions will also include the fix for AMD’s fTPM stuttering issues. Let's move on to the CPU benchmarks. 

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: Zen 4 Ryzen 7000 All We Know
• MORE: How to Overclock a CPU

AMD Ryzen 5 5600 and 5500 Power Consumption and Efficiency

The Intel Alder Lake chips still suck more power than AMD's Zen 3-powered Ryzen 5000 series chips, but pairing the Intel 7 process with the hybrid architecture brings big improvements, particularly in threaded work.

The Zen 3-equipped Ryzen 5 5600G has long been the most efficient chip we've tested, but the Ryzen 5 5500 gives it a run for the money in the HandBrake renders-per-day-per-watt measurements. That isn't entirely surprising, as the 5500 is merely a lower-clocked version of the 5600G, and efficiency tends to improve the further you move to the left on the voltage/frequency curve. 

The Ryzen 5 5600 is also plenty efficient, but the Core i5-12400 takes a slight lead in the efficiency metrics. As you'd expect, the Core i5-12400 and Core i3-12100 draw more power during the heavily-threaded AVX y-cruncher workload. 

Here we take a slightly different look at power consumption by calculating the cumulative energy required to perform Blender and x264 and x265 HandBrake workloads, respectively. We plot this 'task energy' value in Kilojoules on the left side of the chart. 

These workloads are comprised of a fixed amount of work, so we can plot the task energy against the time required to finish the job (bottom axis), thus generating a really useful power chart. 

Bear in mind that faster compute times, and lower task energy requirements, are ideal. That means processors that fall the closest to the bottom left corner of the chart are best. 

AMD Ryzen 5 5600 and 5500 Test Setup

We tested with Windows 11 on an X570 motherboard to maintain a comparable test environment with the rest of the processors in the test pool. Of course, you wouldn't pair this chip with this class of motherboard, but even lower-end 300-series motherboards should provide enough juice for full operation. We also tested with secure boot, virtualization support, and fTPM/PTT active to reflect a properly configured Windows 11 install. We have a full breakdown of the test system configurations at the end of the article.

Our overclocks were rather straightforward — we enabled the auto-overclocking Precision Boost Overdrive (PBO) feature with 'advanced motherboard' settings and adjusted the scalar setting to 10X. You can see details of how to do this in our How to Overclock a CPU feature. Additionally, as outlined in the table below, we made sure to match our memory overclocks with a 1:1 FCLK/memory clock ratio to keep latency low, which games love. For the Ryzen 5 5500, we had to increase the SoC voltage to 1.25V to solidify our DDR4-4000 memory overclock, but that wasn't required for the Ryzen 5 5600, largely because it tapped out at a lower DDR4-3800 in 1:1 mode. We tested the Ryzen 5 5600 and Ryzen 5 5500 in two different configurations each:

• Ryzen 5 5600: Corsair H115i 280mm water cooler, PBO Disabled, DDR4-3200
• Ryzen 5 5600 PBO: Corsair H115i 280mm water cooler, PBO Enabled, Scalar 10X, DDR4-3800, FCLK 1900 MHz (1:1 Ratio)
• Ryzen 5 5500: Corsair H115i 280mm water cooler, PBO Disabled, DDR4-3200
• Ryzen 5 5500 PBO: Corsair H115i 280mm water cooler, PBO Enabled, Scalar 10X, DDR4-4000, FCLK 2000 MHz (1:1 Ratio)

• MORE: AMD vs Intel
• MORE: Zen 4 Ryzen 7000 All We Know
• MORE: How to Overclock a CPU
• MORE: How to check CPU Temperature

AMD Ryzen 5 5600 and 5500 Gaming Benchmarks — The TLDR

As usual, we're testing with an Nvidia GeForce RTX 3090 to reduce GPU-imposed bottlenecks as much as possible, and differences between test subjects will shrink with lesser cards or higher resolutions. You would never see these low-end chips paired with an RTX 3090, but this allows us to highlight unrestrained chip performance. Because most of the titles below show little meaningful differentiation at higher resolutions, we only tested four of the seven titles at 1440p. We typically include Microsoft Flight Simulator in our test suite, but we encountered inexplicable issues with loading the game on our AMD test platforms and didn't have time to rectify the issue prior to the NDA lift (this issue is with the game itself, and is not indicative of an issue with AMD's chips).

At stock settings, the Core i5-12400 is 2% faster than the Ryzen 5 5600 in our overall measurement of 1080p gaming performance, which is a slim victory. After overclocking, the 12400 is 3.9% faster than the Ryzen 5 5600 and ties the 5600X for the overall lead.

The Ryzen 5 5600 is basically a 5600X-killer — it doesn't make sense to spend the extra cash for such slim gains. Both the 12400 and the 5600 retail for $199, and that's a tough position for the Ryzen 5 5600 given its aging AM4 platform with less sophisticated connectivity. Additionally, you could choose the graphics-less 12400F that offers the same performance as the full-featured model for ~$175, helping reduce the extra costs associated with Alder Lake motherboards.

The Ryzen 5 5500 falls within 2 percent of the Ryzen 5 5600G in the 1080p benchmarks, but it isn't as much of a closely-contested battle against the Intel chips. The less-expensive $129 Core i3-12100 is 6.2% faster in gaming at stock settings than the $159 Ryzen 5 5500, and basically ties the overclocked 5500 without any effort.

If gaming is all you're interested in, the Core i3-12100 is the better buy with a much more modern platform — the 5500 is limited to the PCIe 3.0 interface. However, the Ryzen 5 5500 will have advantages for the productivity-minded, as we'll see in the application benchmarks.

Both the new Ryzen chips offer a tremendous step forward over their Zen 2 predecessors. The Ryzen 5 5500 is roughly 10% faster than the Ryzen 5 3600X and 3600, while the Ryzen 5 5600 is ~30% faster, representing a massive generational improvement.

Naturally, moving over to 1440p brings a GPU bottleneck into the equation, so the performance deltas between the chips shrink tremendously. Flipping through the 99th percentile charts for both resolutions also shows larger deltas, but we have to view those with caution as Windows 11 seems to suffer from more framerate variability than our Windows 10 test platform.

Be aware that large performance deltas in a few of the game titles can heavily impact these types of overall measurements. It's always best to make an informed decision based on the types of titles you play frequently, so be sure to check out the individual game benchmarks below.

3DMark, VRMark, Chess Engines on AMD Ryzen 5 5600 and 5500

Synthetic benchmarks don't tend to translate well to real-world gaming, but they do show us the raw amount of compute power exposed to game engines. It's too bad most games don't fully exploit it. 

Far Cry 6 on AMD Ryzen 5 5600 and 5500

F1 2021 on AMD Ryzen 5 5600 and 5500

Hitman 3 on AMD Ryzen 5 5600 and 5500

Horizon Zero Dawn on AMD Ryzen 5 5600 and 5500

Red Dead Redemption 2 on AMD Ryzen 5 5600 and 5500

Watch Dogs Legion on AMD Ryzen 5 5600 and 5500

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: How to Overclock a CPU

AMD Ryzen 5 5600 and 5500 Application Benchmarks — The TLDR

We can boil down productivity application performance into two broad categories: single- and multi-threaded. These slides show the geometric mean of performance in several of our most important tests in each category, but be sure to look at the expanded benchmark results further below. 

The Alder Lake chips lead by healthy margins in our cumulative measurement of single-threaded performance, with the Core i5-12400 being 14.6% faster than the Ryzen 5 5600 at stock settings, and 10% faster after we tune both chips. Meanwhile, the stock Core i3-12100 is 19% faster than the Ryzen 5 5500, and overclocking the latter does little to rectify the disparity. You'll have to look to Intel's own Alder Lake family for faster single-threaded performance. 

Turning to threaded applications, the Core i5-12400 is 2.8% faster than the Ryzen 5 5600 at stock settings, and 5.5% faster after tuning. Meanwhile, the Ryzen 5 5500 is 19% faster than the Core i3-12100 in threaded work, and that increases to 23% after overclocking.

Rendering Benchmarks on AMD Ryzen 5 5600 and 5500

The Core i3-12100 is impressive in single-threaded rendering work, leading all competing Ryzen chips in both Cinebench and POV-Ray benchmarks, but its quad-core design isn't well suited for the heavily parallelized rendering workloads that you'll see in the real world. 

The Core i5-12400 is a closer match for the Ryzen 5 5600 in the threaded benchmarks, but the Intel chip often trails only by a slight amount or leads convincingly. Paired with its absolute dominance in lightly-threaded rendering, the Core i5-12400 is the better all-rounder for this type of work. 

Encoding Benchmarks on AMD Ryzen 5 5600 and 5500

Encoders can be either heavily threaded or almost exclusively single-threaded — there certainly doesn't seem to be a middle ground.

The HandBrake transcoder is heavily threaded and uses AVX instructions to boost performance, with the x265 version featuring a heavier distribution of SIMD instructions than the x264 version. The Core i5-12400 and the Ryzen 5 5600 are closely matched in x265, which naturally plays to Intel's strength in AVX work, but the 5600 pulls ahead slightly in x264. 

Turning to the single-threaded LAME benchmark finds the Intel processors leveraging their generally higher IPC and clock rates to take the lead, while the Ryzen 5 5600 takes the lead in the FLAC encoder.

Overall, you'll find the 12400 or the 5600 jockeying for the lead in this selection of workloads, but the Ryzen 5 5500 easily beats the Core i3-12100 due to its higher core and thread count. 

Adobe Premiere Pro, Photoshop, and Lightroom on AMD Ryzen 5 5600 and 5500

We've integrated the UL Benchmarks Procyon tests into our suite to replace the aging PCMark 10. This new benchmark runs complex Adobe Premiere Pro, Photoshop, and Lightroom workflows with the actual software, making for a great real-world test suite.

Web Browsing, Office and Productivity on AMD Ryzen 5 5600 and 5500

The ubiquitous web browser is one of the most frequently used applications. These tests tend to be lightly threaded, so a snappy response time is critical. We're accustomed to the Alder Lake chips running the table over competing Ryzen chips in the web browser tests, but the Ryzen 5 5600 is competitive. 

Compilation, Compression, AVX Benchmarks on AMD Ryzen 5 5600 and 5500

This section includes a diverse set of workloads, including exceedingly branchy code in the LLVM compilation workload and the massively parallel molecular dynamics simulation code in NAMD. The Core i3-12100 isn't very potent in these heavily-threaded applications, but the Core i5-12400 leads many of the benchmarks. Notably, the Ryzen 5 5600 does well in the NAMD workload due to its healthy slab of L2 cache. AMD benefits in the SHA3, AES, and HASH benchmarks from its cryptographic optimizations. That said, most of these types of workloads in this section aren't well-suited for this class of chip, but we include them as a reference.

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel

The arrival of AMD's Ryzen 5000 found the company with a clear performance lead over rival Intel chips, so the company prioritized high-priced high-margin chips and largely abandoned the budget market. The arrival of Alder Lake has changed the paradigm entirely, spurring AMD to finally bring Zen 3 to the budget arena. 

AMD's non-X $199 Ryzen 5 5600 and $159 Ryzen 5 5500 lower the bar for entry to the Zen 3 architecture and provide access to the lowest-priced AM4 motherboards via retroactive support on older boards, but these chips don't retake the overall performance lead in their price brackets. Instead, they slot in as a solid upgrade for existing AMD system owners or as a value alternative to Intel's Alder Lake for new system builds.

Below, we have the geometric mean of our gaming test suite at 1080p and 1440p and a cumulative measure of performance in single- and multi-threaded applications. We conducted our gaming tests with an RTX 3090, so performance deltas will shrink with lesser cards and higher resolution and fidelity settings.

At stock settings, the Core i5-12400 is 2% faster than the Ryzen 5 5600 in our cumulative measurement of 1080p gaming performance and 3.9% faster after overclocking. In our multi-threaded application benchmarks, the Core i5-12400 is 2.8% faster than the Ryzen 5 5600 at stock settings and 5.5% faster after tuning. It's also 15.6% faster in single-threaded applications. 

Both chips retail for around $199, leaving the Ryzen 5 5600 in a tricky position. If pricing stays in line with the MSRP, the 5600's primary advantage will be its access to cheaper motherboards that lower the total platform cost. However, the AM4 platform is long in the tooth, and you'll sacrifice the PCIe 4.0 interface if you go with older boards.

Additionally, while the 5600 lacks integrated graphics, the 12400 has the built-in UHD graphics 770 engine to provide at least a basic display out. You could also go with the graphics-less 12400F for ~$175, which helps reduce the impact of the extra costs associated with motherboards for Alder Lake. The Ryzen 5 5600 is also basically a 5600X-killer — it doesn't make sense to spend the extra cash for such slim gains.

The $129 Core i3-12100 is 6.2% faster in gaming at stock settings than the $159 Ryzen 5 5500, and it ties after overclocking. The Ryzen 5 5500 is 19% faster than the Core i3-12100 in threaded applications, which increases to 23% after overclocking. Meanwhile, the stock Core i3-12100 is 19% faster than the Ryzen 5 5500 in single-threaded apps. 

The Core i3-12100 is better for gaming-only rigs, but the Ryzen 5 5500 is undoubtedly a better all-rounder. However, the 5500 only supports the PCIe 3.0 interface, so it's best suited for older, lower-end motherboards. 

Critically, AMD has added support for the Ryzen 5000 chips to older 300-series motherboards. That makes perfect sense for the lower-end processors because it can keep the company's existing customers from jumping to Alder Lake instead of waiting for the Zen 4 Ryzen 7000 chips to arrive. 

However, AM4's longevity is both a blessing and a curse — Intel's Alder Lake has a more sophisticated platform that supports the PCIe 5.0 interface and DDR5 memory. Yes, PCIe 5.0 devices aren't available yet, and DDR5 memory remains prohibitively expensive for this class of chip. However, it is nice to have those options in the years to come — PCIe 5.0 SSDs are in the works, and DDR5 pricing will eventually normalize. Additionally, AMD is moving on to the AM5 socket with Zen 4, making support for the 300-series boards feel like too little, too late. 

AMD's new tactic of repurposing its monolithic Zen 3 and Zen 2 APU designs by disabling the graphics units doesn't seem like the best use of die area, but 7nm is mature and yields exceedingly well. This marks a new shift in the company's strategy to using a monolithic die for the low end, a necessity for these lower price points because chiplet-based designs are more costly to produce in terms of packaging and logistics.

AMD has probably built up enough chips with defective iGPUs over the years to meet the initial demand, but we doubt it would purposely defeature chips to maintain the supply. As such, the supply of the defeatured APUs could dwindle rather quickly, much as we've seen with the company's other low-end models (like the Ryzen 3 3300X and 3100). 

The Ryzen 5 5500 is a great chip for upgraders with existing Ryzen systems, but it only makes sense for new builds if you need more threaded heft and don't mind using an older platform. The Core i3-12100 is a better choice for gaming-focused rigs due to its higher performance and friendlier price tag. 

The Ryzen 5 5600 is a great drop-in chip for upgraders that already have systems built around older Zen processors, but it's hard to recommend for a new build unless you're willing to sacrifice modern amenities to maximize the bang for your buck. Overall, the Core i5-12400F is a better choice for both gaming and productivity-focused builds alike. 

• MORE: Zen 4 Ryzen 7000 All We Know
• MORE: How to Overclock a CPU
• MORE: How to check CPU Temperature
• MORE: All CPUs Content

Kopite7kimi, a well-known leaker on Twitter, has published what may be the first specifications of Nvidia's upcoming GeForce RTX 40-series graphics cards family. Since we are several months away from Nvidia's launch of its codenamed Ada Lovelace architecture, it's possible there will be changes, but given the leaker's track record, the specs are at least worth considering. 

Nvidia's initial GeForce RTX 40-series family will consist of three graphics cards: the GeForce RTX 4090, based on the AD102-300 graphics processor paired with 24GB of GDDR6X memory; the GeForce RTX 4080, powered by the AD103-300 GPU mated with 16GB of GDDR6/GDDR6X memory; and the GeForce RTX 4070 featuring the AD104-275 chip equipped with 10GB of GDDR6. 

The number of CUDA cores featured by Nvidia's Ada Lovelace GPUs show that the new graphics boards will likely have a considerable performance increase when compared to Nvidia's existing GeForce RTX 30-series, which are among best graphics cards around. 

Rumored Nvidia GeForce RTX 40-Series Specifications

All these initial GeForce RTX 40-series graphics boards will allegedly be quite power hungry with the top-of-the-range product featuring a thermal board power (TBP) of 450W and other rated for 420W and 300W, respectively. When it comes to power consumption, the information corroborates with earlier rumors indicating a major increase of TBP for Ada Lovelace family. 

While power consumption of Nvidia's next-generation GeForce RTX 40-series products will increase compared to current-generation graphics cards, not everything is clear about pricing of the new boards. Kopite7kimi tells us not to expect GeForce RTX 4000 boards to feature lower MSRPs. Price is a particularly murky topic, keeping in mind higher production costs on TSMC's N5 process technology (or rather customized 4N) when compared to Samsung's 8LPP. 

It is noteworthy that this time around Nvidia will use three different graphics processors for three initial cards. This might be a result of decent initial yields of Nvidia's Ada Lovelace GPUs at TSMC, or perhaps due to some changes in the company's internal chip design schedules. Anyhow, the gaps between the number of CUDA cores featured by the AD102, AD103 and AD104 GPUs might indicate that Nvidia will be very flexible with configurations for refresh (Ti-branded or Super-branded products). 

Of course, Nvidia typically does not comment on specifications of unreleased products, so take the information with a grain of salt.

Intel's four-core eight-thread Core i3-12100 comes with an incredibly competitive $129 price tag that earns a spot on our list of best CPUs for gaming and Best Cheap CPUs as Intel finally addresses what has become the most ignored part of the PC market — the sub-$200 segment. That's not to mention that the chip also comes as a $104 F-series Core i3-12100F that Intel ships with deactivated integrated graphics for $25 less than the full-featured model. In fact, with no clear current-gen competitor from AMD and stellar performance for its price point, the Core i3-12100 easily leads our CPU benchmark hierarchy in the $105 to $130 bracket.

Intel refreshed its Comet Lake Core i3 lineup when it released its 11th-Gen Rocket Lake chips in 2020, but those models didn't come with a new architecture or any meaningful performance improvements. Rather, they came as refreshed 10th-Gen models with a paltry 100 MHz clock speed increase. Not that it mattered — given the realities of the chip shortages, we rarely saw those chips at retail anyway.

Speaking of chips that don't really exist, AMD's last budget model came as the incredibly impressive Ryzen 3 3300X that landed back in 2020. The quad-core 3300X brought an unheard-of level of performance for a $120 chip, promising new levels of gaming performance for budget builds. Unfortunately, that didn't come to fruition as the chip was a ghost and never appeared in any meaningful volume at retail.

Things haven't improved in the interim, either. AMD abandoned the sub-$200 market when it launched its Ryzen 5000 processors, leaving its older 3000-series processors to hold the line. However, as you'll see in our benchmarks below, they aren't relevant. AMD's lowest point of entry into its Zen 3-powered Ryzen 5000 series comes in the form of the $259 Ryzen 5 5600G. At twice the price of the 12100, it's a non-factor for lower-end gaming rigs unless you plan to use integrated graphics.  

• Ryzen 7 5800X3D vs Core i7-12700K and Core i9-12900K Face-Off: The Rise of 3D V-Cache
• Intel Core i9-12900K vs Ryzen 9 5900X and 5950X: Alder Lake and Ryzen 5000 Face Off
• Intel Core i5-12600K vs AMD Ryzen 5 5600X and 5800X Face Off: Ryzen Has Fallen
• Intel Core i7-12700K vs AMD Ryzen 9 5900X and 5800X Face Off: Intel Rising
• Intel Core i5-12400 vs AMD Ryzen 5 5600X Face-Off: The Gaming Value Showdown

That leaves Intel unchecked in the budget segment, adding to the company's newfound dominance with the Alder Lake chips that even outperform more expensive Ryzen 5000 chips. Intel's advantages also extend to the motherboard ecosystem too, with B660 and H610 motherboards providing a great pairing for the Core i3-12100. So even though these boards do cost more than we're accustomed to for the lowest-end models, they provide plenty of connectivity for budget systems.

Alder Lake's performance advantages come even without its support for DDR5 memory and PCIe 5.0 interfaces (both of which Intel brought to market first). As such, you can use standard DDR4 memory and PCIe 4.0 devices and still have superior performance and connectivity options over AMD's aging AM4 platform. There are also plenty of B- and H-series boards that leverage less-expensive DDR4 memory, which is a saving grace given the ongoing DDR5 shortages.

Alder Lake also brings another innovation — the hybrid x86 design. The higher-end Alder Lake chips have big and fast Performance cores (P-cores) for latency-sensitive work paired with clusters of small and powerful Efficiency cores (E-cores) that chew through background processes. The Golden Cove architecture powers the 'big' P-cores, while the 'little' E-cores come with the Gracemont architecture.

However, the Core i3-12100 doesn't have a hybrid architecture, instead coming with a more traditional design with only four Golden Cove P-Cores active. That means this four-core eight-thread processor doesn't need Intel's new Windows 11-exclusive Thread Director technology to place workloads on the correct cores. As a result, unlike Intel's hybrid models, the 12100 is just as potent in Windows 10 as it is in Windows 11.

As you'll see in our benchmarks below, the Core i3-12100 doesn't have a similarly-priced competitor from AMD. However, despite a total lack of competition, it still brings impressive generational performance gains to the table. In fact, in 1080p gaming, the $129 Core i3-12100 delivers 88% of the $299 Core i5-12400's performance, but for 56% less cash. The Core i3-12100 also trails the previous-gen $262 Core i5-11600K by a mere 3% in gaming, but at half the price. 

Overall, the quad-core i3-12100's potent combination of price, performance, and improved stock cooler dominates the $100 to $130 price range while punching up against more expensive competitors. 

Intel Alder Lake-S Core i3-12100 Specifications and Pricing

We have deep-dive coverage of the Alder Lake design and microarchitectures here, along with a broader overview in our Alder Lake all we know article. Additionally, Intel now assigns a Processor Base Power (PBP) spec instead of using the 'TDP' (Thermal Design Point) nomenclature. The company also added a secondary Maximum Turbo Power (MTP) value to represent the highest power level during boost activity. You can read more about that here.

Intel fabs Alder Lake on the 'Intel 7' process. We previously knew this 'Intel 7' manufacturing tech as 10nm Enhanced SuperFin, but Intel recently renamed its process nodes to match industry nomenclature. Technically, 'Intel 7' is the second generation of Intel's 10nm process, but it's a first for desktop PCs.

All Alder Lake chips support DDR4-3200 or up to DDR5-4800 memory, but caveats apply. PCIe support will vary by motherboard, but Alder Lake chips expose up to 16 lanes of PCIe 5.0 (technically for storage and graphics only, no networking devices) and an additional four lanes of PCIe 4.0 from the chip for M.2 storage. Intel's Alder Lake drops into Socket 1700 motherboards from the 600-series, including Z690, H670, B660, and H610.

The Core i3-12100 comes with a 60W PBP (base) and 89W MTP (peak) power rating. The chip clocks in with a 3.3 GHz base and boosts up to 4.3 GHz. It also comes with 12 MB of L3 cache.

The Core i3-12100 is a locked chip, so it isn't overclockable. However, Intel supports memory overclocking on Z690, H670 and B660 motherboards (Z690 doesn't make sense for this class of chip, though). Manipulating the power limits serves as a quasi-overclock that can eke out some additional performance in some gaming and threaded work, but you don't get much of a benefit with chips this far down on the low end.

Intel has revamped its stock air coolers with Alder Lake. These coolers are designed to address two major deficiencies with Intel’s stock coolers: Thermal dissipation limitations and aesthetics. AMD’s stock coolers have long beat Intel in both of these departments, so this is a sorely-needed upgrade. The Core i3-1100 ships with the Laminar RM1 cooler that comes without RGB lighting but has a decorative blue plastic ring lining the fin stack. Intel rates this cooler for ‘quiet performance’ at 3.9 BA. 

We tested with both the stock heatsink and a Corsair H115i watercooler to gauge the strength of the air cooler. We didn't measure any meaningful difference between the two, so as long as you're not experiencing severe chip bowing issues, you can use the stock cooler without worry.

The standard Core i3-12100 comes with the UHD Graphics 730 engine with 24 EUs. The engine runs at 300/1400 MHz base/boost frequencies. If you're looking to save some coin, the graphics-less Core i3-12100 comes with a $25 price reduction and has the same specs as the 12100, which is incredibly attractive if you plan on using a discrete graphics card. The only difference between the standard 12100 and the 12100F is that the latter has a 58W PBP rating, so performance is identical with both models. Notably, going with the 12100F means you will lose Quick Sync capabilities and the iGPU fallback that you can use for troubleshooting in the event of an issue with a discrete GPU.

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: Zen 4 Ryzen 7000

Intel Alder Lake Core i3-12100 Power Consumption and Efficiency

The Intel Alder Lake chips still suck more power than AMD's Zen 3-powered Ryzen 5000 series chips, but pairing the Intel 7 process with the hybrid architecture brings big improvements, particularly in threaded work.

As we can see, the 12100 jockeys with the quad-core Comet Lake Core i3-10100, with the latter often consuming less power. But that comes at the cost of performance. As you can see in our renders-per-day measurements, the Core i3-12100 is more efficient, which comes as a byproduct of its higher performance within a similar power envelope.

The Zen 3-equipped Ryzen 5 5600G takes the crown as the most efficient chip in the test pool and often finds itself in the mix with the Core i3 models in the average power measurements. This six-core 12-thread chip also serves up quite a bit more performance than the i3's, so it takes a big lead in our renders-per-watt-per-day metrics. 

However, the Core i3-12100 doesn't have a modern quad-core AMD competitor, and it takes the win against MD's only quad-core entrant, the Zen 2-powered Ryzen 3 3300X. 

Here we take a slightly different look at power consumption by calculating the cumulative energy required to perform Blender and x264 and x265 HandBrake workloads, respectively. We plot this 'task energy' value in Kilojoules on the left side of the chart. 

These workloads are comprised of a fixed amount of work, so we can plot the task energy against the time required to finish the job (bottom axis), thus generating a really useful power chart. 

Bear in mind that faster compute times, and lower task energy requirements, are ideal. That means processors that fall the closest to the bottom left corner of the chart are best. 

As you can see, Intel's chips have descended from the undesirable upper right of the chart down closer to the lower left hand, indicating improved efficiency. The gap between the Core i3-12100 and the Core i3-10100 illustrates just how much the Golden Cove architecture paired with the 'Intel 7' process has improved the company's standings in our efficiency measurements. 

Test Setup

We tested with Windows 11 and DDR4 memory on a Z690 motherboard to maintain a comparable test environment with the rest of the processors in the test pool. Of course, you wouldn't pair this chip with this class of motherboard, but the 600-series B- and H-series motherboards also support overclocking memory and removing power limits. Given the 12100's tame power consumption, even lower-end motherboards will provide the chip with enough juice for full operation. We used DDR4 memory for testing, as DDR5 pricing removes it from consideration for this class of chip.

The Core i3-12100 is a locked chip, meaning you can't overclock the CPU cores, but you can overclock the memory on Z- and B-series motherboards. We also tested with secure boot, virtualization support, and fTPM/PTT active to reflect a properly configured Windows 11 install. We have a full breakdown of the test system configurations at the end of the article. We tested the Core i3-12100 in two different configurations:

• Core i3-12100 DDR4-3800: Corsair H115i 280mm water cooler, power limits removed, memory overclocked to DDR4-3800 in Gear 1 mode (Gear 2 results in performance regressions)
• Core i3-12100: Stock cooler, Intel recommended stock power limits (60/89W), Stock DDR4-3200 in Gear 1

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: How to Check CPU Temperature 

Intel Core i3-12100 Gaming Benchmarks — The TLDR

As usual, we're testing with an Nvidia GeForce RTX 3090 to reduce GPU-imposed bottlenecks as much as possible, and differences between test subjects will shrink with lesser cards or higher resolutions. You would never see the Core i3-12100 paired with an RTX 3090, but this allows us to highlight unrestrained chip performance. Because most of the titles below show little meaningful differentiation at higher resolutions, we only tested four of the seven titles at 1440p.

We paired the Core i3-12100 with affordable DDR4 memory for our testing. We removed the 12100's power limits and overclocked the memory for the 'Core i3-12100 DDR4-3600' entry, but only registered a 2.2% improvement. That means you won't benefit much from investing in a more expensive memory kit.

The 12100 doesn't need too much help, though: The chip was a whopping 29.5% faster than the Core i3-10100 in our cumulative gaming measurement, representing a massive leap forward for budget 1080p gaming. 

The 11600K, last-gen's fastest Core i5, was only 3.5% faster than the stock Core i3-12100, but overclocking the 12100's memory narrowed that to 1.4%. That's an impressive gen-on-gen improvement given the 11600K is twice the price of the 12100. Naturally, overclocking the 11600k would give it the lead, but that also requires a much more expensive cooler and other accommodations. 

The 12100 is even more impressive against AMD's lower-end models. Moving on to the only comparably-priced AMD chip, the mythical quad-core Ryzen 3 3300X, finds the 12100 beating it by 19.2% and 18.8% at stock and overclocked settings, respectively. The 12100 is also 19% and 9% faster than the six-core $199 Ryzen 5 3600 and $240 3600X, respectively, showing that it has the chops to take on AMD's entire sub-$250 roster.

That means we have to move up into the $260 range to find an AMD chip that can compete with the 12100, but there isn't a great AMD comparable at that price point. AMD's Ryzen 5 5600G APU isn't designed as a direct competitor for the 12100 — it's designed for gaming on its integrated graphics, and there it will easily outmaneuver the 12100. However, when paired with a discrete GPU, the 12100 is 6% and 1% faster than the 5600G at stock and overclocked settings, respectively. So at twice the price, it's clear that the Ryzen 5 5600G isn't a suitable competitor for the 12100 if you plan on using a discrete GPU.

The Core i5-12400 is the next step up the Intel ladder. At $199, the 12400 is 13% and 16% faster than the 12100 at stock and unlocked power settings, respectively. Put another way, the 12100 delivers 88% of the 12400's gaming performance, but for 56% less cash. The Core i5-12400 delivers much more performance in threaded application benchmarks than the 12100, though, making it a better all-rounder. 

Naturally, moving over to 1440p brings a GPU bottleneck into the equation, so the performance deltas between the chips shrink tremendously. Flipping through the 99th percentile charts for both resolutions also shows larger deltas, but we have to view those with caution as Windows 11 seems to suffer from more framerate variability than our Windows 10 test platform.

The Core i3-12100 easily beats the Ryzen comparables, but be aware that large performance deltas in a few of the game titles can heavily impact these types of overall measurements. It's always best to make an informed decision based on the types of titles you play frequently, so be sure to check out the individual game benchmarks below.

In either case, the Core i3-12100 is now the budget gaming champion, offering a superior level of performance at its price point with no clear competitors.

3DMark, VRMark, Chess Engines on Intel Core i3-12100

Synthetic benchmarks don't tend to translate well to real-world gaming, but they do show us the raw amount of compute power exposed to game engines. It's too bad most games don't fully exploit it. 

Far Cry 6 on Core i3-12100

F1 2021 on Intel Core i3-12100

Hitman 3 on Core i3-12100

Horizon Zero Dawn on Core i3-12100

Microsoft Flight Simulator 2021 on Core i3-12100

Red Dead Redemption 2 on Core i3-12100

Watch Dogs Legion on Core i3-12100

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: How to Check CPU Temperature

Core i3-12100 Application Benchmarks — The TLDR

We can boil down productivity application performance into two broad categories: single- and multi-threaded. These slides show the geometric mean of performance in several of our most important tests in each category, but be sure to look at the expanded benchmark results further below.

The Core i3-12100 is surprisingly agile in our cumulative measure of single-threaded performance. The 12100 is 27% and 5% faster than the Core i5-10100 and i5-11600K,  respectively, but the 12400 leads by roughly 2%.

Compared to Ryzen, the 12100 dominates in single-threaded applications, with its lead stretching between 25% over the Ryzen 5 3600 to 11% over the Ryzen 5 5600G. 

Overall, the Core i3-12100 offers great performance in single-threaded workloads for its price point, but if you're looking for the closest thing to a "catch," you'll find it in threaded application workloads. 

In multi-threaded work, the Core i3-12100 continues to assert its dominance over comparably-priced chips with an 18% lead over the Ryzen 3 3300X and a 30% lead over the Core i3-10100. However, the 124100 isn't as impressive in multi-threaded work against the six-core chips as we saw in our gaming benchmarks. The Ryzen 5 3600 and 3600X lead by ~11%, while the 5600G leads by 19%. 

Those six-core chips obviously lead in threaded productivity applications, they do carry much higher price tags after all, but they pale in comparison to the Core i5-12400 as it takes a 32% lead over the 12100. Unfortunately, unlocking the power limits and tuning the memory didn't yield any performance increases in threaded work for either the 12100 or 12400.

Overall the Core i3-12100 offers a solid blend of performance in both single- and multi-threaded apps given its price point, but its single-threaded performance stands out as exceptional. You'll have to look to Intel's own Alder Lake family for faster single-threaded performance. The 12100 also dispatches the comparably-priced Ryzen 3 3300X and Core i3-10100 easily in threaded work. 

Rendering Benchmarks on Core i3-12100

The Core i3-12100 is impressive in single-threaded rendering work, leading all competing Ryzen chips in both Cinebench and POV-Ray benchmarks, including those that cost more than twice the price.

The 12100 also easily beats AMD's price-comparable Ryzen 3 3300X throughout the full gamut of threaded rendering benchmarks and shows massive gains over the Core i3-10100. The six-core Ryzen models take the lead in the threaded workloads over the quad-core Core i3-12100, but that isn't a fair comparison because they are in a much higher price class. 

Encoding Benchmarks on Core i3-12100

Web Browsing on Intel Core i3-12100

The ubiquitous web browser is one of the most frequently used applications. These tests tend to be lightly threaded, so a snappy response time is critical. The Core i3-12100 beats the competing Ryzen processors in its price bracket easily.

Adobe Premiere Pro, Photoshop, and Lightroom on Core i3-12100

We've integrated the UL Benchmarks Procyon tests into our suite to replace the aging PCMark 10. This new benchmark runs complex Adobe Premiere Pro, Photoshop, and Lightroom workflows with the actual software, making for a great real-world test suite. 

The Core i3-12100 is incredibly impressive as it takes the lead over all the Ryzen chips in Adobe Lightroom and Photoshop and trails only the vastly more expensive $299 Ryzen 5 5600X in Adobe Premiere Pro. 

Office and Productivity on Core i3-12100

The Core i3-12100 provides snappy application load times, but the Ryzen 5 5600G and Ryzen 3 3300X are faster. Conversely, the 12100 beats the entire Ryzen roster in the Microsoft Office suite. 

Compilation, Compression, AVX Benchmarks on Core i3-12100

The Core i3-12100 easily beats the comparably-priced Ryzen 3 3300X in nearly all of these workloads, including exceedingly branchy code in the LLVM compilation workload and the massively parallel molecular dynamics simulation code in NAMD. That said, most of these types of workloads aren't well-suited for this class of chip, but we include them as a reference.

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: Zen 4 Ryzen 7000

Budget Gaming Dominance

Like the rest of the Alder Lake family, the $104 to $130 Core i3-12100 comes to market with disruptive pricing as Intel continues to attempt to claw back market share from AMD at any cost.

Frankly, given that Intel has virtually no competition from AMD in the sub-$200 market, it's surprising that Intel has delivered this much performance at such attractive pricing. Below, we have the geometric mean of our gaming test suite at 1080p and 1440p and a cumulative measure of performance in single- and multi-threaded applications. We conducted our gaming tests with an RTX 3090, so performance deltas will shrink with lesser cards and higher resolution and fidelity settings. 

The Core i3-12100 eschews the hybrid Alder Lake design, instead going with 'only' four P-cores. That doesn't hold the chip back in gaming, though, and the Core i3-12100 now reigns as the fastest budget gaming CPU on the market.  

The Core i3-12100 represents a massive leap forward for Intel in budget gaming performance, beating the Core i3-10100 by a whopping 29.5% at 1080p. The Core i5-11600K, last-gen's fastest Core i5, was surprisingly only 3.5% faster than the stock Core i3-12100, but at twice the price. The 12100 is also equally impressive compared to Intel's higher-end Alder Lake models, delivering 88% of the 12400's gaming performance but for 56% less cash.

AMD's chips can't compete, but that's because the company has completely abandoned the sub-$250 market. The Core i3-12100 easily beat AMD's previous-gen Ryzen 5 3600 and 3600X, not to mention the venerable 3300X, by margins ranging from 9% to 19%, respectively, showing that the 12100 has the chops to take on AMD's entire sub-$250 roster in gaming.

The Core i3-12100 is plenty impressive in lightly-threaded apps, too. In fact, the $800 Ryzen 9 5950X is the only Ryzen chip that can match the Core i3-12100's single-threaded performance in our CPU benchmarks hierarchy. You'll have to look to other Alder Lake chips to find faster performance in single-threaded work.

The Core i3-12100 is impressive in threaded productivity workloads for its price point, easily beating the price-comparable Ryzen 3 3300X in nearly every benchmark and establishing a commanding 30% lead over the Core i3-10100. Of course, AMD's more expensive six-core chips provide more performance in threaded work, but they should given their higher price tags.

You can also pick up the graphics-less $104 Core i3-12100F for ~$25 less than the full-featured model, but it provides the same level of performance. In addition, both the Core i3-12100 and 12100F also come with a capable Laminar RM1 cooler that delivers the full performance of the chip, making a sweet deal even sweeter for budget builders.

You should pair the Core i3-12100 with a B- or H-series motherboard, though the latter doesn't allow memory overclocking. That said, memory overclocking only imparted a 2.2% gain in 1080p gaming performance and no gain in most applications, so it doesn't make too much sense — especially for budget builds.

The Core i3-12100 also has a much more modern platform than AMD's AM4 ecosystem. Leading-edge DDR5 and PCIe 5.0 interfaces add too much cost for this class of chip, so look for DDR4 B660, H670 and H610 motherboards for the best value. You won't need DDR5 memory to unlock the best gaming performance, and that's a good thing because DDR5's high pricing doesn't make sense for sub-$250 chips.

Alder Lake has delivered a decisive blow to AMD's entire Ryzen 5000 family, and it doesn't look like we'll have a chance to see competitive new budget offerings until the Zen 4 'Raphael' Ryzen 7000 chips arrive later this year. If you're looking for a bit more performance in threaded workloads than you'll get with the 12100, the Core i5-12400 remains the undisputed value champ for an all-rounder chip, but you'll have to shell out an additional $65.

Overall, the Core i3-12100 is a balanced chip that offers exceptional performance in gaming and lightly-threaded work in tandem with leading performance for its price point in multi-threaded workloads. If you're looking for an unprecedented amount of gaming and application performance from a $105 to $130 chip, the Core i3-12100 is the hands-down winner and takes a spot on our list of Best CPUs for gaming. 

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: Zen 4 Ryzen 7000

The eight-core 16-thread $299 AMD Ryzen 7 5700X comes to market as a slightly modified yet lower-cost version of its predecessor, the $335 Ryzen 7 5800X, but it offers nearly the same gaming and application performance after a bit of no-hassle tuning. The 5700X debuts as part of AMD's newest line of seven Ryzen 5000 models that are designed to shore up the company's rankings in CPU benchmarks and retake its position on the Best CPUs for gaming list. That's a critical need after Intel's Alder Lake upset the Ryzen lineup with a better blend of both pricing and performance. 

The Ryzen 7 5700X leverages the same Zen 3 architecture and 7nm process as its counterparts and drops into the existing ecosystem of AM4 motherboards. Its predecessor, the 5800X, has always been an oddly-positioned chip, with its price point making it the lone Ryzen 5000 processor that didn't make much sense for just about anyone due to competing products from both Intel and AMD. In fact, the 5800X's positioning was so poor, and the Ryzen 7 5700X's absence so conspicuous, that we asked where the 5700X was right in the title box of our original review (seen below). 

It took AMD eighteen months, but it has now finally released the Ryzen 7 5700X. It's certainly late, though. As we explained in our 5800X review back in 2020, AMD really needed the 'missing' Ryzen 7 5700X to plug the big pricing gap in its product stack and make it easier for its customers to jump from Ryzen 5 to Ryzen 7 instead of buying an Intel processor. 

However, that was back when AMD competed with Intel's 10th-Gen processors. The game has changed entirely since then — Intel's disruptive 12th-Gen x86 hybrid Alder Lake chips are now well established as the overall performance and value leader at every price point, and the company's 13th-Gen Raptor Lake chips are purportedly on track for release this year. AMD also has its 5nm Ryzen 7000 ‘Raphael’ Zen 4 chips slated to arrive at the end of the year, but they'll arrive with the new AM5 platform. Meanwhile, the Ryzen 7 5700X drops into the long-lived Socket AM4 platforms that have shepherded the Ryzen chips from their infancy with the Ryzen 7 1800X in 2017 to the current day, but that cuts off any future upgrade path for adopters.

Naturally, AMD has a different pricing strategy today than we would have seen back in 2020: The 65W Ryzen 7 5700X is $150 less than the launch price of its full-fledged 105W sibling, the Ryzen 7 5800X. That isn't relevant to today's pricing situation, though — the 5700X is only $35 less than the 5800X's average pricing at retail. That isn't much of a discount. 

Given the similarities we'll see throughout our benchmark suite, the Ryzen 7 5700X is really just a price cut for the Ryzen 7 5800X, but it comes disguised as a new product.

The Ryzen 7 5700X is a solid upgrade choice if you already have a system built around a Ryzen 1000- or 2000-series processor and need more threaded horsepower. However, if you're a Ryzen upgrader that's only interested in gaming, you could save some cash with the Ryzen 5 5600, which offers comparable gaming performance at a much friendlier $199 price point. For new builds, you should look to Intel chips, like the Core i5-12400 or Core i5-12600K, and their more modern accommodations.

Let's take a quick look at the specs, then get right to our full gaming and application test results. 

AMD Ryzen 7 5700X Specifications and Pricing

The Ryzen 7 5700X slots into the Ryzen stack with eight cores and 16 threads. At $299, the 5700X is the highest-priced 65W part from AMD, filling in the gap between the powerful 105W Ryzen 7 5800X that retails for $335 and the $225 Ryzen 5 5600X that also comes with a 65W TDP. 

The Ryzen 7 5700X comes with a 3.4 GHz base, a 4.6 GHz boost clock, and the same chiplet-equipped 'Vermeer' design as the existing Ryzen 5000 models. As such, the 65W Ryzen 7 5700X's eight-core 16-thread design is identical to the 105W Ryzen 7 5800X — they both even have 32MB of L3 cache.

To differentiate the two, AMD merely trimmed the Ryzen 7 5700X's clock rates to accommodate its 40W lower TDP. As a result, the TDP ratings and clock rates are the only difference between the two chips. This is likely purely the result of artificial segmentation; given the maturity of the TSMC 7nm process, it is unlikely that AMD has many dies that couldn't reach the extra 100 MHz boost that would make them suitable for the 5800X. If so, the company could have simply used them for the higher-priced and higher-margin Ryzen 7 5800X3D.  

Compared to the 5800X, the 5700X's 100 MHz lower boost clock rate will be nearly indistinguishable in most work, but the 400 MHz difference in base clocks will be more pronounced in heavily-threaded workloads at stock settings. However, as we'll demonstrate below, that's extremely simple to rectify with AMD's one-click Precision Boost Overdrive (PBO) auto-overclocking feature. If you engage PBO and have a suitable cooler, there is little to no difference between the 5700X and 5800X in threaded workloads.

AMD broke with tradition at the Ryzen 5000 launch and scuttled one of its biggest value-adds; except for its 65W TDP models, AMD stopped providing 'free' bundled coolers with its Ryzen processors. Unfortunately, now AMD has inexplicably ditched that policy, too, so the 65W Ryzen 7 5700X also comes without a cooler.

The $299 Ryzen 7 5700X faces intense pressure from Intel's Alder Lake from both above and below: The less-expensive $175 Core i5-12400 offers slightly faster gaming performance but trails in threaded applications. However, it comes with a cooler, magnifying the value prop for gamers. Meanwhile, Intel’s $270 Core i5-12600K beats the Ryzen 7 5700X in every facet and also doesn't have a cooler, but it retails for $30 less to offset Intel's higher motherboard costs. This is to say that the Ryzen 7 5700X would have certainly been more compelling if it came with a bundled cooler. 

The Ryzen 7 5700X fully supports overclocking the CPU, including core clocks, memory, and the Infinity Fabric, and will drop into existing 400- and 500-series motherboards (Socket AM4). AMD’s upcoming BIOS updates will also enable support on most older 300-series platforms. You'll need a BIOS with AGESA 1.2.0.6b (or newer) for the Ryzen 7 5700X. AMD says that Ryzen 5000 support will vary by vendor, as will the timeline for new BIOS revisions. However, we should see them all by the end of May 2022. These BIOS revisions also fix AMD’s fTPM stuttering issues.

The 5700X also doesn't support the leading-edge connectivity options, like DDR5 and PCIe 5.0, that you'll find with Alder Lake, but it does support up to DDR4-3200 and PCIe 4.0. AMD won't be able to match intel's connectivity tech until its 5nm Ryzen 7000 ‘Raphael’ Zen 4 CPUs arrive later this year.

• MORE: Best CPUs for Gaming
• MORE: CPU Benchmark Hierarchy
• MORE: AMD vs Intel
• MORE: Ryzen 7000
• MORE: How to Overclock a CPU