The price of DDR5 memory is setting new highs these days as demand badly outstrips supply. In a bid to save money, Meta is recovering legacy DDR4 memory from used servers and is installing it into new machines using its in-house developed Vistara ASIC that enables it to connect old memory modules to its latest servers running AMD EPYC 'Turin' processors that only support DDR5 memory.

Interestingly, Meta is not the only company developing such a solution. Panmnesia, a startup from South Korea, has developed an off-the-shelf CXL controller and switch that enables servers to attach considerably larger memory pools without extending latency, which differentiates Panmnesia’s solution from competing CXL offerings.

Custom ASIC enables DDR4 memory to work with new servers

Vistara is Meta’s first-gen custom CXL memory expander ASIC designed to attach outdated DDR4 memory to modern servers. The chip implements a CXL 2.0 Type-3 memory expander over a PCIe 5.0 x16 interface and bridges standard DDR4 RDIMMs to host processors. Each ASIC supports two independent 72-bit DDR4 memory channels and can provide up to 256 GB of capacity using 64 GB DIMMs. At present, Meta deploys 128 GB per ASIC using 32 GB DDR4 modules recovered from decommissioned servers.

Meta deploys Vistara in its MemServer platform, where two ASICs connect to a single 158-core AMD Turin processor over PCIe 5.0 x8 links. Each server combines 768 GB of DDR5-6400 local memory with 256 GB of CXL-attached DDR4-2400, which expands memory capacity to 1 TB. The software stack transparently exposes CXL memory as a separate NUMA node and enables Linux to migrate cold pages to the slower DDR4 tier (with 76 GB/s of bandwidth) and retain frequently accessed data in local DDR5 (with 614 GB/s of bandwidth).

The ASIC is based on three RISC-V processor cores for secure boot, device initialization, firmware management, and health monitoring. Meta claims it has optimized its CXL controller and memory pipeline to reduce protocol overhead, minimize queuing delays, and lower idle round-trip latency to around 50ns. The chip also incorporates advanced reliability features, including Reed-Solomon two-symbol error correction and x4 chip-kill support.

Not only Meta's Vistara

Meta is not the only company that wants to attach legacy DDR4 memory to newer servers that rely on DDR5 memory and save some money. While Vistara is available exclusively to Meta, there is a new CXL expander solution from Panmnesia that will be available to other companies.

"There has been a perception that putting a switch between the CPU and devices makes it hard to meet the memory-access latency these systems expect, so directly attached multi-headed devices (MHDs) stayed the norm even though they were harder to scale," said Myoungsoo Jung, chief executive of Panmnesia. "Our work shows this is not an inherent limit of CXL or CXL switches — it is a trait of early-stage CXL, and one that fades as the standard and the products around it mature. With a fabric switch that carries our next-stage CXL controller, scalability, low latency, and stable performance can come together."

CXL is a protocol that sits on top of the PCIe physical interface. As a result of this, many early CXL implementations were built by modifying existing PCIe IP, which is why such implementations inherited architectural characteristics optimized for PCIe rather than for memory-semantic communications, which added substantial latency, according to Panmnesia. By contrast, its new CXL controller IP features a redesigned data path that replaces separate per-layer buffers with shared buffers to eliminate much of the synchronization overhead. In addition, it features additional latency optimizations throughout the protocol stack that offset the additional hop introduced by the switch.

The accompanying CXL fabric switch introduces Port-Based Routing (PBR), which removes the tree-topology limitations of conventional Hierarchy-Based Routing (HBR) used by PCIe and early CXL implementations. The fabric switch still supports both PBR and HBR to enable flexible system topologies, optimized traffic routing, and stable performance. In practice, it enables companies like Meta to install more DDR4 memory into their modern servers without major performance degradation because of high latency.

Panmnesia claims that while early CXL deployments could connect only a handful of compute nodes to shared memory pools, its fabric scales to up to 64 nodes, which means greater flexibility for hyperscalers that tend to run thousands of servers, but which now have to rationalize usage of expensive DRAM.

Panmnesia says its next-generation CXL technologies are progressing toward commercialization. The company has pre-release silicon for its PCIe 6.4/CXL 3.2 Fusion Switch and has completed development of its PCIe 7.0/CXL 4.0 Combo IP, which supports the latest features introduced by the CXL 4.0 specification.

The PC industry is bracing for a slip back to DDR4 memory among enthusiasts. Tom’s Hardware has learned at Computex 2026 that both motherboard brands and many module houses — the companies that produce the DIMMs you can buy — are shifting their strategy toward a resurgence in DDR4 platforms as unprecedented memory shortages and price increases continue to raise the entry point into building a PC. Those pain points are particularly acute with DDR5 memory.

We confirmed with over half a dozen sources: motherboard manufacturers and module houses are seeing an increase in demand for DDR4 platforms and shifting production accordingly. This could create challenges, however, as high-performance DDR4 dies, such as the famous Samsung B-dies, are no longer in production. As such, most of the revamped DDR4 kits will top out at a rather pedestrian DDR4-3600.

On the motherboard side of things, at least two vendors confirmed to Tom’s Hardware that they are ramping production of DDR4-supporting motherboards for the second half of the year and into 2027, which makes sense, given that motherboard sales have seen a “collapse” this year, with sales declining by as much as 37% with some vendors. Others have confirmed that they plan to either refresh or re-release DDR4-supporting options later in the year. Many of these products had been in end-of-life (EOL) status, so production lines had long ago shifted to other products. Now new manufacturing capacity will be dedicated to restoring those product families.

This comes as demand for DDR4 platforms has increased. One motherboard brand cited a double-digit increase in sales over the last quarter, which Tom’s Hardware was unable to corroborate, while others simply said the demand has increased significantly.

AMD and Intel have geared up for a shift back to DDR4, as well. AMD launched the Ryzen 7 5800X3D 10th Anniversary Edition at Computex, which the company says it will continue to sell as long as it makes sense; in other words, it’s not a limited edition run. AMD has also shifted the hybrid bonding process of the 5800X3D, setting it up for a more long-term production run. It joins AMD’s Zen 3 XT chips, which it released in 2024 and continues to sell.

Intel continues to sell its Raptor Lake and Raptor Lake Refresh CPUs, as well, and the company told Tom’s Hardware it “will continue to make sure that there are products which can take care of older memory technologies.” One motherboard vendor said it was specifically ramping production of LGA 1700 motherboards with DDR4 support, as options have slowly dried up in the market.

Although there is already DDR4 in the market, it’s also easier to produce, which would help elevate some of the bottlenecks in the current memory supply chain. One of the key shortages right now is advanced packaging, which DDR5 requires with an integrated PMIC. DDR4, by comparison, is much simpler to package and sell, which should help keep prices from climbing into the DDR5 range.

The major chokepoint for DDR4 is wafer allocation, which is a bottleneck that one memory manufacturer (not a module house) pointed to as a potential issue with a shift back to DDR4 platforms. Wafer allocation is a broader bottleneck within the PC industry, however, with Intel even shifting allocation toward the data center as an unprecedented demand for data center CPUs takes hold.

Unfortunately, there appears to be no end in sight. If there’s one thing that nearly all of our sources agreed on, it is that DRAM and NAND shortages will continue throughout all of 2027.

If you want to buy a CPU for $100 today, you have three options. There’s AMD’s Ryzen 5 5500 at $80, alongside Intel’s Core i3-12100F at $90 and Core i3-14100F at a clean $100. Of course, you can go back in time to pick up something like the Core i7-6700 or Ryzen 3 4100, but if you’re after the best CPU for gaming, those chips won’t cut it. I wanted to see how much performance you could get out of a chip for $100, brand new, without mucking around eBay or dusting off DDR3 DIMMs.

Although AMD and Intel have mostly abandoned the sub-$150 price point with newer generations, it’s still a popular segment among budget builders. That’s evidenced by the fact that the Ryzen 5 5500 is (and has been for months) the second best-selling CPU on Amazon. And despite being much lower in the rankings, Intel’s Core i3-12100F ranks higher on that chart than the newer Core Ultra 5 250K Plus.

This segment has only become more important over the past six months, as well. Rising SSD and RAM costs, along with GPU shortages, have bloated the price of building a PC. These chips not only represent a cheap entry point into a relatively modern PC, they also represent a big cost-savings with memory; all three CPUs support DDR4, which is about half as expensive as a kit of DDR5 right now.

We put all three chips through our full gauntlet of benchmarks, as featured in our CPU reviews, to see how they stack up to each other. The goal here is to look at the entry point for building a PC today, taking into account the full cost of a build. You can bump up your budget and get a much better CPU, such as the Ryzen 5 7600X, but that carries with it the baggage of DDR5 prices. We’re capping the budget at $100 to truly segment what the state of entry-level CPUs looks like in 2026.

To that end, we’re mainly comparing these CPUs to each other. In our testing below, we’ll call out some neighboring chips at the budget end of the market, but we’re really focused on our three main test subjects. We evaluate all aspects of the chips as they relate to each other more so than the broader market that starts to pick up closer to the $200 price point. If you want to see more comparisons, make sure to look at our CPU benchmark hierarchy.

$100 CPU Shootout: Specs and Features

$100 CPU Shootout: Specs and Features

The Ryzen 5 5500 and Core i3-12100F are both from 2022, and the Core i3-14100F is from 2024, but the specs for these CPUs feel ripped straight out of 2015. We have two quad-core Intel chips with relatively high boost clocks facing off against an AMD hexa-core chip, and despite AMD packing more cores, it has the cheapest price of the three CPUs in our lineup at only $80 on sale.

Architecturally, the Ryzen 5 5500 is the most dated CPU here, packing six Zen 3 cores that originally debuted in late 2020. However, it doesn’t come from the Vermeer range like most other Ryzen 5000 CPUs. Rather, it falls under the Cezanne codename for Ryzen 5000 APUs, and in particular, the Ryzen 5 5500 lives in a small lineup of those APUs with the integrated graphics disabled. Vermeer and Cezanne are identical when it comes to the node and architectural design. However, Cezanne only supports up to PCIe 3.0, while Vermeer supports PCIe 4.0.

Outside of the PCIe limitation, the Ryzen 5 5500 is a familiar Zen 3 CPU. It comes with a relatively large L3 cache at 16 MB, and you’ll find 512 KB of L2 cache per core. Power demands are low with a 65W TDP and PPT of 88W, and the chip comes bundled with AMD’s Wraith Stealth cooler, which is enough to keep the chip within normal operating temperatures. It exclusively supports DDR4 memory, with official speeds up to 3,200 MT/s, and it slots into AM4 motherboards. As usual with AMD’s long-standing AM4 support, you should check compatibility with your motherboard; the Ryzen 5 5500 is technically supported on 500- and 400-series chipsets, and even some 300-series chipsets, but support varies from board to board. You will likely need an older AM4 chip to flash a BIOS update, as well (short of BIOS Flashback features available on some boards).

The Core i3-12100F and Core i3-14100F are both similar chips, leveraging the same silicon with minor modifications. Starting with the Core i3-12100F, it’s a quad-core chip from Intel’s Alder Lake range. Unlike other 12th-gen CPUs, however, it doesn’t come with a hybrid architecture. It comes with four Golden Cove P-cores and no E-cores, giving you access to a total of eight threads.

Although the Core i3-14100F is two generations newer, it uses the same die with H0 stepping as the Core i3-12100F, just with higher boost clocks and a more aggressive MTP, along with elevated base clocks. Both chips come with a pool of 12 MB of shared L3 cache, along with 1.25 MB of L2 cache per core, totaling 5 MB across the chip.

Both chips are compatible with LGA 1700 socket motherboards, including 600- and 700-series chipsets, though you may need a BIOS update to use the former. Both chips also support DDR4 and DDR5 memory given you have a compatible motherboard. DDR4 and DDR5 are physically incompatible, so make sure you have the proper motherboard version before picking out your memory.

Like AMD, Intel offers a bundled cooler with both chips: the Laminar RM1. Given the low power demands and locked multiplier, the RM1 is enough to keep both chips within operating temperatures. However, Intel’s 12th- through 14th-gen chips tend to run hot, so investing in an inexpensive tower cooler ($20 - $30) is worth it. Unlike the Ryzen 5 5500, both Intel chips support PCIe 4.0 for storage and graphics.

We’re looking at the F-series variants of the Core i3-12100 and Core i3-14100 without integrated graphics, which are the versions you’ll still find available for around $100. The pricier versions with integrated graphics perform identically, though they’re way too expensive to recommend right now with resellers asking north of $220 for them.

$100 CPU Shootout: Gaming Performance

$100 CPU Shootout: Gaming Performance

When we review CPUs at Tom’s Hardware, we use the most performant gaming GPU available to consumers in order to isolate CPU performance as much as possible — that’s currently the Nvidia RTX 5090 FE. This approach has a flaw when we’re looking at CPUs as weak as the Ryzen 5 5500, Core i3-1200F, and Core i3-14100F, however. We are forcing the system into a full CPU bottleneck and ignoring the influence of the GPU in a budget system running one of these chips, which is an influence that we need to account for in this situation.

Performance with the RTX 5090 FE is important, as we’ve isolated CPU performance as much as possible to get an accurate view of how these chips compare to one another. However, we also ran a test pass with an Asus Dual RTX 4060 8GB as a more grounded, “real-world” comparison point. The choice to use the RTX 4060 was deliberate, as it only has 8GB of VRAM, and some of the games in our test suite have performance issues with 8GB graphics cards. If you’re buying one of these CPUs, there’s a good chance you’re pairing it with a GPU that has 8GB (or less) of VRAM, so we wanted to reflect that situation in our testing.

You can see both configurations represented in our chart above, with a natural split between the RTX 5090 and RTX 4060. We’re testing with DDR4 here to keep the playing field level, as well as focus on the budget builders that don’t want to shell out for a DDR5 kit. Based on our testing, adding DDR5 into the mix with the Core i3-12100F or Core i3-14100F represents around a 5% to 8% improvement, depending on the title.

Performance falls in-line with price, with the Ryzen 5 5500 at the bottom and the Core i3-14100F at the top. With the RTX 5090, the Core i3-14100F was a minor 3.8% faster than the Core i3-12100F, but 11.8% faster than the Ryzen 5 5500. With the RTX 4060, the difference between the two Intel CPUs is less than two frames (about 2%), while the Core i5-14100F is 10.7% ahead of the Ryzen 5 5500. The scaling is slightly less dramatic with the RTX 4060, but it’s still very much present.

In our other geomeans, you can see the Core i3-14100F consumed the most power out of our test pool, averaging 55.1W across our suite of games. This chart is interesting because it shows the influence of the GPU, and how the CPU becomes the bottleneck as it tries to keep pace with a much faster GPU.

The Core i3-14100F was the least efficient of our test pool, while the Core i3-12100F was the most. But really, we’re dealing at the extreme low end here. All of these CPUs are exceptionally efficient in games, so much so that we had to reset the X-axis on our efficiency chart to fit the data in. As we’ll get to later, none of these chips come close to 100W, even under a full workload. And in games, less than 50W is the norm.

Despite drawing the most power, the Core i3-14100F sits in the middle of our temperature rankings. Although these are great thermal results, we test with a 360mm all-in-one liquid cooler. If you plan on using the stock cooler included with any of these CPUs, expect much higher temperatures under load.

Baldur’s Gate 3 Benchmarks

At 1080p, Baldur’s Gate 3 is completely CPU-bound with this test pool, as we can see virtually identical performance across both the RTX 4060 and RTX 5090 passes. The Core i3-14100F is a clear favorite here, outpacing the Ryzen 5 5500 by just over 10% overall, and coming in a few frames ahead of the Core i3-12100F.

Borderlands 4 Benchmarks

Borderlands 4 is one of the reasons we opted to test the RTX 4060, as this game struggles with 8GB GPUs. With the RTX 5090, we see the Core i3-14100F only 3% ahead of the Ryzen 5 5500 and 6% ahead of the Core i3-12100F. However, the Intel chips handled the 8GB RTX 4060 more gracefully, with the Core i3-14100F outperforming the Ryzen 5 5500 by 21.9%.

Regardless of your CPU choice, playing this game with an 8GB graphics card is a bad experience with consistent stuttering, as shown by the 1% lows.

Crimson Desert Benchmarks

Crimson Desert is another game where it helps to have data from the RTX 4060. With the RTX 5090, the Core i3-14100F outpaces the Ryzen 5 5500 by 13.5%, but that lead shrinks to just 4% with the RTX 4060 as the GPU becomes a bigger influence on performance.

Counter-Strike 2 Benchmarks

We see the opposite behavior in Counter-Strike 2, with the CPUs offering identical performance with the RTX 5090, but the Core i3-14100F coming out 6.8% ahead of the Ryzen 5 5500 with the RTX 4060, likely on the back of its boosted clocks. Despite similar overall performance, the Intel CPUs show better stability in 1% lows.

Cyberpunk 2077

In Cyberpunk 2077, there’s a wall that the Core i3-14100F runs into around 68 FPS, but we see scaling with the different GPUs with the Core i3-12100F and Ryzen 5 5500. Regardless, the Core i3-14100F clearly leads here.

Doom: The Dark Ages Benchmarks

Doom: The Dark Ages is mostly GPU-bound with its always-on ray tracing, and that’s clear immediately in our chart. Still, the Core i3-14100F beat the Core i3-12100F by 3.2% and the Ryzen 5 5500 by 4.5% with the RTX 5090. With the RTX 4060, the two Intel chips are in lockstep, beating out the Ryzen 5 5500 by around 4%.

F1 2024 Benchmarks

We can see our test pool is once again bound by the GPU with the RTX 4060 in F1 2024, with the Core i3-14100F marginally outclassing the two other chips by around 2%. Shifting the load back to the CPU with the RTX 5090, the Core i3-14100F is a massive 17% ahead of the Ryzen 5 5500, as well as 5% ahead of the Core i3-12100F.

Far Cry 6 Benchmarks

Final Fantasy XIV Benchmarks

Flight Simulator 2024 Benchmarks

Flight Simulator 2024 is one of the more interesting benchmarks for this test pool. With the RTX 4060, you can see we’re completely GPU-bound. However, with the RTX 5090, the Core i3-14100F and Core i3-12100F post identical results, which are more than 20% ahead of the Ryzen 5 5500.

Hitman 3 Benchmarks

Hitman 3 shows strange results at first glance, but that’s mainly a consequence of including two GPUs in the same chart. What we’re really seeing here is that the Core i3-12100F and Core i3-14100F offer identical performance in this title, which is marginally ahead of the Ryzen 5 5500 with the RTX 5090 and about 13% ahead with the RTX 4060.

Marvel Rivals Benchmarks

We run into the exact same situation with Marvel Rivals, though the data is a bit cleaner. The two Intel CPUs are about 8.5% ahead of the Ryzen 5 5500 with the RTX 5090, and just shy of 4% ahead with the RTX 4060. This is an Unreal Engine 5 game, and UE5 is generally GPU-bound, but we still see a bit of scaling here.

Spider-Man 2 Benchmarks

Again, the two Intel CPUs are in lockstep here, posting identical performance with the RTX 4060. With the RTX 5090, the Core i3-14100F shows a minor 2.1% jump over the Core i3-12100F. The Ryzen 5 5500 really struggles in this game. Intel is around 20% ahead with the RTX 5090, and about 25% ahead with the RTX 4060.

Starfield Benchmarks

We run into a GPU-bound situation with the RTX 4060 in Starfield, and all three chips posted similar performance with the RTX 5090. Still, the Core i3-14100F technically leads here, beating out both other chips by around 7% with the RTX 5090.

$100 CPU Shootout: Productivity Performance

$100 CPU Shootout: Productivity Performance

The CPUs in our test pool are on the lowest-end of chips you can still find in stock at retailers, and because of that, they’re going to end up at the bottom of the charts below. That’s how these chips stack up to some more expensive options around $150 to $200, but again, our focus here is to look at your options if you only have $100 to spend. So, although we’ll show the context of other CPUs in our charts, we’re mainly focusing on the comparison between the three chips in our test pool.

As with games, we stuck with 32GB of DDR4 memory running at 3,200 MT/s for the three chips we’re looking at. In our charts, we’ve also included the Core i3-13100F running with DDR5-4800 memory as a comparison point for what you can expect out of the Core i3-12100F and Core i3-14100F if you opt for a DDR5 motherboard. However, we’re focused mainly on the DDR4 performance here given how largely that influences the total cost of a budget build.

Starting with multithreaded performance, The Ryzen 5 5500 unsurprisingly tops the charts with a 50% increase in core/thread count compared to the two Core i3s. Those extra two cores drive a 13.7% improvement over the Core i3-14100F and an 18.2% jump over the Core i3-12100F.

Looking at Intel, the Core i3-14100F has a minor 3.9% improvement over the Core i3-12100F, matching the Core i3-13100F running with DDR5 memory. All three of these CPUs use the same silicon, just binned differently, and that really shows up in our multithreaded geomean. Unlike games, where DDR5 memory can represent a decent jump in performance, you shouldn’t expect more than a 5% improvement in heavily-threaded workloads.

The script flips in single-threaded performance, with the Ryzen 5 5500 sliding into last place with its limited 4.2 GHz boost clock. The Core i3-12100F is 11.9% faster, while the Core i3-14100F is 22.1% faster. Comparing the two Intel CPUs, the Core i3-14100F is 9% ahead of the Core i3-12100F.

Single-threaded performance levels the playing field in our test pool a bit, with the Core i3-14100F actually coming out ahead of the Ryzen 5 7600X. However, the two Intel CPUs have locked multipliers, so you can’t squeeze out extra single-threaded performance by overclocking. The Ryzen 5 5500, on the other hand, has an unlocked multiplier and supports Precision Boost Overdrive (PBO). Manually tuning the Ryzen 5 5500 won’t change the performance story here broadly, but it could help close the gap in single-threaded workloads.

Rendering Benchmarks

Starting with rendering tasks, the Core i3-12100F and Core i3-14100F broadly take a backseat to the Ryzen 5 5500 in multithreaded workloads, with the two Intel chips sitting close to each other in the rankings. Despite how close they are, the Core i3-14100F offers some meaningful improvements, such as an 8.1% jump in Blender Monster. Otherwise, the Ryzen 5 5500 wins in multithreaded rendering tests, short of V-Ray 6, where it matched the Core i3-14100F.

The Ryzen 5 5500 falls into last place when looking at single-threaded rendering via Cinebench and POV-Ray. The Core i3-12100F is 33% faster in POV-Ray, while the Core i3-14100F is 45.7% faster. Again, this feels like a flashback to generations past, with Intel excelling in single-threaded performance with a quad-core chip, while AMD makes up lost ground with weaker single-core performance but a large array of cores to work with.

Encoding Benchmarks

The performance picture with encoding workloads is almost identical to what we can see with rendering, mainly due to the fact that encoding workloads are either heavily-threaded or exclusively single-threaded. In video encoding via Handbrake, AMD leads by upwards of 17% depending on the codec. Outside of Handbrake, we can see the Ryzen 5 5500 leading by a margin of 13% in our HEVC encode, and by a margin of 25% in our AV1 encode.

Once again, single-threaded encoders show big leads for Intel. In a standard LAME run, the Core i3-14100F outpaces the Ryzen 5 5500 by 13.8%, which shrinks to a 12.8% lead when looking at an extended run. In a single-threaded JPEG-XL decode, the Core i3-12100F is 8% faster than the Ryzen 5 5500, while the Core i3-14100F is 18.8% faster.

Creative Application Benchmarks

The Adobe suite features a broad range of workloads that helps drill down on which CPU from our test pool really stands out when considering tasks that aren’t exclusively single- or multithreaded. Starting in Photoshop, there’s a bit of an upset. AMD’s new CPUs usually excel in Photoshop, but we can see the Core i3-14100F taking a marginal lead over the Ryzen 5 5500 overall. Still, AMD managed to beat the Core i3-12100F by 4.6% in this test.

In video editing, the Ryzen 5 5500 claims the lead in Premiere Pro, outpacing the Core i3-14100F by 3.9%. The same is true in DaVinci Resolve, though the Ryzen 5 5500 holds a more commanding 8.1% lead. After Effects mirrors what we can see in Photoshop, with the Core i3-14100F marginally leading the pack.

Broadly, the two extra cores available to the Ryzen 5 5500 help out in video editing workloads. However, the margins aren’t as large here, with AMD usually leading by less than 10%, and in many cases, less than 5%. Although it’s interesting to inspect the difference between the chips in our test pool, context is important here. None of these CPUs are well-suited for creative workstations, so you’ll likely run into performance limitations regardless of which chip you use.

Web and Office Benchmarks

Web and office performance is important for our test pool. Although you should expect some bumpy performance with more demanding apps like the Adobe suite, basic web and office applications are firmly in the wheelhouse of budget CPUs. Most of our tests here are lightly-threaded, giving Intel a leg up with its superior clock speeds.

Starting with WebXPRT 4, the Ryzen 5 5500 and Core i3-12100F match each other at the bottom of the chart, while the Core i3-14100F claims a lead of 8.8%. In Tesseract OCR, both Intel chips come out ahead, with the Core i3-12100F completing the text-to-image scan 5% faster, and the Core i3-14100F completing it 23.7% faster.

There’s a big divergence in the Microsoft Office suite, as well, with both Intel chips coming out ahead of the Ryzen 5 5500. In Excel, the Core i3-12100F is 10% ahead of the Ryzen 5 5500, while the Core i3-14100F is 23% ahead. The story is similar in Powerpoint, with the Core i3-12100F beating the Ryzen 5 5500 by 12.5%, and the Core i3-14100F beating AMD by 20.5%.

Security, Compression, Chess Engines, and Workstation Benchmarks

For this test pool, we’re putting a lot of emphasis on our main suite of encoding, rendering, creative, and general productivity benchmarks. However, we still ran these chips through our normal suite, which includes a broad range of workstation tasks, spanning everything from data science to web server workloads.

We’ve included the results in the album above if you’re interested in taking a look, but none of the chips we’re looking at today are well-suited for workstation applications.

$100 CPU Shootout: Power and Efficiency

$100 CPU Shootout: Power and Efficiency

All three chips in our test pool barely sip power. So much so that you can reasonably get by with the bundled cooler included with each CPU. Throughout our power testing, the highest result we recorded across the test pool was 78W, which the Core i3-14100F climbed to during an all-out render via Blender. Even with more power available to the platform, these chips play things safe.

In some power tests, you can see the three chips we tested in lockstep. There’s no meaningful difference in power consumption between them in a multithreaded Y-Cruncher pass, nor a significant difference in Linpack. We can see more significant differences in Cinebench, Blender, and Handbrake, however.

In multithreaded Cinebench 2024, the Core i3-12100F consumed 16.6% more power than the Ryzen 5 5500, while the Core i3-14100F consumed 23.3% more power. In Blender, the Core i3-12100F is 24% ahead of AMD and the Core i3-14100F is 56% ahead. Even in these workloads, the Ryzen 5 5500 isn’t climbing over its rated 65W TDP, which is a good thing. Given that it’s unlocked for overclocking, you certainly have some power headroom to play with.

There’s a big divergence in demanding workloads between our three chips, but an even bigger split between Intel and AMD when looking at idle power consumption. In a true idle state, both Intel chips consumed nearly triple the power of the Ryzen 5 5500. And in an active idle situation (YouTube playback), Intel consumed more than double the power. In both cases, we’re looking at a difference of 10W, but that’s still significant considering just how little power the Ryzen 5 5500 requires.

Turning to efficiency, the Ryzen 5 5500 comes out on top with its overall lower power draw, though the margins are tighter. We’re mainly looking at heavily-threaded workloads when talking about peak power consumption, so it’s no surprise to see the Ryzen 5 5500 dominate in efficiency. It offers better multithreaded performance due to packing six cores, and it requires less power than the Intel chips overall.

$100 CPU Shootout: Pricing and Platform Considerations

$100 CPU Shootout: Pricing and Platform Considerations

These CPUs are all around the same price, ranging from $80 to $100 depending on availability and sales. However, there’s a broader pricing conversation when looking at the overall platform. With Intel, you have the option between DDR4 and DDR5 along with an accompanying motherboard, while with AMD, you’re locked to DDR4 but with more plentiful motherboard options.

Starting with the motherboard, you can find an AM4 board for the Ryzen 5 5500 for as little as $60, though you should expect to spend around $80 to $100 on a decent board. AMD’s B550 chipset works here, as you’re able to overclock the Ryzen 5 5500 on that chipset. Technically, you can use 400-series and even some 300-series chipsets, though you should double-check compatibility and be prepared to flash a new BIOS using an older AM4 CPU. Despite the age of AM4, you can still find motherboards in stock at just about any retailer.

Intel is trickier. You can find Socket LGA 1700 motherboards for as little as $70, though most LGA 1700 boards only support DDR5 memory. Surprisingly enough, you’ll spend a premium on a DDR4 motherboard, likely due to low inventory. The cheapest board we could find is the Asus B760M-AYW Wi-Fi D4 for $90, though as the name suggests, it’s a Micro ATX board. For full ATX, you can pick up the MSI Pro B760-P Wi-Fi DDR4 for $140. If you want to jump to DDR5, you can do so for the same price with the MSI MAG B760 Tomahawk.

Both 600-series and 700-series chipsets work for both Intel chips, though you’ll need to flash a new BIOS if you plan on pairing the Core i3-14100F with a 600-series chipset.

One of the big reasons to go with one of these CPUs is DDR4 support given that DDR5 prices have shot through the room. A kit of Teamgroup T-Force Vulcan Z memory will run you about $130 for a 16 GB (2 x 8 GB) kit at 3200 MT/s. If you want something fancier, the Corsair Vengeance RGB Pro 32 GB kit (2 x 16 GB) clocks in at $210 at the time of writing, also at 3200 MT/s.

If you spring for DDR5, expect to spend about double what you spend on DDR4. Currently, one of the cheaper DDR5 kits around is the Corsair Vengeance 32 GB (2 x 16 GB) kit at 6000 MT/s, which is around $450.

Using DDR4 as the baseline, the Ryzen 5 5500 ends up around $310 for the full platform, while the Core i3-14100F will run you about $370 ($10 less if you go for the Core i3-12100F). That’s about 19% more expensive going with one of Intel’s chips compared to the Ryzen 5 5500.

There’s some upgrade potential regardless of the platform you go with, though Intel certainly has a leg up with availability. Intel’s fastest gaming CPU is still the Core i9-14900K (the new Core Ultra 7 270K Plus is marginally slower), and it’s available for sale alongside most of the 14th-Gen lineup. You’ll need a Z-series chipset in order to overclock a K-series SKU, however, along with a much beefier cooler.

Despite the long-standing legacy of AM4, there aren’t a ton of Ryzen 5000 CPUs available for sale. Outside of the lower-end offerings like the Ryzen 5 5500, you’ll mainly find AMD’s XT refresh chips. Rumors suggest that AMD is planning on re-releasing the Ryzen 7 5800X3D, but that hasn’t been confirmed. In fact, you don’t have a reasonable way to unlock AMD’s coveted 3D V-Cache without spending top dollar on the secondhand market for an X3D AM4 chip.

Regardless of the platform, both AMD and Intel have moved onto newer sockets and chipsets, so you won’t have an opportunity to upgrade to a newer generation down the line. Intel has an edge if you’re moving within the existing product lineup, mainly due to the availability of Raptor Lake chips at retailers.

Test Setup and Notes

We used the same testing procedure for this $100 CPU shootout that we follow in our CPU reviews, including identical test benches short of the CPU and motherboard. We also make some tweaks to the BIOS and operating system to maximize performance while limiting sources of variation between different chips.

That includes turning off Virtualization-Based Security, enabling Resizable BAR, turning on XMP/EXPO (or DOCP in this case), and disabling any automatic boosting features. Intel doesn’t cover its Extreme power profile under warranty, nor does AMD cover Precision Boost Overdrive (PBO), so we manually disable these features.

$100 CPU Shootout: Choosing a Winner

$100 CPU Shootout: Choosing a Winner

Out of the three CPUs under $100 we looked at, the Core i3-14100F is the winner. It’s not as powerful as the Ryzen 5 5500 in multithreaded workloads, but then again, none of these CPUs are great for heavily-threaded tasks. It still manages to trade blows with the Ryzen 5 5500 app-to-app, and it takes a lead when looking at lightly-threaded workloads.

Gaming makes a big difference at this price, however; the less you spend on a CPU, the more you can invest in a better GPU. In games, the Core i3-14100F leads, no caveats required. Even when going down to the RTX 4060 for our more grounded testing scenario, the Core i3-14100F shows consistent scaling over the Ryzen 5 5500. Against the Core i3-12100F, that advantage is less consistent across titles, but it’s still present.

The gaming results bring up an interesting point about core count in the context of gaming, as well. For years, a quad-core was the gold standard for a gaming PC; anything more than that was generally considered a waste if you were just gaming. That slowly crept up to six- and eight-core chips becoming the standard, which might dissuade you from considering a pure quad-core like the Core i3-14100F.

I’m not advocating for more quad-core CPUs. We’ve moved past that point with modern architectures, but a head-to-head battle like this shows just how much architecture overshadows specs, even in a relatively tame comparison between a quad-core and hexa-core CPU without any hybrid architectures or chiplets to contend with.

Outside of performance, the platform plays a big role in choosing among our budget CPUs. You’ll spend more on an Intel platform, not only on the CPU, but also on a DDR4 motherboard. However, there are some significant advantages to Intel’s platform in the context of our test pool.

Most importantly, you get PCIe 4.0 with Intel. Most AM4 boards support PCIe 4.0, as well, but you’ll need to upgrade beyond the Ryzen 5 5500 to unlock it. PCIe 4.0 support of the box is important for a budget platform. We’ve seen budget GPUs like the AMD RX 6500 XT that run into severe performance limitations on PCIe 3.0 due to only using an x4 interface. And, of course, you have access to faster storage.

Upgrades are easier on the Intel platform, as well. Although AMD bolstered AM4’s longevity with its XT refreshes, it moved exclusively to DDR5 and a new socket with Zen 4. During that same period, Intel was releasing LGA 1700 CPUs on both DDR4 and DDR5 platforms. The ceiling is higher on an LGA 1700 platform right now, and short of some sort of re-release, that will likely remain the case.

A lot of the results and conclusions here are counterintuitive if you’re accustomed to the battle we see play out between Intel and AMD in the main product stack each generation. AMD has often been defined by platform longevity and leadership in gaming performance, while Intel excels in application performance with its hybrid architecture and peak clock speeds. In this test pool, we see the opposite. AMD excels with heavily-threaded workloads, while Intel takes the lead in gaming and offers a platform with solid upgrade potential.

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

PC memory manufacturer Team Group has agreed to settle a class action lawsuit to the tune of $1.1 million over allegations it advertised deceptive speeds for its DDR3, DDR4, and DDR5 memory products purchased in the U.S. between May 3, 2020, and April 8, 2026. Similar to a recent case involving G.Skill, the lawsuit claims that consumers were led to believe that the advertised speeds on Team Group’s memory kits could be achieved out of the box without requiring BIOS tweaks or overclocking profiles.

It was alleged by the plaintiffs that the company marketed its RAM kits using rated speeds that could only be achieved after enabling XMP or EXPO memory profiles via the motherboard BIOS/UEFI menu. Team Group has denied all allegations of wrongdoing and says that its “products were appropriately labeled and performed as represented.”

An individual can claim as part of the settlement class if they purchased Team Group DDR3, DDR4, or DDR5 memory products while living in the United States between May 3, 2020, and April 8, 2026. The settlement only applies to individual consumers, while purchases made directly by a business, such as a company, LLC, corporation, or partnership, do not qualify for compensation. However, if you personally bought the memory as an individual consumer, you may still qualify even if you later used the product for work or business purposes.

As there is no fixed amount out of the total settlement fund of $1.1 million, the money will be divided among every individual who submits a valid claim. Payments will be made based on how many eligible Team Group DRAM products each person claims. The settlement also clarifies that one may claim compensation for up to five memory products per household without providing proof of purchase, while claims beyond that require supporting documentation.

Eligible customers can head to the Claim Hub website and follow the instructions carefully to file an official claim. You can also submit an objection or exclude oneself from the lawsuit completely. The deadline for the claim is currently set for July 7, 2026.

Earlier this year, G.Skill settled a similar $2.4 million lawsuit involving its DDR4 and DDR5 memory products. Plaintiffs argued that advertised speeds above JEDEC defaults required additional BIOS adjustments and overclocking settings that were not clearly disclosed on packaging or product pages. Most memory kits for modern PC platforms ship with conservative default speeds based on JEDEC standards, while higher advertised frequencies, such as DDR5-6000 or DDR5-7200, require users to manually enable XMP or EXPO profiles. The lawsuit argued that average consumers may not realize these additional steps are necessary to achieve the marketed performance levels.

G.Skill was also required to change its packaging and be clearer about overclocking and BIOS adjustments as part of its agreement. The Team Group settlement doesn't seem to include any such provisions and is rather purely offering a payout to affected buyers.

If you're building projects based on Nvidia Jetson modules, don't panic; it's fairly likely that you're working with the newer Orin and Thor modules. If, however, your project is based around the older Xavier or TX2 boards, then we have some bad news: Nvidia has apparently brought forward its EOL timelines for those families of Jetson products due to the RAMpocalypse making LPDDR4 harder to get. That's according to Connect Tech, a Canadian supplier and system integrator for AI systems that says Nvidia has moved these devices to Non-Cancelable, Non-Returnable (NCNR) status due to the change. (Shout out to CNX Software for spotting this.)

The specific devices affected at Connect Tech are the Jetson TX2 NX, the Jetson TX2i (all SKUs), the Jetson AGX Xavier 32GB Industrial variant, and the Jetson Xavier NX in 8GB and 16GB versions, but the supplier actually says that Nvidia has marked all TX2 and Xavier models as NCNR on its side. It also says that final purchase orders for those modules must be in by July 1st, existing purchase orders convert to NCNR on July 15th, and the last time it will ship any orders featuring those products is July 15th next year.

Strictly speaking, those timelines are from Connect Tech, not Nvidia itself, but the company does say that they are "based on Nvidia timelines." In any case, it's not really that surprising; all of these are older models, with the TX2 having been introduced in 2017 and the Xavier parts originally introduced in 2018, though some of the specific variants included arose as late as 2021.

Given that, it's probably time to start thinking about moving to newer hardware if you're working with these parts. Nvidia's Orin NX is close to a drop-in replacement for the older Xavier NX platform as long as you are not relying on more specialized I/O configurations, and the overall form factor and power envelope remain in the same class. Moving from AGX Xavier to AGX Orin is even more straightforward since both use the same 699-pin connector family, although power delivery and thermals still need to be validated.

These are hardly the first casualties of the RAMmageddon, but they do complicate the idea that only DDR5 is under pressure. DDR4 pricing has skyrocketed due to lacking supply since memory manufacturers moved on to DDR5. What is happening here is not a simple shortage in the traditional sense but a reallocation of manufacturing capacity. Memory vendors are prioritizing higher-margin parts for AI accelerators, particularly HBM and newer DDR5, and that shift pulls capacity away from legacy nodes that produce LPDDR4.

That leaves older embedded platforms in an awkward position. They depend on memory that is no longer the focus of the industry, yet they still require long lifecycle guarantees that newer consumer products do not. When supply tightens, those platforms are often the first to be pushed into constrained ordering terms, and NCNR status is usually the earliest visible signal.

Seen in that light, this is less about an abrupt discontinuation and more about a market reality finally catching up with aging hardware. The TX2 and Xavier families were already living on borrowed time, and the current memory crunch simply accelerates a transition that was going to happen anyway. For developers and integrators, the takeaway is that if a design still depends on LPDDR4-based Jetson modules, the window to secure supply is closing, and migration to Orin is no longer just a performance upgrade but a matter of long-term viability.

The spot price of a 16GB DDR4 chip slipped roughly 5% over the past month to around $74.10, the first monthly decline since February 2025 and the first crack in a rally that pushed the same part from around $3.20 a year earlier, according to DigiTimes. This marks a notable decline from the 2,200% increase in pricing seen earlier this year.

16GB DDR5 fell by a comparable amount to roughly $37.20 over the same period, with the sharpest drops showing up in Chinese channel listings and on Amazon, where some 32GB DDR5 kits came down by as much as 30%. Even after the pullback, 16GB DDR4 is trading at more than 20 times its price just 12 months ago.

None of this has reached the contract market, where PC OEMs and system builders actually source memory. TrendForce's latest survey, which we covered last week, projects another 58% to 63% quarter-on-quarter rise in conventional DRAM contract prices in Q2 2026 and a 70% to 75% jump in NAND Flash, following record Q1 increases of 90% to 95% on DRAM.

DigiTimes reported that neither the three major suppliers nor Chinese memory makers CXMT and YMTC have eased pricing, and that customers are still signing long-term supply agreements. Spot sales represent a small fraction of total memory shipments, so a correction in channel listings has little direct bearing on OEM pricing.

It’s the Chinese channel that has moved the most, however, with DigiTimes reporting 32GB DDR5 kits down 27% over the month, while 8GB and 16GB DDR4 modules posted weekly declines of 25% in data the outlet attributed to China Flash Market.

Two factors appear to be driving the sell-off. First is that distributors who had built inventory at the top of the rally began clearing it once smaller module vendors ran out of headroom to pass costs downstream. Weak consumer demand has been a running theme since late 2025, when Samsung signed a non-cancellable DDR4 supply agreement that wouldn’t flow through to consumer buyers.

The second factor is Google's TurboQuant announcement in late March, a memory compression technique the company's research arm claims can cut key-value cache memory use during large language model inference by a factor of at least six. DigiTimes said the disclosure prompted some stockpilers to unload positions on concern that hyperscaler memory demand could soften if similar techniques reached production at scale.

DDR3 and legacy MLC NAND continue to climb despite the broader spot-side cooling, tracking the end-of-life pricing pattern we’ve seen as manufacturers pull the plug on DRAM in the pursuit of AI profits. Several suppliers are exiting those nodes entirely, and tight residual supply tends to support pricing on the way out. Retail DDR5 kit listings remain volatile, with triple and, in some cases, quadruple increases over the last three months.

Angeion Group, a settlement administration firm, is calling for eligible G.Skill customers to claim their cash from the company. According to the company’s X post, people who purchased a G.Skill DDR4 or DDR5 RAM kits may be eligible to receive a part of the $2.4 million that the company paid to settle the deceptive advertising case filed against it.

The plaintiffs said that “they were lead to believe that the advertised speeds were 'out of the box' speeds requiring no adjustments to their PCs," and they alleged “that G.Skill deceptively advertised and labeled the speed of its DDR-4 and DDR-5 DRAM (non-laptop) memory products with rated speeds over 2133 MHz or 4800 MHz, and that G.Skill is liable for violations of consumer protection statutes and breach of express warranty." While the company refused to admit any wrongdoing, it decided to settle the case to avoid “the uncertainties, burdens, and expenses associated with ongoing litigation.”

Eligible claimants must have purchased a G.Skill DDR4 memory module with a rated speed of over 2133 MHz or a G.Skill DDR5 rated above 4800 MHz between January 31, 2018, and January 7, 2026, while residing in the United States. You also do not need to present proof of purchase if you’re making five or fewer claims. So, if you think you’re qualified to file a claim, you can fill out the claim form on the official Settlement Website until April 7, 2026.

However, even though the settlement amount is relatively large at $2.4 million, the amount each claimant will get will vary depending on the total number of claims. Claim Depot also reports several deductions from the total amount, including administration costs of $295,000, attorney’s fees of up to $800,000, attorney’s expenses (to be determined), and service awards to class representatives of up to $10,000. Because of this, the amount you get (if you’re eligible for the claim) could be significantly less than the price of one of the best RAM kits you can get today. But even if that is the case, every penny still counts, especially as we struggle with skyrocketing memory and storage prices.

After months of consistent decline, Intel’s consumer CPUs have enjoyed a small but significant spell of market share growth, at least according to the latest Steam Hardware Survey (January 2026 data). The latest statistics raise a couple of key questions. Firstly, could this mark the beginning of an Intel comeback? Secondly, why might Intel platforms be more attractive in the current PC market?

So, the change in share is quite small, at 0.25% up in Intel’s favor, in January 2026 data. We know the SHWS hasn’t got the strongest reputation for rigorous statistical accuracy, despite the size of the user base. Nevertheless, it is entertaining to ponder whether something larger is happening in the CPU market, and why the balance could be shifting.

An Intel comeback?

Though Intel has broken its downtrend in this latest set of survey statistics, it isn’t possible to say whether this is truly a change in fortunes. As Aristotle (probably) said, “one swallow does not a spring make.” So, one result doesn’t provide a lot of certainty for extrapolations.

We’d need three months of data to confirm a popularity plateau, resistance point, or reversal of fortunes for Intel in the gaming CPU market. That means we are still a long way off the establishment of a reliable trend.

Why might Intel gain market share in 2026?

If we assume that Intel is once again going to consistently grow its consumer CPU market share, it is interesting to look at its current strengths and recent platform announcements.

Ahead of the RAMpocalypse, PC DIY forums and social media generally seemed to steer builders toward AM5 plus DDR5 for creating a potent ‘future-proof’ PC system. After several months of exploding RAM pricing, this particular path to PC Nirvana has lost its glossy shine.

In the PC DIY space, pricing and availability may have worked in Intel’s favor over recent months. The iconic PC chipmaker’s 13^th and 14th-generation processors appear to have remained in stock, often discounted. They can be used with cheaper DDR4 RAM (that upgraders may already have), with plentiful well-priced DDR4 motherboards (e.g. B760) still at retailers. Moreover, these remain decent platforms for most gamers, who will typically be performance-limited by their GPU choice. We've also seen some fairly drastic and generous discounts on Intel's current generation processors, with Core Ultra 200 series chips bundled with motherboards, coolers, popular games, or even just slashed in price in the second half of 2025.

A cultural shift from future-proofing to buying the best value $/FPS system today seems to have taken hold in Q4 2025 and persists to this day.

At CES 2026, we also noticed more positives for Intel. Its next-gen Panther Lake chips for laptops were far more warmly welcomed than AMD’s Ryzen AI 400 series refresh. This won’t have started to trickle through to Steam survey results, of course, but Intel traditionally has stronger laptop-maker support, and Panther Lake should ensure that isn’t eroded in 2026.

In today's market, memory has become one of the most critical and expensive components in any build, and prices have only worsened thanks to the ongoing global DRAM shortage. We're keeping an eye on the volatile memory pricing to help identify any good deals that may arise.

Now, even more so than before the shortage began, price tags are fluctuating significantly from day to day, and it's becoming increasingly complex to tell when you're getting a fair deal versus overpaying for the same memory kit. We've compiled this comprehensive RAM price tracker that monitors the availability and pricing of different DDR5 and DDR4 memory kits across major U.S. retailers. Our goal is to take the guesswork out of memory shopping and help you make an informed purchasing decision.

We update this index regularly with the lowest prices available in the U.S. market for each memory kit. Our tracker covers a range of capacities and speeds across both budget and premium brands. We list the lowest price for each available memory kit, regardless of manufacturer, so you see the absolute best deals currently on offer.

The general availability of many memory kits is limited, with truly exceptional deals scarce and often selling out within seconds of appearing. High-capacity kits, particularly DDR5, frequently show the most severe stock shortages and price premiums. Therefore, timing and vigilance are especially important when shopping for memory in current times. Whether you're budget-conscious and want the best bang for your buck, or you're building a premium system and want the best RAM for gaming available, our index gives you the data you need to make the right purchase.

You can also check out our list of the best RAM and our RAM benchmark hierarchy to see evergreen performance data to help you make an informed decision when choosing a new GPU for your system.

Lowest DDR5 Memory Kit Prices

DDR5 memory prices have exploded over the last trimester, with triple and, in some cases, quadruple increases. For instance, a conventional 32GB (2x16GB) DDR5 memory kit sold for around $100 to $200 in October 2025, but the same kit now starts at $350, if it's even in stock.

Retailers are taking advantage of the DRAM shortage, just as they did with the graphics card shortage a few years ago. They've started raising DDR5 prices to align with current market conditions and sometimes bundle DDR5 memory kits with processor or motherboard purchases.

Nonetheless, these DDR5 memory kits sometimes sell for more reasonable prices. However, you have to be quick to pull the trigger as scalpers and automated shopping bots tend to gobble up what little stock is left.

Lowest DDR4 Memory Kit Prices

The rising prices of DDR5 memory kits have forced some consumers to stick with DDR4. However, DDR4 has also seen price hikes over the last three months, though not as drastic as DDR5. We've seen the pricing double to triple.

For reference, it was common to find 32GB (2x16GB) DDR4 memory kits selling for between $60 to $90 back in October 2025. Jump to January 2026, and the same memory kit could cost anywhere between $150 and $180, so DDR4 isn't exactly a safe haven for consumers.

We've seen more reasonably priced DDR4 memory kits on the market than DDR5. Logically, it makes little sense to build a DDR4-based system now due to processor and motherboard availability. However, if you're in desperate need of a DDR4 memory upgrade, there are viable options on the market.

More Tech Deals

Best Tech and PC deals | Best gaming PC deals | Best RAM combo deals | Best 3D printer deals | Best RAM deals | Best gaming laptop deals | Best monitor deals | Best Wi-Fi Router deals | Best GPU deals | Best SSD deals | Best hard drive HDD deals | Best CPU deals | Best gaming chair deals | Best PC building tool deals | Best PC peripherals deals | Best filament and resin deals | Best motherboard deals | Best CPU cooler deals | Best PC case deals | Best Dell and Alienware deals | Best USB charger deals | Best gaming and productivity laptop deals under $1,000 | Best laptop PC deals

Also, you can join the Tom's Hardware deals Discord for up-to-the-minute hardware deals.

Now might be one of the worst times ever to build a new PC. Not only are GPU prices climbing, particularly at the high-end, but the RAM and NAND pricing apocalypse has pushed memory and storage — two components that are traditionally among the cheapest in many builds — to new highs. That makes building a new PC from scratch with DDR5 particularly tough right now. Upgrading, though, is a different story. If you have an old kit of DDR4 kicking around, and maybe an SSD, you can make a substantial upgrade to your PC without shelling out for insanely priced hardware right now.

We’ve put together two builds based on the latest chipset and processors that support DDR4, from both AMD and Intel. Intel certainly has the edge, as it supported DDR4 all the way up to its last-gen Raptor Lake Refresh chips. AMD stopped supporting DDR4 with Zen 3, and if it weren’t for some well-timed refreshes a couple of years back, you probably wouldn’t be able to find Zen 3 chips at all. If you have RAM and a decent CPU already, now is the time to buy a new video card, as prices are expected to further increase over the coming months. A modern Nvidia RTX 5060 Ti 16 GB balances budget and performance at 1080p, and even into 1440p.

The builds here are complete, short of an operating system (which you can get for free or cheap), but pick and choose components based on what you have to carry over from a previous build. If you have an older AM4 motherboard, you might be able to use it with a Ryzen 7 5800XT. Just make sure to check for compatibility with your motherboard manufacturer. Especially as we ride out the surge in RAM pricing, piecemeal upgrades are your best bet if you’re trying to take the value of your PC dollar the furthest. You could even buy 32GB today and sell your 16GB kit to offset costs.

If you’re coming from a really old build, you may not have a choice but to buy from scratch, and we have you covered there, too. For under $1,300, you can build a well-performing machine that not only plays games well at 1080p (or even 1440), but also serves as a solid productivity PC. In fact, you won’t be able to reach these performance levels until our $2,000 ($2,400 with RAM factored in) best PC build. Compared to DDR5, you’re saving at least $200 by using (or sticking with) DDR4.

Unfortunately, DDR4 builds are effectively a dead end. AMD moved past AM4 and DDR4 a few years back, and Intel abandoned DDR4 as it sunset the LGA1700 socket. However, these builds set you up for a full platform upgrade in the future. You might not be able to drop in a new CPU, but you’ll at least have all the other components you need once RAM pricing is under control.

DDR4 Intel Build

Intel had some fortunate (if unintentional) foresight by supporting both DDR4 and DDR5 on its 12th through 14th-gen CPUs. The Core i5-14600KF we chose for this build still remains an excellent processor. It’s less than 6% behind the Core i7-14700K on average in games at 1080p, and about 7% behind the newer Core Ultra 5 245K in multi-threaded performance. DDR4 represents an additional performance loss, though only around 5% (or less) in most cases. It depends on the workload.

We paired the Core i5-14600KF with a B760 chipset, which should (rightfully) send up alarm bells for regular Tom’s Hardware readers. This is an unlocked SKU, and Intel still doesn’t support CPU overclocking on its B-series chipsets. The Core i5-14600KF is the best deal we could find right now, though. The Core i5-14600 isn’t available, while the base Core i5-14600K sells for around $290. At $230, the Core i5-14600KF is the clear pick for this build.

Although a Z-series chipset is ideal to give you the full performance of the chip, there aren’t a lot of options available. Most Z-series motherboards still available at reasonable prices only support DDR5. For the few Z-series motherboards that have DDR4 support, you can expect to spend north of $250.

Along with the CPU, we chose the Thermalright Peerless Assassin 120, both because of how inexpensive it is and because it’s one of the best CPU coolers. Raptor Lake Refresh is known for getting hot, but that’s less of an issue for a chip like the Core i5-14600KF.

For the GPU, we’re going with the Gigabyte Windforce OC RTX 5060 Ti 16GB. This card has more than enough grunt for 1080p, and it can scale up to 1440p with its 16GB frame buffer and a little help from DLSS 4 (and even the newer DLSS 4.5). If you can spend around $150 more, the RTX 5070 is available for true 1440p sweet spot, or you can go with the AMD RX 9070 XT, which is currently our top pick for the best graphics card.

Rounding out the build is an inexpensive Montech XR midtower that posted solid thermal performance in our Montech XR Wood review (this version is $10 cheaper and ditches the wood but is otherwise identical). We also chose the Crucial P310 for its solid pricing and performance, along with the ASRock Steel Legend SL-850G, mainly due to the price and because it’s fully modular.

DDR4 AMD Build

The full build at Newegg when we wrote this was $1,262.84. For the money, you get the latest-gen AM4 processor, in this case a Ryzen 7 5800XT with an 8c/16t configuration and a max boost of 4.8 GHz. Priced at $219.99, the CPU isn’t a great deal (it sold as low as $124.99 previously), as people are already looking for DDR4/AM4 upgrades or new systems. Worse yet, for gamers, AM4 X3D options are either sold out at popular e-tailers, or third-party vendors are charging significantly higher prices. If you want a 5800X3D today, a third-party seller on Newegg has you covered…for over $600 (MSRP at launch was $450). Yikes.

We paired the Ryzen 7 5800XT with the budget-friendly Asus TUF Gaming B550-Plus Wi-Fi ($119.99). This motherboard supports 3000- and 5000-series processors (including 4/5000 G series), offers PCIe 4.0, dual M.2 sockets (one PCIe 4.0, one PCIe 3.0), and six SATA ports. Connectivity includes 2.5 GbE, Wi-Fi 6E, and eight rear USB ports, including a 10 Gbps Type-C. It utilizes the Realtek ALX897 audio solution and has robust power delivery for compatible CPUs.

We also chose Gigabyte’s RTX 5060 Ti Windforce OC 16G for this budget AM4 build (or upgrade) for its ample VRAM, important for the future, and solid 1080p performance across a wide variety of games, even AAA titles. The dual-fan solution should keep things running cool and quiet, and its all-black look blends in with most build themes.

For RAM, we stuck with the same 32GB from the Intel build, so you can have the best experience possible. We chose Patriots’ Viper Steel DDR4-3200 (2x16GB CL16) for the capacity, tight timings, and ‘low’ price of $172.99. You can go faster, but this is a budget build, and you generally wouldn’t notice the difference. Save the money here for spending elsewhere.

Storage-wise, Patriot’s Viper VP4300 1TB PCIe 4x4 fits the bill at $139.99. The drive uses DRAM cache for quick bursts, and lists speeds up to 7,400/6,100 MB/s R/W for longer transfers, which is plenty fast for loading your games and applications quickly. You can use SATA-based SSDs or hard drives for additional storage at a lower price per TB if necessary.

The machine is powered by an ASRock Steel Legend SL-850G 80Plus Gold (Cybenetics Platinum) 850W power supply. This mid-range unit supports ATX 3.1 and PCIe 5.1 standards, including a native 600W 12V-2x6 cable. At under $90, this 'future-proof' PSU is backed by a 10-year warranty.

We selected the Montech X3 Mesh (black) case ($69.90), which includes six preinstalled, fixed-lighting RGB fans for excellent airflow. The chassis supports ITX to E-ATX motherboards, multiple drives, PSUs up to 160mm, coolers up to 160mm, and GPUs up to 305mm. A side-swivel tempered glass panel provides easy access and displays the internals. For under $70, it’s a good-looking RGB case with ample space and cooling for high-performance components.

Conclusion

As you can see, building a new gaming PC today using DDR4 can get you plenty of performance for the money – and all the more so if you have RAM you can carry over from a previous system. Intel’s i5-1400KF and AMD’s 5800XT are formidable processors by nearly all measures. They’ll game and do well with most highly threaded work, too. Nvidia’s RTX 5060Ti 16G is arguably one of the better price-to-performance-to-VRAM ratio cards around, doing well at 1080p, and even stretching its legs to 1440p if you choose.

While these systems aren’t the pinnacle of DDR4-based PCs, they are full, well-rounded builds from Intel and AMD. Swap in your existing RAM and/or storage where you can, to help you get even more out of your build budget.

AMD and Intel have both moved past DDR4 at this point, so some element of building a DDR4-based PC is based on what’s available, not necessarily what’s objectively best. With that in mind, also check your local Micro Center and Best Buy, particularly when it comes to CPUs and motherboards. You might be able to score a deal if you shop at a local retailer.

A major Japanese PC and electronics store is pleading with customers to sell their old PC gear. “As a favor, if you buy a new one, please sell your gaming PC to our company,” begged the X-account of Sofmap Gaming in Akihabara, the Electric Town district of Tokyo (machine translation, h/t PC-Watch). The store shared a photo of some almost barren shelves, presumably taken at its triple-floor retail establishment.

“Gaming PCs, even used ones, are really out of stock right now,” wrote Sofmap, as an explanation for its call for old rigs. In the above Tweet, it asks customers to come in and sell their old PCs, highlighting that “We buy them back at pretty high prices...”

Moreover, the company underlined that it wasn’t going to be fussy. “Whether it's a gaming desktop or a laptop, or even a regular non-gaming one, we pretty much buy any PC...”

These are clearly the words of a PC retailer facing consumer demand that it just can’t meet. We reported on Akihabara store trying to limit new RAM, SSD, and HDD sales back in November.

Old becomes gold

The memory supply crunch impacted the PC industry faster and more deeply than many would have predicted. The insatiable demand for memory from AI data center makers, with their deep circular-funded pockets, caused the first pricing jolts in the PC memory market. That’s reasonable, as consumers and industry both need to be fed product from the same big-three memory makers.

Consumers saw the first impacts on modern DDR5 pricing. Some DDR5 kits, if you can find them in stock, like this Corsair Vengeance RGB DDR5-5200 16GB (2x8GB) on Amazon is now $235. That price is more than 3.5X what it cost last October ($66).

However, there remains some hope that DDR4 pricing and availability, thanks to old stocks and upgraders already having DIMMs, could provide a safe haven for continued PC building. This perception even seems to permeate PC component makers, with more DDR4-supporting motherboards being manufactured, plus hints about new processors for DDR4 platforms.

However, we are continuing to feel RAM crunch aftershocks. Prices of pre-built PCs were the next market affected. Graphics cards with more generous VRAM quotas are also strongly rumored to be facing constraints. We should at least expect a price rise for GPU-restocks, with next-gen GPUs rumored to be delayed…

Now, underlined by this Japan retail report, it even seems like stocks of old used PCs are being snapped up by consumers.

How old is too old?

Of course, some old PCs are too old for retailers like Sofmap, even during today’s PC drought. We’d expect retailers that dabble in used PCs for non-enthusiast users to limit their purchases to DDR4 platforms, with hardware support that slots above the Windows 11 minimum requirements (Intel 8^th Gen, AMD Ryzen 2000).

There’s an entirely different market for really old PCs, though. Vintage computers of certain eras have been increasingly pricey for quite a long time now. I was in Japan this time last year and astonished by the bountiful supplies of old PCs at used electronics retailers like Hard-Off. Hopefully, these computing gems (see the above picture), many of which live in the awkward zone between vintage and modern, will remain plentiful and affordable for PC retro-fans and tinkerers alike.

Memory prices are booming, pricing out many prospective PC builders in 2026. But, according to AMD's SVP & GM of the Client Business Unit, Rahul Tikoo, it's nothing to be worried about. During a roundtable interview with Tom’s Hardware at CES 2026 in Las Vegas, Nevada, we asked Tikoo how the company is responding to skyrocketing DRAM and storage prices, and how that might affect AMD going into 2026, as market intelligence firm IDC forecasts that PC shipments to fall by as much as 9% this year.

“We don’t see an issue there other than, you know, tightness leads to higher prices, eventually. So, from that perspective, I’m not seeing any impact to our business this year,” the executive told Tom’s Hardware. “I think that people who need the technology need the technology, and they’re going to buy the technology — though consumers might decide that they have a choice to make on how much memory, what CPU.” He also added, “Which is okay because we have a wide assortment of products available.”

AMD currently offers 215 SKUs, which are "AI PC" capable. There are also an equal or larger number of non-AI PC options available from the chipmaker. This means prospective PC buyers across a host of budgets and form factors may be able to find an AMD CPU to fit their build and budget. Old AM4 CPUs still in production have been topping Amazon’s sales charts, with another AMD executive even teasing the return of older Zen 3 processors.

Tikoo’s answer is logical — after all, AMD does not influence the memory chip supply chain, and they’re also seemingly not directly affected. Though the company sells Radeon GPUs, which are gaining ground on Nvidia’s RTX graphics cards, its market share is still small, sitting at just 7%. On the other hand, Nvidia dominates the GPU space, with a staggering 92% footprint. Therefore, it's likely that Nvidia is heavily affected, especially since Nvidia is reportedly no longer supplying VRAM to its GPU board partners. So, the only thing that it can really do is to offer a wider range of SKUs that utilize older DDR4 memory, which is currently slightly more affordable than DDR5.

Unfortunately, it offers little comfort to the average consumer — especially those who have already invested in a Zen 4 or Zen 5 chip, as the AM5 platform exclusively uses DDR5 memory. In line with what industry players are saying, you should only buy RAM kits or SSDs right now if you need them. If you don’t have an urgent requirement for more memory or storage, then we recommend you wait out the RAM-pocalypse before eyeing up a new upgrade, or find a solid bundle deal.

Although Tikoo doesn't see a major impact to AMD's business from DRAM shortages, the reality of higher pricing is hard to ignore. Estimates on when the DRAM shortage could ease range anywhere from six months to multiple years. As we begin 2026, we'll keep a close eye on memory pricing and where it's headed.

Samsung has reportedly delayed the shutdown of its DDR4 production line, which it initially announced earlier this year. According to DigiTimes [machine translated], the South Korean chipmaker is doing this as 16GB DDR4 modules hit a record of $60 in the spot market. This pricing made the DDR4 market lucrative enough for Samsung to delay the shutdown of its existing DDR4 production line. However, it will not set up new ones to accommodate increasing consumer demand. In line with this, reports indicate that one customer has already signed an “NCNR contract” with the company to secure its DDR4 supply.

NCNR (non-cancellable, non-returnable) contracts mean a client will receive a fixed number of memory modules at a fixed price, and that the price or volume cannot be modified in the future. In other words, the customer is contractually obliged to buy the product regardless of any other factors.

This type of agreement ensures stability in both volume and cost for the customer so that they won’t be affected by the current volatility in the memory market driven by the AI rush. On the other hand, this helps Samsung insulate itself from uncertainty in the memory market. Even if the AI bubble bursts and there is an oversupply of HBM and DDR5 memory modules, causing their prices to drop, this ensures that its DDR4 production line remains profitable.

Sources say the NCNR's price is expected to exceed $20 per 16GB DDR4 module, but Samsung is still evaluating the market and may increase it further. Still, this is only a third of prevailing spot market prices for short-term contracts. Unfortunately, the client said to be gunning for the NCNR contract mainly focuses on server applications, and no consumer-facing company has signaled its intent to secure a long-term contract for DDR4 chips. This means that end-users should not expect any relief from the ongoing memory shortage.

Samsung originally planned to end DDR4 production to focus on HBM and DDR5, but it seems that the booming AI demand for HBM has caused DDR5 memory pricing to skyrocket. This led to higher demand and prices for older DDR4 memory, causing the big three memory makers to delay their planned shutdown of DDR4 production lines. However, these are expected to serve industry clients, not consumers, so PC builders and enthusiasts should not expect any form of relief from the RAM shortage in 2026.

If you’ve missed out on Black Friday and you’re now looking for a deal so you can get a decent gaming PC for the holidays without breaking the bank, the ABS Flux II Aqua Gaming PC might be just the thing you’re looking for. It’s currently $999.99 on Newegg, giving you a 28% discount off its $1,399.99 regular price and saving you $400. This gaming PC comes with an Intel Core i5-14400F processor, an Asus Dual GeForce RTX 5060 graphics card, a 1TB Kingston SNV35 PCIe 4.0 x4 M.2 NVMe SSD, and, perhaps most importantly, 32GB (4 x 8GB) of TeamGroup T-Force Vulcan Z DDR4-3200 RAM.

• Check out this deal on Newegg

This might not be the latest or the best-of-the-best hardware available today, but it offers good enough performance for a reasonable price, especially now that memory and storage are in short supply (and pricey!). One of the things that we love about ABS is that it shows you the exact make and model of the parts that it’s going to put into your pre-built, thus avoiding any unpleasant surprises.

Aside from the components specified in the intro, you’re also getting an ASRock B760M-C/D4 motherboard, a Gamdias Boreas E1 air cooler, a Gamdias Flux II 650W 80+ Gold PSU, and a Gamdias Flux II PC case. We also checked out the individual prices of each component and found that you’d actually save more than $200 compared to buying each part and building this system yourself.

You will miss out on DDR5 when you go for this pre-built PC, but you’re also saving a lot of money with the way memory prices are going these days. The Intel i5-14400F also falls behind comparable AMD processors, but the rush to build gaming PCs with older (and cheaper) memory modules has gotten to the point that AM4 X3D processors are now more expensive than brand-new AMD Ryzen 7 9800X3D chips because of demand skewed by the memory market. As for its graphics card, the RTX 5060 is Tom’s Hardware’s chosen option for the best GPU at the $300 price range, giving you a lot of bang for your buck.

Even though this new gaming PC does not have cutting-edge hardware, it gets its points by being a value king, especially in this day and age. Sure, you might be disappointed because you can’t play 4K ray tracing titles on this gaming PC, but your enjoyment shouldn’t be diminished as it can still play some of the latest titles at 1080p with decent quality. So, if you really need a new gaming PC this holiday season, don’t wait and get this affordable gaming PC from Newegg today.

If you're looking for more savings, check out our Best PC Hardware deals for a range of products, or dive deeper into our specialized SSD and Storage Deals, Hard Drive Deals, Gaming Monitor Deals, Graphics Card Deals, Gaming Chair, Best Wi-Fi Routers, Best Motherboard, or CPU Deals pages.

The RAM-pocalypse is still very much in full swing, with RAM prices continuing to rise daily, and supplies running low, as cheaper kits get snapped up by either enthusiasts or scalpers. There are already lots of eye-watering prices on listings from retailers like Amazon and Newegg, and eBay brings up some even more ludicrous entries. We've covered quite a few bundled deals since before Black Friday, and that continues to be the trend for finding some of the best deals on RAM. Much like today's combo deal from Newegg that pairs an Asus Tuf Gaming B550 Plus Wifi II motherboard with 32GB (2x16GB) of G.Skill Ripjaws V Series DDR4-3200 memory for $199.98.

• Grab this deal at Newegg

With today's combo deal, you're saving $110, rather than buying them both separately. On their own, the Asus motherboard is listed at $119.99 on sale from $149.99, and the G.Kill Ripjaws RAM kit is $189.99. But, as a bonus, you're also getting the free gift of a Cooler Master Hyper 212 Spectrum V3 CPU air cooler worth $25.85 (until stocks last, or the offer expires). This is a great start for a budget gaming PC build, and you could pair this with a 5000-series AMD Ryzen CPU, and toss in a GPU to make up the core of a new build.

Although AM4 is an older platform that has been around for many years, it's still very capable and will run pretty much all modern games with a decent graphics card. CPUs like the 5800X3D and 5700X3D are still extremely potent, and if you can't get your hands on one of the X3D models, then a normal 5700X/5800X also does a very good job. Check our CPU hierarchy and look at reviews for 5000-series CPUs to pair with this bundle.

G.Skill's Ripjaws V-Series memory is DDR4 3200MHz (PC4 25600) with timings of 16-18-18-38, and a CAS Latency of CL16, with a voltage of 1.35V. The Asus Tuf Gaming Plus WiFi II motherboard has an AM4 socket and supports 3000, 4000, and 5000-series CPUs. The board features dual M.2 slots, the primary one with PCIe 4.0 x4 connectivity. There is a front panel USB 3.2 Gen 1 connection, USB 3.2 Gen 2 Type-C and Type-A ports on the IO plate, and also a Thunderbolt 3 header. Connect to the internet via WiFi 6 (802.11ax), or fast 2.5Gb LAN port. This memory bundle might not package up the latest and greatest parts, but it's a solid offer, given what we've seen out there on the market currently.

Best gaming laptop deals | Best monitor deals | Best PC and laptop deals | Best SSD deals | Best CPU deals | Gaming Chair | Best hard drive deals | Best PC case deals | Best Dell and Alienware deals | Best 3D printer deals | Best PC peripherals deals | Best motherboard deals | Best CPU cooler deals | Best gaming chair deals

The ongoing structural change of the DRAM market caused by the shift of manufacturing capacities to production of high bandwidth memory (HBM) for AI accelerators has already caused a massive price hike of commodity DDR and LPDDR memory — but the worst is yet to come.

According to the general manager of Chinese memory giant TeamGroup, contract prices of DRAM and NAND products have almost doubled recently. Supply of commodity memory is set to worsen in early 2026, and normalization is unlikely before 2027 – 2028 when more production capacity emerges, reports DigiTimes.

December contract prices of some categories of DRAM and 3D NAND increased 80% to 100% month-on-month, according to Gerry Chen, general manager of TeamGroup, a prominent maker of memory modules, solid-state drives, and products based on 3D NAND. Spot prices tell a similar story. A 16Gb DDR5 chip was priced at $6.84 on average at DRAMeXchange on September 20. On November 19 average spot price was $24.83, but on December 1 average spot price of a 16 Gb DDR5 IC increased to $27.2 (session low was $19, session high was $37).

Essentially, memory alone for a 16 GB memory module costs around $217.6. A PCB, assembly, and testing, additional parts like PMIC will add $8 – $10, so a 16 GB memory module now costs $225 – $228 without manufacturer premiums, logistics, and taxes.

Chen expects availability of DRAM and NAND to worsen in the first and second quarters of 2026 once distribution stockpiles are exhausted. At that point, he cautions, obtaining allocation could become difficult regardless of willingness to pay. In his view, relief would not come quickly: he projects the current shortages to extend into late 2027 and potentially beyond.

The reason for shortages of commodity memory is well known: DRAM makers reallocate their production capacities to HBM (which uses larger DRAM dies than commodity types of memory) that is consumed by AI accelerators, like Nvidia's B300 or custom accelerators by large cloud service providers, such as AWS, Google, and Microsoft. These companies tend to book supply years in advance, so at some point, DRAM makers will not have enough capacity to meet demand for commodity DRAMs.

Building a new greenfield fab takes at least three years, so even if companies like Micron, Samsung, or SK hynix made a decision to build a memory fab today, it would come online in late 2028 at the earliest and would be fully ramped only sometime in 2029.

When it comes to NAND, NAND suppliers also prioritize large customers, which happen to be makers of AI servers. Chen does not expect capacity to swing back to PCs, smartphones, and other consumer devices in 2026, which will affect the prices of these devices.

The effects are clear to see. Enthusiasts are seeing RAM prices for custom-built PCs increase by orders of magnitude week on week, with 64GB of DDR5 RAM now costing more than a PS5 in some cases. This week's Black Friday and Cyber Monday RAM deals might be the last chance to buy RAM before prices skyrocket even further.

The DRAM market is flashing mixed signals as 2025 enters its final stretch. According to findings published by Digitimes, DDR4 spot prices have stabilized after a frenzied second quarter, yet contract pricing continues to rise as supply remains tight.

DDR4 rebounds, DDR5 weakens

This has created a paradox for PC builders, with older DDR4 kits now coming in more expensive than their newer, faster DDR5 counterparts.

DRAM spot market industry data compiled by Digitimes shows that DDR4 16GB (2Gx8) parts increased by nearly 7% in August after a slow July, pushing prices to $9.17 per chip. DDR5 (2Gx8) 16GB slid 3% to $5.99, widening the gap between generations to almost 50%. DDR4 8GB (1Gx8) continued its decline, while DDR3 (512Mx8) surged again, extending July’s 20% rally with a further 13% jump as Samsung’s 2024 production exit and limited output from Taiwanese firms triggered panic buying.

This might look like a simple supply squeeze on the surface, but a closer look reveals that top suppliers like Samsung, SK hynix, and Micron are all winding down DDR4 production to free up space for DDR5 and high-bandwidth memory (HBM), both of which serve the booming AI accelerator market. This pivot has left DDR4 users footing the bill for engineered scarcity.

An AI tax on legacy hardware

This pricing surge is more about supply being deliberately throttled than demand suddenly spiking. Larger fabs are prioritizing HBM, which is far more profitable due to its applications in GPUs and AI inference.

Every wafer shifted to feed Nvidia’s H100s or AMD’s Instinct accelerators cuts into supply for DDR4. That deliberate reallocation throttles supply, which in turn forces system builders and distributors to stockpile what’s left.

The result is a market inversion whereby DDR4 now costs comparatively more than DDR5, despite being an older, slower legacy technology. In effect, mainstream builders are subsidizing the AI boom, shouldering higher costs on legacy parts because memory makers know they can extract maximum profit before pulling the plug.

Contract prices still heading higher

DDR4 memory is outdated, but there are plenty of PCs with this type of SDRAM — some of which were built as recently as 2023 — waiting for an upgrade. Due to reports about the end of DDR4 production at leading DRAM makers, pricing skyrocketed in the global market.

That leaves owners of DDR4 systems wondering whether they should upgrade their rigs now, as it was due to vanish from shelves soon. Thankfully, it appears that DDR4's lifespan has been extended, according to multiple industry sources.

The transition to a new type of consumer PC memory usually takes around seven to eight years. The decline of the previous generation DRAM standard usually happens within two to three years of the new standard's release.

Cut-throat competition between memory manufacturers means that they are keen to wind down production of previous generation products, to cut down costs. It also means that companies don't have to use aging technology with the latest nodes. Therefore, there will be no DDR4 and LPDDR4 devices made on 6th Generation 10nm-class nodes, such as 1c/1γ (1-gamma).

As a result, while there are plenty of entry-level platforms that support DDR4 today — including rather decent 13/14 Generation Core or Core 2-series based on the Raptor Lake silicon or all-new Bartlett Lake silicon — and can take advantage of lower-cost memory, they will not have a lot of time to do so.

Samsung and SK hynix will maintain production of DDR4 memory until the end of 2025 and Q1–Q2 2026, respectively. So DDR4 products are not going to vanish overnight, and will remain widely available well into 2026. After that period, it's safe to assume that DDR4 will become niche memory for applications with a very long lifecycle. Even after that, smaller players will keep producing DDR4 for those who need it.

Micron to ship final DDR4 chips in early 2026

Micron issued end-of-life notices for DDR4 and LPDDR4 products several months ago to customers in multiple high-volume segments, like consumer PCs, data centers, and mobile devices. Micron notified them that final shipments of mainstream DDR4 and LPDDR4 devices will occur within the next two to three quarters. Or in other words, early 2026.

"Micron's leading-edge DRAM nodes such as 1ß and 1γ are focused on the latest-generation products such as DDR5, LPDDR5, and HBM, and are not utilized to produce DDR4 and LPDDR4," said Sanjay Mehrotra, chief executive of Micron, during the company's latest conference call. "DDR4 and LPDDR4 products are largely produced in our 1α (1-alpha) DRAM node."

As with previous memory transitions, Micron will continue to support long-term, lower-volume customers in the automotive, defense, industrial, and telecom sectors, with 1α DRAM for several more years, at appropriate prices. As a result, expect shortages of DDR4 and LPDDR4 memory from Micron to occur in the coming quarters. But the good news is that there are plenty of DDR4 memory modules available now from Micron's Crucial brand.

Samsung to sustain DDR4 production until December 2025

Samsung uses its 3^rd Generation 10nm-class (1z) process technology to produce its inexpensive DDR4 devices. Due to strong demand and increasing prices of DDR4, the company is postponing the shutdown of its DDR4 1z DRAM production lines until the end of 2025, according to TheElec.

Samsung initially informed clients that it would end DDR4 1z DRAM production by late 2025, the report claims. However, to meet demand for previous-generation memory and grab some extra cash, the company decided to keep its 1z DRAM manufacturing line running for several more weeks, which equals millions of dollars. These lines are already fully depreciated, so the cost of keeping them up is low.

Normally, Samsung needs to install new production lines as soon as possible to meet demand for more advanced and popular products. However, just like Micron, Samsung is obliged to support its long-term clients from the automotive, industrial, and telecommunications sectors with low volumes of DDR4 for years to come. So mass-producing DDR4 for just a little longer makes sense for the company. Samsung's revised plan may also provide short-term relief for module makers and device manufacturers that do not have a long-term supply contract with the company.

Samsung is poised to sell DDR4 chips for several months (if not quarters) into 2026, assuming that the final mass-produced DDR4 device will be made in December 2025.

SK hynix to keep DDR4 production till Q2

SK hynix's transition away from DDR4 is expected to proceed as planned. According to a source with knowledge of the matter speaking to Tom's Hardware, SK hynix's production is winding down between the first and second quarters of 2026. This will likely make SK hynix the last high-volume producer of DDR4 memory. However, it remains to be seen whether it will prolong volume production.

Of course, the company will also keep supporting its clients with long-term contracts and niche applications with DDR4 chips for years to come. But, by 2027, it'll be hard to come by unbuffered DDR4 memory modules made by SK hynix.

The outlook on DDR4

For the first time in history, last-gen's DRAM has soared to over double the price of the current generation. The price of DDR4 modules is rapidly outpacing that of DDR5 modules, thanks to a perfect storm of tariff fears and stock uncertainty.

According to TrendForce, some DDR4 kits and configurations have increased by up to 40% in the last week alone, rapidly widening the gap between DDR4 and DDR5 prices. The spot price for DDR4 16Gb (1Gx16) at 3200 MHz from Samsung/SK hynix grew to an average price of $12.50 via DRAMeXchange, with highs reaching $24.00. Comparatively, the price for DDR5 2x8G modules at 4800/5600 MHz has stuck close to $6, with daily highs only hitting $9.

Never before have we seen the outgoing generation of DDR memory chips surge this high on their way out. The rapid growth spurt for DDR4's trade value comes shortly after Micron announced its exit from producing new DDR4 memory modules this year. Micron's DDR4 lines are entering end of life, with the firm to wind down production over the next 6-9 months.

The firm is one of the final dominoes for DDR4 to fall, as a few memory producers have priced most of the major producers out of building more DDR4 chips. Micron joins Samsung, which announced its imminent DDR4 retirement in April, citing a need to focus on incredible demands for DDR5 and HBM production. Chinese firm CXMT also teased the end of its DDR4 production, even though the firm just reached peak DDR4 production.

Taiwanese DRAM firm Nanya stands to win big in this topsy-turvy DDR4 market. The firm is one of the largest suppliers of DDR4, currently. Recent reports found that in Q1, Nanya's DDR4 inventory was worth a staggering NT$37.6 billion ($1.2 billion USD), and while this was initially thought to be a major money pit for the company, recent upturns in prices have left Nanya standing on top. The firm recently stopped DDR4 spot price quotes as prices began to climb, leaving them ready to sell high.

Samsung, SK hynix, and Micron are now reportedly raising their prices to their customers for their last-ever waves of DDR4, so it's likely this will cause the bubble to continue growing. On top of this, tech industry tariff concerns have left things uncertain enough to spark some panic buying.

The U.S. government's trade interventionism with China is well-documented, and many speculators believe the current administration will see China's huge existing output of DDR4, ready to capitalize on the price hikes, and level retaliatory tariffs to cinch supply from the nation. This would send DDR4 prices even higher — likely to more than triple the cost of DDR5.

In a market dominated by DDR5 and HBM memory demand, it's easy to forget that the average consumer is not chasing the newest memory in quite the same way. DDR5 adoption accounts for only 60% of new notebook models, with many laptop buyers happy to stick with the older generations because they (used to) cost less, and the technology is more mature (more reliable). This means many laptop OEMs are still looking to source DDR4 memory to pair with, for example, Intel's older Raptor Lake chips, which are currently outselling newer generations in an embarrassing fashion. Unfortunately, it looks like this money-saving hack may be ending, as DDR4 laptops may soon cost more.

DDR4 memory prices have seen a “rapid spike” reports Taiwanese tech industry journal DigiTimes. It asserts that, in the second half of May alone, DDR4 spot market prices have soared by around 50%. Several factors are said to be at play here, exerting inflationary pressure on DDR4 prices.

You can see in our table, above, the rather extreme price action affecting the DDR4 spot market in recent weeks. DigiTimes also says that the contract prices paid by tech manufacturers have also skyrocketed by between 22–25% for 8 GB and 16 GB chips, compared to prices negotiated in early May. During Q3, it is expected that DDR4 prices will continue to rise by an estimated 10 to 20%.

What is behind the sudden DDR4 price increases?

One well-documented memory industry trend that is behind the price increases seen is said to be makers shutting down their DDR4 production in favor of DDR5 and other more profitable lines. In February, we noted that the likes of Micron, Samsung, and SK hynix were being rudely elbowed out of the DDR4 market by Chinese players (such as CXMT and Fujian Jinhua) ruthlessly undercutting them in this segment.

Samsung was seen to flinch in late April, as reports circulated that the South Korean technology and manufacturing giant had scheduled to cease DDR4 production in early June.

Now there are indications that oversupply from Chinese ‘dumping’ is at an end, as CXMT has been instructed by the Chinese government to abandon DDR4 manufacturing. Thus, the reported spikes in DDR4 pricing in recent weeks may stem from a perfect storm of the above supply-side factors all exerting an effect over a relatively short period of time.

DigiTimes doesn’t lay the DDR4 pricing spiking blame entirely at the door of cuts in supplies and the anticipation of spot market traders. It also cites upward pricing pressure coming from US-China tensions and other market uncertainties.

The source doesn’t go into much detail about why US-China tensions have contributed to the recent DDR4 memory spike. However, we reported on US tariff-inspired stockpiling of DDR4 in late May, causing a significant price surge. Additionally, we note that CXMT’s move to shadow South Korea’s lead in shifting away from DDR4 will have been deemed necessary for China’s increasingly self-sufficient semiconductor industry to keep up with tech trends.

DDR4 still required by some industrial and embedded platforms

Leading DRAM makers — Micron, Samsung, and SK hynix — may cease production of DDR3 and DDR4 memory by the end of the year. The decision is due to low prices caused by dumping (by Chinese makers) and declining demand, reports DigiTimes. This may potentially cause shortages of DDR4 memory that is still used in entry-level PCs and consumer electronics.

Chinese DRAM manufacturers Changxin Memory Technology (CXMT) and Fujian Jinhua have been ramping up production of DDR4 while aggressively lowering prices. Late last year their DDR4 chips were priced at half the cost of similar products from South Korean competitors. In some cases, Chinese DDR4 memory ICs were even 5% cheaper than refurbished chips. As a result, leading DRAM producers may no longer earn profit by selling DDR4, which is why they may have to phase out this type of memory while focusing on more profitable DDR5 (well, until Chinese companies start to produce it in volumes and dump on the global market) and HBM.

However, there is a catch, as demand for DDR4 is quite strong, and Chinese makers may not meet it. Thus, as soon as Micron, Samsung, and SK hynix cease to make DDR3 and DDR4 memory for the spot market, DDR4 supply constraints could emerge after mid-2025, according to DigiTimes analysts.

Taiwanese producers may step in to bridge the gap, but both Nanya Technology and Winbond Electronics only produce specialized types of DRAM, which means relatively low volumes and high prices. Yet, some industrial customers are unlikely to adopt China-made DRAMs and are more likely to adopt specialized memory from Nanya and Winbond instead.

Nanya Technology predicts the market will reach its lowest point in early 2025, followed by a recovery in the second quarter. This rebound will be fueled by stronger demand, better stock management, and economic stimulus. AI-related cloud computing is expected to remain a major driver, while general consumer demand is projected to see only a slight improvement.

Winbond Electronics, facing weakening demand for older DDR versions, is moving to a more advanced 16nm process in late 2025. This upgrade from 20nm will allow the company to produce higher-capacity 8Gb DDR chips.

Chinese DRAM makers Changxin Memory (CXMT) and Fujian Jinhua are aggressively expanding production and cutting prices, reports DigiTimes. Right now, the two manufacturers sell their DDR4 components with a 50% discount compared to similar ICs made by South Korean makers and in some cases, these DRAMs are even cheaper than reballed (recycled/re-used) memory chips.

CXMT's production capacity has surged from 70,000 wafers per month (WPM) in 2022 to 200,000 WPM in 2024, the report says. The company aims to increase its output to 300,000 wafers per month and capture 11% of the global DRAM market in the coming years. Fujian Jinhua, despite being sanctioned by the U.S. government, has also managed to ramp up its DDR4 production.

The oversupply of DDR4 chips has resulted in sharp price competition, with Chinese manufacturers offering up to 50% lower prices than Micron, Samsung, and SK hynix and 5% below even second-hand reballed memory chips. This aggressive pricing is pushing prices in the DDR4 market down, particularly impacting the consumer segment. Yet, industrial clients remain cautious with regard to adopting Chinese DRAM.

Reacting to this intense price pressure, Korean DRAM firms are reducing their DDR4 production and shifting to DDR5 and HBM3 memory - ICs that Chinese vendors cannot mass produce right now. This move is aimed at maintaining profitability and countering the cost-competitive advantage Chinese manufacturers currently hold in the DDR4 segment. This shift to DDR5 and HBM3 is expected to stabilize the DRAM market by addressing oversupply and redirecting resources toward more advanced, high-demand memory technologies.

The Chinese government has actively supported this DRAM expansion through subsidies and national policies, fostering a domestic production surge. This support enables Chinese manufacturers to offer highly competitive prices, with little concern for immediate profitability, as they focus on establishing dominance in both local and international markets.

Chinese DRAM manufacturers are also exploring indirect entry strategies into regions like India, where they hope to leverage partnerships with Taiwanese brands. This approach is intended to mitigate political and economic risks associated with the 'Red Supply Chain' and to expand their footprint in markets wary of direct Chinese products.

Changxin Memory recently faced a production setback due to a human error that resulted in 65,000 wafers being scrapped, impacting its ability to meet delivery schedules. The error led to penalties for several executives. Nonetheless, Changxin’s long-term expansion remains on track, with capacity growth and ongoing upgrades at both its Hefei and Beijing fabs.

Memory makers have halted disclosing contract prices of DRAM in Q2 2024, following a major earthquake in Taiwan, which may point to potential price hikes. The earthquake disrupted wafer fab operations, impacting supply as Micron, Samsung Electronics, and SK Hynix are among those affected, reports DigiTimes. This situation may lead to sharper price increases than anticipated, as the industry was already aiming for higher prices before the quake, in a bid to recoup losses memory makers incurred in the previous years.  

Micron ceased offering product quotes for the second quarter, shortly after the earthquake, with its Korean rivals quickly following suit. Industry insiders revealed that these leading memory chip manufacturers had already initiated measures to maintain a rising trend in their product prices before the April 3 quake. Now, it seems highly probable that the price hikes may be more pronounced than previously expected.  

In the past, memory prices have typically surged following production disruptions caused by natural calamities or accidents. The latest incident could provide the vendors with chances to compensate for the losses incurred in 2022 and 2023. Prior to the quake, second-quarter DRAM and NAND flash contract prices were experiencing slower growth compared to the first quarter. Specifically, NAND flash prices had risen by 15% - 20% for the second quarter, as opposed to over 20% in the first quarter. 

Following the earthquake, Micron announced that it would evaluate the impacts on its local production and supply chain. The company stated that it would then discuss supply schedules with its customers, hinting at potential changes in supply. Industry sources mentioned that memory fabs typically halt production during such strong earthquakes, which affects supply of memory wafers. 

Memory module manufacturers, who are reportedly low on chip inventory, are now preparing for increased chip purchase costs. They had hoped that the rising price trends would decelerate, given the rising capacity utilization rates at chip suppliers. However, chip suppliers had significantly reduced production to sustain pricing last year, when demand was stagnant. Downstream customers, including server makers, might become aggressive in stocking memory products in anticipation of shortages, further driving up prices. 

Although contract prices might be soaring, the spot market remains weak due to low demand from the consumer sector following the Lunar New Year. Apparently, there is still an abundant supply in the spot market, despite the disruptions and anticipated price hikes in the contract market, the report claims.

Consumers should be wary that they're getting DDR4 DRAM that's more smoke and mirrors than an actual straight-from-the-manufacturing-line product. As reported by TrendForce, the worldwide DRAM spot market is being hit by a wave of used DDR4-3200 MT/s memory chips that are being desoldered from old server RAM sticks, re-flashed, and then put to use in new, for-sale consumer RAM kits and devices.

Besides putting additional downward pricing pressure on a market that has almost imploded compared to its historic levels, there's also a justified concern that products packing these chips might fail earlier than expected.

Usually, there's not enough of a price difference between used and new memory chips that justify going to the trouble of actually separating (desoldering) all the individual DRAM chips from a used stick of memory. But for DDR4 memory, the usual economics don't apply; there's enough of a difference in performance between server CPU platforms offering either DDR4 or DDR5 support (remember that Zen 4 and Intel's Sapphire Rapids are DDR5-exclusive architectures) that enough professional environments are making the jump. And all those decommissioned high-capacity DDR4 memory chips have to go somewhere.

Most of the identified recycled chips are said to use "two major South Korean suppliers' legacy processes,' heavily implying that these chips originally came from Samsung and SK hynix. 

TrendForce says there's an 'influx' of used chips flooding the market — enough to impact spot market pricing — so it's clear there is quite a volume of used chips finding their way onto the market. That means whoever is recycling the modules must have somehow automated the chip desoldering, extraction, reprogramming, and repackaging process, as many of these steps — particularly desoldering — would be arduous and time-consuming to do manually for large volumes of chips.

These companies are essentially exploring the delta between how low they can buy bulk DDR4 memory chips and DDR4's average spot market pricing. If their operational costs are lower than the difference, they can sell these used products at a price premium (compared to what they bought them for) that's still competitive with newly-manufactured DDR4 chips. And paying less so you can buy more is always an attractive business loop.

Some questions about product longevity are warranted, but we'd expect a limited warranty to still be put in place for chips acquired this way. It's likely that we're dealing with a temporary business arrangement - the companies' profits will start to decline the moment the used DRAM prices market increases. But while it's impressive that companies are finding ways to lower the recycling costs of chips low enough that they can engage in this niche business model, it isn't an attractive prospect to buy a new PC or DDR4 stick that actually uses old re-reprogrammed flash that might be a bit dodgy. 

The DRAM price drops seen in recent quarters are slowing, according to a new report by TrendForce. The computer industry market researchers say that, in the current quarter, we will only see DRAM prices decline in the region of 0-5%. This figure contrasts with Q2 2023 DRAM price decline of between 13-18%. These headline figures are for the DRAM market as a whole (including sever, mobile, graphics), but also hold true for the segment that interests us most - PC DRAM.

According to TrendForce, one of the reasons that DRAM price declines have slowed is that the big three manufacturers have been cutting their production. Such action may provide temporary relief for the producers, but attempting to artificially create scarcity, and running production much slower than it could be, isn't usually a sustainable strategy in a competitive market. We would rather DRAM makers spur demand by making higher capacity RAM configurations more affordable, for example.

Another fly in the ointment for DRAM producers is that TrendForce asserts there are high DRAM inventory levels. Holding excessive inventories is also not a solid long-term strategy. Any release of pressure from the production line or inventory by any of the big players could precipitate further big price declines. In the meantime, they may hold in hope of a new device or platform inspiring lots of DRAM purchases.

With the Q2 and Q3 figures, and market knowledge, TrendForce says it doesn't see the chance of a rebound in DRAM prices "until 2024."

Looking more closely at PC DRAM in particular, TrendForce says that there is still DDR4 oversupply - and in the current quarter (Q3) prices are expected to slide 3-8%. PC DDR5 will see a more modest 0-5% drop in Q3, says the market researcher. Thus we can see the influence of PC makers and DIYers shifting gradually to systems using the newer, faster memory standard, and perhaps even early adopters now splashing out on higher-capacity DDR5 kits.

Graphics DRAM is a similar story to PC DRAM's decelerating Q2 vs Q3 pricing declines. TrendForce notes that GDDR6 16Gb chip (2GB VRAM) demand is pretty high due to recent RTX 40 launches. Moreover, we are approaching the peak season for component makers to integrate VRAM, so graphics RAM might bottom out (and rise) before PC system memory.

In conclusion, these figures seem to indicate that the decline in PC RAM prices we have seen over recent quarters might be slowing. For some that will be a sign that it is a good time to buy - and this decision might be right for RAM upgraders, but there are so many other PC component prices to consider for a new full system.

If you are interested in PC RAM right now, it is worth a look at our frequently updated Best RAM for Gaming 2023 guide which lists our favorite DDR4 and DDR5 kits, includes specifications, mini-reviews, full review links, and buying links.

Our RAM benchmark hierarchy helps you find the fastest DDR5 and DDR4 memory kits for your system. People frequently underestimate RAM (Random Access Memory), yet it is among the most vital hardware components. Choosing the appropriate or inappropriate memory kit can significantly influence your system's performance in gaming, productivity, and, if insufficient, lead to overall sluggishness. Remember, consult our Best RAM list for the best memory deals.

Companies often introduce new memory kits with different speeds, timings, capacities, and ranks, making it tedious to compare numerous models. Our RAM benchmark hierarchy aims to provide a simple database categorizing the best memory kits we've reviewed by raw performance.

To maintain objectivity in our rankings, we utilize the geometric mean of our memory benchmarking results, which helps eliminate subjective factors such as aesthetics and overclocking potential. Detailed insights are available in individual RAM reviews. We believe many of you prioritize performance above all else. However, the geometric means alone may not provide a complete picture: we suggest consulting the full review of the memory kit you’re considering to optimize performance in specific workloads or gaming scenarios.

The score results are based on the geometric mean of our RAM benchmark suite, which includes both scripted and real-world tests. This suite features applications such as Adobe Photoshop, Adobe Premiere, and Adobe Lightroom, along with benchmarks like Cinebench 2024, Corona 12, Blender 4.0, V-Ray 6, 7-Zip for compression and decompression, HandBrake for x264 and x265 conversion, LuxMark v4, Windows AI inference, and Y-Cruncher.

For simplicity, we have organized the memory kits into distinct categories by density. Next, we ranked the memory kits by capacity for both Intel and AMD systems, from best to worst. The score in our hierarchy may differ slightly from the geometric mean presented in the individual review. This variation occurs as we strive to provide results based on the most recent and relevant Intel and AMD platforms. We retest each memory kit to ensure the metrics in the table are as up to date as possible. We conduct retests whenever there are significant updates to either of our testing systems, including new processors, motherboards, graphics cards, or firmware revisions.

Memory pricing is highly volatile. We have refrained from providing a price-to-performance ratio because it reflects pricing at a single, static moment, and memory pricing varies significantly. Nonetheless, we strive to provide the most current pricing in each table below; however, the performance measurements do not account for price.

If you're stuck between two specific kits, you can easily compare your options by dividing the real-time pricing by the geometric mean. Ideally, you'll want to aim for the lowest value, as a smaller figure indicates better value for your money.

Update June 15, 2026: The results are valid for comparisons of memory kits within the same category. Please avoid comparing memory kits of different capacities, as we're still updating the results for the latest Z890 and X870 platforms.

Best DDR5 RAM For Intel CPUs

Best 32GB (2x16GB) DDR5 Memory Kits for Intel CPUs

Best 48GB (2x24GB) DDR5 Memory Kits for Intel CPUs

Best 64GB (2x32GB) DDR5 Memory Kits for Intel CPUs

Best 96GB (2x48GB) DDR5 Memory Kits for Intel CPUs

Best 128GB (2x64GB) DDR5 Memory Kits for Intel CPUs

Best 192GB (4x48GB) DDR5 Memory Kits for Intel CPUs

Best DDR5 RAM For AMD CPUs

Best 32GB (2x16GB) DDR5 Memory Kits for AMD CPUs

Best 48GB (2x24GB) DDR5 Memory Kits for AMD CPUs

Best 64GB (2x32GB) DDR5 Memory Kits for AMD CPUs

Best 96GB (2x48GB) DDR5 Memory Kits for AMD CPUs

Best 128GB (2x64GB) DDR5 Memory Kits for AMD CPUs

Best 192GB (4x48GB) DDR5 Memory Kits for AMD CPUs

Best DDR4 RAM For Intel CPUs

Best 16GB (2x8GB) DDR4 Memory Kits for Intel CPUs

Best 32GB (2x16GB) DDR4 Memory Kits for Intel CPUs

Best 64GB (2x32GB) DDR4 Memory Kits for Intel CPUs

Best DDR4 RAM For AMD CPUs

Best 16GB (2x8GB) DDR4 Memory Kits for AMD CPUs

Best 32GB (2x16GB) DDR4 Memory Kits for AMD CPUs

Best 64GB (2x32GB) DDR4 Memory Kits for AMD CPUs

CPUs and Supported Memory Frequency

Once upon a time, the memory controller resided on the motherboard. As processors and motherboards evolved, chipmakers eventually moved the memory controller onto the chips. Therefore, Intel and AMD specify the official supported memory frequency on each generation of their processors. Unless you have an excellent sample, drifting far away from the supported frequency will require manual tweaking to get the memory to play nicely with your processor. For the best plug-and-play experience, you should pick up a memory kit that doesn't surpass the supported frequency.

There are two essential things to consider if you want to run memory faster than the official specification for your processor. For starters, the quality of your processor's IMC (integrated memory controller) determines if it can handle high-speed memory. Unfortunately, since no two processors are equal, and you can't judge a chip's IMC until you test it, you're at the mercy of the silicon lottery.

Next, verify that the motherboard is compatible with the frequency of your chosen memory kit. Quickly looking at your motherboard's user manual or product page will provide the needed information. Additionally, the motherboard's Qualified Vendors List (QVL) can be handy, as it indicates whether a particular memory kit has been tested and approved for use with the motherboard. However, remember that your chip's IMC remains a critical factor.

If you're unsure of the memory frequency your processor supports, we've provided the different standards for the last few generations of AMD and Intel processors.

AMD Desktop Processors

AMD has officially announced the Ryzen 9000 (Granite Ridge) and Ryzen 9000X3D processors powered by the chipmaker's latest Zen 5 execution cores. Similar to the company's previous Ryzen 7000 (Raphael) lineup, the latest Ryzen 9000 processors only support DDR5 memory. Native memory support on Zen 5 has improved, embracing DDR5-5600 right out of the box, a 4% improvement over Zen 4. The maximum memory supported on Ryzen 9000 processors is 128GB. We have been able to get up to 192GB working on the X870 platform, but your mileage may vary.

Zen 5 natively supports DDR5-5600 in a 1 DPC configuration regardless of whether the memory modules have a single- or dual-ranked design. However, in a 2-DPC structure, the official data rate drops to DDR5-3600.

AMD introduced EXPO (EXtended Profiles for Overclocking) profiles with Zen 4. AMD EXPO is equivalent to Intel's XMP 3.0 technology, allowing one-click overclocking to the predefined data rates. AMD didn't make any changes to EXPO; therefore, EXPO-certified memory kits that work on Zen 4 will also work flawlessly on Zen 5.

Intel Desktop Processors

Intel's Core Ultra 200S (Arrow Lake) processors natively support DDR5-6400 with the latest CUDIMMs. With regular UDIMMs, the native support remains at DDR5-5600, just like the 13th Generation Raptor Lake and 14th Generation Raptor Lake Refresh processors. However, you cannot combine CUDIMMs and UDIMMs on the same motherboard. One crucial change with Arrow Lake is that Intel has officially axed support for DDR4.

In a 1 DPC setup, Arrow Lake enables DDR5-5600 using single- and dual-rank memory modules. In contrast, the 2 DPC configuration officially supports DDR5-4800 for single-rank memory modules, while it decreases to DDR5-4400 for dual-rank modules.

Arrow Lake maintains the gear ratio system, which Intel introduced with Rocket Lake. Gear 1 has the processor's memory controller and memory speed in a 1:1 balance. At the same time, Gear 2 runs the memory controller at half speed (0.5:1). Gear 4 is helpful with memory kits that exceed DDR5-9600 at the expense of an even higher performance penalty than Gear 2.

Test Systems

Our Intel DDR5 testbed includes the Core Ultra 9 285K (Arrow Lake) and the MSI MEG Z890 Unify-X. In contrast, the AMD DDR5 system features a Ryzen 9 9900X (Granite Ridge) and the MSI MPG X870E Carbon WiFi. The Corsair iCUE Link Titan 360 RX LCD CPU liquid cooler keeps our chips cool.

We utilize the MSI GeForce RTX 4080 16GB Gaming X Trio for our gaming benchmarks. With speeds reaching up to 7,400 MB/s, the TeamGroup A440 Lite 2TB stores our Windows 11 24H2 installation, benchmarking software, and games. A Corsair RM1000x Shift power supply delivers clean power and facilitates easy cable management for our DDR5 test platforms.

Our previous Intel DDR4 system featured the Core i9-10900K (Comet Lake) and the Asus ROG Maximus XII Apex. The AMD equivalent utilized the Ryzen 9 5900X (Vermeer) processor and the Asus ROG Crosshair VIII Dark Hero. For several years, the MSI GeForce RTX 2080 Ti Gaming Trio reliably provided us with its graphical capabilities. Crucial's MX500 500GB SSD served as our primary drive for Windows installation and benchmarking software, while the 2TB drive was dedicated to our more storage-intensive games. Meanwhile, the Corsair RM650X was selected to power our DDR4 systems.

No matter the testbed, the Streacom BC1 open-air test bench has been excellent, providing a comfortable space for all our hardware.

RAM Pricing Trends

The memory shortage, which began in late 2025, has caused DDR4 and DDR5 memory kits to increase substantially in price. Pricing is fluctuating more rapidly than usual. Some memory kits, especially the high-capacity ones, have tripled in value over this period. The shortage has impacted both DDR5 and older DDR4 memory kits, but more DDR5 than DDR4.

The AI boom is far from over, as many experts believe memory prices likely won't stabilize until 2028. As a result, Intel and AMD are reviving older DDR4-based platforms to somewhat rescue consumers from surging prices. Many U.S. retailers have also started bundling memory kits with motherboard or processor bundles that sometimes represent the closest you'll get to normal pricing.

Whether you're shopping for a memory kit on our list of Best RAM or one that's not, you may find savings by checking out the latest Newegg promo codes, Best Buy promo codes, and Micro Center coupon codes.

MORE: CPU Benchmarks and Hierarchy 2025

MORE: GPU Benchmarks and Hierarchy 2025

Whether you're looking for the best RAM for gaming to upgrade a PC that struggles with today’s titles or building a new PC from scratch, the most suitable RAM kit for your budget depends on the platform you choose and the workloads you plan to run.

The tricky part is figuring out whether faster memory improves your system’s performance. For instance, if you’re using an Intel system with the best graphics cards, most programs won’t respond significantly to faster or slower memory. However, certain workloads, such as specific games and software, benefit from higher data rates. For example, file compression programs favor fast memory. Check out our RAM benchmark hierarchy and a ranking of the best RAM.

AMD’s Zen-based processors benefit more from higher memory frequencies, which you can read about here. Increased memory speeds on AMD Ryzen- and Threadripper-based systems often lead to real-world performance gains. In gaming, this results in higher frame rates at common resolutions, such as 1080p (1920x1080), and smoother gameplay at higher resolutions. However, the number of extra frames gained with faster RAM can vary greatly depending on the game.

Lastly, memory speed greatly affects gaming performance with integrated graphics, whether using an Intel or AMD processor (you can see how they compare in our CPU Benchmark Hierarchy). The graphics engine built into most of the best CPUs for gaming typically lacks dedicated memory, unlike discrete graphics cards, so faster RAM can also improve performance. However, if you have to pay a premium for the fastest RAM to get playable frame rates, investing in slower system memory and a discrete graphics card is a smarter choice.

In short, the best RAM is usually the fastest, especially if you’re gaming without a dedicated graphics card, using an AMD Ryzen system, and sometimes with Intel chips in specific cases. If you're looking for the best plug-and-play compatibility, DDR5-5600 works directly with AMD Ryzen 9000 (Granite Ridge) and Ryzen 9000 X3D processors. At the same time, DDR5-6400 CUDIMMs are compatible with Intel Core Ultra 200S (Arrow Lake) processors.

The global memory market is currently experiencing a significant shortage driven by the explosive growth in artificial intelligence and data center expansion. Some memory kits have doubled and even tripled in price. Despite these market fluctuations, our recommendations remain valid for both DDR5 and DDR4 platforms. While it would be unrealistic to expect these memory kits to maintain the exact pricing we observed during our original testing and review period, the value proposition of each kit holds. If you find any of our recommended memory kits at a competitive price, we strongly encourage you to act quickly. Memory deals are hard to come by, and they sell out within hours, or even minutes, of being listed.

Best RAM For Gaming You Can Buy Today

Best Performance 32GB DDR5 Kit

It's hard to find reasons not to like the Trident Z5 Neo RGB DDR5-6000 C30. The memory kit functions flawlessly out of the box and delivers solid performance. That alone is enough to win over the majority of buyers. Looks, as usual, are subjective, but you can't dispute the Trident Z5 Neo RGB's premium exterior. Like its competition, G.Skill utilizes SK hynix M-die ICS for the memory kit, so some headroom remains for tweaking in memory modules.

DDR5 pricing still fluctuates, but current DDR5-6000 C30 memory kits are available starting at $97.97. When you consider that, the Trident Z5 Neo RGB DDR5-6000 C30's $117.99 price tag suddenly doesn't look so bad. The Trident Z5 Neo RGB DDR5-6000 C30 has proven to be a formidable memory kit for AMD and Intel platforms. There will ultimately be faster memory kits down the road, but in the meantime, the Trident Z5 Neo RGB DDR5-6000 C30 is as good as it gets.

Read: G.Skill Trident Z5 RGB DDR5-6000 C30 review

Best 32GB DDR5 Kit

Best 192GB DDR5 Kit

While not focused on speed, the Vengeance DDR5-5200 C38 is the kind of memory kit that satisfies users who need a lot of memory for their work. This use case includes professionals like content creators, software developers, and data scientists—just to name a few. Current AMD and Intel processors support DDR5-3600 and DDR5-4400, respectively, in a four-DIMM setup. Therefore, DDR5-5200 is technically overclocked, so your results may vary depending on your processor and motherboard.

Corsair currently sells the Vengeance DDR5-5200 C38 for $649.99. The company also offers a more eye-catching version with RGB that adds a $10 premium. It's clear that the memory kit costs a lot, but it's a valuable asset if you need plenty of memory for your work.

Read: Corsair Vengeance DDR5-5200 C38 review

Best 96GB DDR5 Kit

The Trident Z5 RGB DDR5-6400 C32 is a fast-performing memory with generous capacity for professional users. Occupying just two memory slots, the memory kit is the ideal solution for compact workstations and enthusiast small-form-factor builds based around mini-ITX motherboards or simply conventional ATX motherboards that can only house two DIMMs.

G.Skill sells the memory kit for $339.99, a reasonable price for a 96GB memory kit of this caliber. Cheaper options are available on the current market, but they typically run at lower frequencies and don't deliver the same level of performance as the Trident Z5 RGB DDR5-6400 C32.

Read: G.Skill Trident Z5 RGB DDR5-6400 C32 review

Best 32GB DDR5-4800 Kit

The Samsung DDR5-4800 C40 memory kit targets consumers who want to stick to the JEDEC baseline speed. The memory modules don't look pretty but have shown excellent overclocking headroom. There's a good potential to hit breakneck speeds with tight timings. The B-die ICs will help you push the memory modules. Our sample got to DDR5-5800 C36 easily. However, there is no guarantee overclock, so this memory kit belongs to the high-risk, high-reward category.

The memory modules sell for $54.50. That's the consumer pricing, of course, not the IC pricing. As a result, a two-DIMM configuration will set you back $109, so it's not a bad price if you want to run JEDEC speeds. The Samsung DDR5-4800 C40 memory kit can be your ticket into the B-die overclocking world.

Read: Samsung DDR5-4800 C40 review

Best 16GB Kit For Ryzen Owners

TeamGroup did a great job with the Xtreem ARGB DDR4-3600 C14 memory kit -- It certainly ticks all the right boxes. The memory kit looks fantastic and performs equally well when lit up or powered down. The Xtreem ARGB is the fastest DDR4-3600 C14 memory kit we've tested.

The memory market only has a handful of DDR4-3600 C14 memory kits at the 16GB (2x8GB) capacity. The Xtreem ARGB is the least expensive, at $169.99. So, our only gripe is its availability. Unfortunately, Newegg is the only retailer that lists the memory kit, so finding it could be challenging.

Read: TeamGroup T-Force Xtreem ARGB DDR4-3600 review

Best High-Speed 16GB Kit

For those with a board that can handle its top speed and games or workloads that can take advantage of it, Patriot’s Viper Steel DDR4-4400 16GB kit is an excellent high-performance option that also skips RGB.

Keeping the kit simple has allowed Patriot to equip the Viper Steel with enhanced timings that dramatically boost the performance of specific programs, including some games. The Viper Steel DDR4-4400 excels in a market filled with gaming rigs and builds focused on singular tasks.

Read: Patriot Viper Steel DDR4-4000 review

Best High-Speed RGB 16GB Kit

Patriot bucks the trend of pairing cosmetic features with mainstream DRAM ICs instead of pushing data rates up to 4,133 megahertz (MHz) on its Viper RGB. As a result, this DDR4-3600 kit is only $10 more than the white-LED version and several dollars cheaper than competing products with similar latency. That makes it a great value at this speed, though slower kits have more significant pricing advantages.

The fastest DDR4-3600 kit we’ve tested, Patriot’s Viper RGB DDR4-3600 kit, provides excellent value to buyers who want to go (overclocking capability) and show (RGB LEDs).

Read: Patriot Viper RGB DDR4-3600 review

Best Overclocker Value 16GB Kit

It may not have fancy software-controlled RGB lights, and faster kits are available for higher prices, as are budget-priced kits that cost less. But for many who don’t want or need their memory to glow like a rainbow, Patriot’s Viper 4 DDR4-3400 C16 16GB (PV416G340C6K) sits in a sweet spot of price and performance.

The kit includes two 8GB modules rated with XMP values of DDR4-3400 CAS 16-18-18-36. Those last three numbers aren’t great, but DDR3 lovers must remember that 16 cycles at a 3400 MHz data rate have the same latency time as eight cycles at 1700 MHz. So this kit has an excellent DRAM overclocking value and provides some stylish red heatsinks to ensure your memory looks good enough to show off in your windowed case.

Read: Patriot Viper 4 DDR4-3400 review

Best 32GB RGB Kit

Builders who prioritize aesthetics face a tough choice between the best-looking and best-performance parts. Corsair offers a bit of both in its Vengeance RGB DDR4-3200 kit, providing four 8GB DIMMs (32GB total) at CAS 16 timings for a reasonably moderate price, given recent market trends. And this kit isn’t all about looks; it has the goods where benchmarks are concerned, too.

Superb performance and moderate pricing earn the Vengeance RGB Pro DDR4-3200 our Editor’s Choice Award for RGB-equipped memory, though its pastel colors might be off-putting to a few builders.

Corsair’s kit beat our previous favorite, the HyperX Predator RGB, in overall performance at every speed, barring the DDR4-4000 setting it didn’t reach. Corsair also provides a better-performing XMP value and a lower price than the competition, making the Vengeance RGB Pro the uncompromised winner and an excellent addition to your next RGB build.

Read: Corsair Vengeance RGB Pro DDR4-3200 review

Best Two-DIMM 32GB Kit

Buyers in the performance PC market have only two excuses for stopping at DDR4-3200: Either they’re trying to save money or coax good performance from a system that can’t go much farther.

For those in either category, but the former in particular, Patriot Patriot’s 32GB Viper Steel 3200 kit outshines the competition in terms of pricing price, beating its closest competition by approximately 18%. That’s without any performance penalties beyond having the exact mid-market timings as its competitors.

Even though the Patriot Viper Steel’s performance victories are less than 1% overall, its low price puts it well ahead of even the least-expensive competitor in our basic performance-to-price comparison. As a result, value seekers within the performance PC market have just found their new champion.

Read: Patriot Viper Steel DDR4-3200 review

Best Two-DIMM 64GB Kit

The Viper Steel DDR4-3600 C18 is a terrific memory kit for content creators or professionals who don't have the luxury of many DDR4 memory slots. Performance isn't a problem because the memory kit excels at everything you throw at it.

Patriot practically binned these modules to the max, so overclocking headroom is almost non-existent, even if you're willing to go wild on the voltage. However, running the Viper Steel at the advertised frequency should be sufficient in most scenarios.

Patriot prices the Viper Steel DDR4-3600 C18 64GB memory kit very attractively. At $239.99, the memory kit is neither cheap nor expensive. Viper Steel finds itself right in the middle of the competition. Given that it stands tall in terms of performance, this kit is easy to recommend for those needing speed and density in a dual-DIMM scenario.

Read: Patriot Viper Steel DDR4-3600 C18 2x32GB review

Best High-Speed 32GB Kit for Manual Tuning

It's easy to write G.Skill's Trident Z Neo DDR4-3600 C16 kit off if you don't look beyond the published specifications. However, during our review, the RAM kit proved its performance and value in different workloads.

The real value is hidden below the heat spreader—the memory modules use Samsung B-die ICs. With patience and time, you can optimize the timings for better performance. As always, your overclocking mileage will vary, but we got our sample down to CL13.

The Trident Z Neo memory modules come with XMP timings of 16-16-16-36, which aren't the worst in the memory world. However, with the DRAM voltage set to 1.45V, we tightened the timings to 13-14-14-35 before the kit became unstable.

Read: G.Skill Trident Z Neo DDR4-3600 C16 2x16GB review

Best 16GB Kit for H370 and B360

Intel’s H370 and B360 chipsets instruct its Core i5 (and above) processors to lock out any memory settings above DDR4-2666, which is particularly unfortunate in a market that’s moved way past that setting. DDR4-3200 is now mainstream within the enthusiast PC market and is often treated as such by the memory sellers that cater to enthusiasts and gamers. Thus, the best way to get a top-performing brand-new DDR4-2666 kit would be to dial the way-back machine to 2016 and get the high-performing equipment from that time. Barring that, Corsair has a workaround.

Corsair’s easy workaround for the performance problem of Intel’s DDR4-2666 limit was to use older, low-density chips to populate its 8GB DIMMs with two ranks rather than the single rank of its competitors. The best part is that they did this without a significant price increase. We recommend it for anyone whose XMP-compatible platform has a maximum DDR4-2666 data rate, including most retail boxed H370 and B360 motherboards.

Read: Corsair Vengeance LPX DDR4-2666 review

Quick RAM Shopping Tips 2026

Savings on the Best RAM

Whether you're buying DIMMs that made our list of the best RAM or not, you may find savings by checking out our lists of Newegg promo and Corsair coupon codes.

TrendForce reports that DRAM and NAND flash prices are receiving even more price cuts this quarter, despite major price discounts and production cuts already occurring within the industry right now. Overall, the industry is expecting another 18% discount starting in Q2 2023, which will make DDR4 prices even cheaper for PC users. However, the storage discounts will only apply to the mobile and server markets, not necessarily for the Best SSDs in the consumer space.

Prices are dropping again due to a continued reduction in demand for these components across the PC, server, and mobile spaces. The PC industry in particular is suffering heavily from ample DDR4 supply, which is not selling in large enough quantities. Meanwhile, the server market is suffering from weak server demand overall leading to further DDR4 oversupply, forcing suppliers to reduce prices even more.

However, the pricing impact on DDR5 is expected to be very little, since the DDR5 market share remains low. DDR5 is still relatively new and will take time to fully replace DDR4 as a mainstream item. The same applies to server DDR5 but is compounded by PMIC compatibility issues narrowing DDR5 server prices.

The story is a little different for mobile DRAM (i.e. memory for smartphones, tablets, handhelds, etc.) Smartphone brands have actually corrected their memory inventories and killed off any excess DRAM inventory they once had. But despite this, inventory pressure on suppliers continues to remain high due to buyers merging Q3 demand with Q2 to increase their bargaining leverage.

Overall, the average price of DRAM memory in the PC and server market is expected to drop by 15–20 percent, and 13–18 percent for the mobile DRAM market in Q2 2023.

For server and mobile storage, the same issues affecting both markets are also affecting SSDs. Declines in server demand, paired with reduced orders for server ODMs, have led to high inventory levels of enterprise SSDs. This is causing prices to decline by 10–15 percent. Mobile UFS storage discounts are expected to expand by 10–15 percent as well, again due to buyers merging Q3 demand with Q2 demand to increase bargaining power, causing high inventory pressure on suppliers.

The memory shortage is taking hold, fueled by the massive AI data centers that are soaking up the world's memory stocks. This is causing a supply shortage, sending prices skyrocketing, and all indicators point to further price increases over the coming months. In fact, prices are expected to double throughout next year. There's no better time to buy a memory kit before the price hikes become even more severe, and our Best Black Friday RAM Deals article will tell you where you can buy the best RAM at affordable prices.

It's now become one of the biggest questions for new builders: Which type of memory wins the DDR4 vs DDR5 battle? This has become especially important as some of the best CPUs for gaming support DDR4 or DDR5 memory. But, of course, you also have to determine if there's a big enough performance gain to justify the upgrade. But sticking with the DDR4 memory kit you have is now an option that builders are taking far, far more seriously due to price increases.

DDR5 comes with many promises, but one of its most significant selling points is the higher bandwidth it can deliver to processors with tons of cores. Memory bandwidth has become increasingly important as modern chips can reach up to 16 cores in mainstream PCs. The advent of new memory kits with non-binary capacities, including 48GB and 96GB kits, provides increased flexibility in pricing and capacity options. Memory manufacturers have also increased the capacity per module, with conventional DIMMs now scaling up to 64GB.

We'll check out the details below to compare DDR4 vs DDR5 in the spec sheets, then dive deep into testing to see where each type of memory is most effective.

DDR5 vs DDR4 Specifications

The DDR5 memory standard promises a future with denser memory sticks, ultimately resulting in more memory capacity in your system. DDR4 stopped at 16-gigabit memory chips, but DDR5 can use up to 64-gigabit memory chips. The latter also supports die stacking with up to eight dies per chip, meaning DDR5 can top out at 2TB per module. That'll be for servers that come with LRDIMMs, RDIMMs, or MRDIMMs. DDR5 will probably stop at 128GB per stick in the mainstream market. Consumer DDR5 memory modules currently max out at 64GB, paving the way for 256GB (4x64GB) memory kits.

According to the JEDEC (Joint Electron Device Engineering Council) specification, DDR4 data rates range from DDR4-1600 to DDR4-3200. Therefore, it's easy to think of DDR5 as a continuation of DDR4 since initial DDR5 starts at DDR5-3200 and spans up to DDR5-6400. Modern DDR5 now incorporates a clock driver (CKD) that acts as a buffer, further increasing the data rate. The highest-clocked DDR5 memory kit presently sits at DDR5-9600. At the rate at which DDR5 is progressing, it shouldn't be long before we cross the DDR5-10000 mark.

However, looking back at the beginning of the DDR4 era, DDR4-1600 memory never really took off. Instead, DDR4-2133 served as the baseline for DDR4. DDR5 follows a similar pattern. Although JEDEC has specified data rates as low as DDR5-3200, the starting point for many, if not all, mainstream DDR5 products is DDR5-4800.

Unlike the last transition from DDR3 to DDR4, DDR5 doesn't have more pins than its predecessor. Instead, DDR5 retains the 288-pin arrangement, but with different pinouts. As a result, the notch position has changed and will help prevent less-experienced users from trying to insert a DDR5 module into a DDR4 slot, or vice versa. That's only a small change, though. The real game-changer lies at an architectural level you don't see on the DIMM's exterior.

DDR4 memory modules have a single 64-bit channel (72-bit if ECC is enabled). In contrast, DDR5 memory modules feature two independent 32-bit channels (40-bit with ECC). JEDEC also doubled the burst length from eight bytes (BL8) to 16 bytes (BL16). As mentioned earlier, the upgrades improve efficiency and reduce data access latency. In a dual-DIMM setup, this transformation essentially converts DDR5 to a 4 x 32-bit configuration rather than the conventional 2 x 64-bit configuration on DDR4.

To continue the push for better power efficiency, DDR5 operates at 1.1V, down from DDR4's 1.2V. However, the only memory kits you'll find at 1.1V conform to JEDEC's timings. For example, the standard operating voltage for DDR4 is 1.2V, but overclocked or higher-binned memory kits with tighter timings are more demanding of voltage. Just like how we've seen DDR4 scale up to DDR4-5000 at 1.6V, DDR5 will likely climb the voltage ladder as well. It isn't a contest, but 1.45V is the highest DDR5 has gone so far (DDR5-9600).

Intel's Extreme Memory Profile (XMP) extension has evolved alongside DDR4, so we now have the third iteration. So, what has changed with XMP 3.0? Well, there are now up to five XMP profiles, and users can modify and save two custom XMP profiles directly on the SPD. Rivaling chipmaker AMD has introduced the AMD EXPO with a similar purpose as Intel XMP 3.0. Nowadays, DDR5 memory kits are available with support for either one or the other, or both.

DDR5 marks a radical change in voltage regulation, too. The motherboard is no longer responsible for voltage regulation, as the memory modules include a power management IC (PMIC). (12V on server-grade DIMMs and 5V on mainstream DIMMs.)

The PMIC takes the 5V input from the motherboard and converts it to usable voltages for the voltage rails, comprising VDD (1.1V), VDDQ (1.1V), and VPP (1.8V). The PMIC helps improve voltage regulation and signal integrity and reduce noise. However, the change is a double-edged sword. The voltage regulator on the DDR5 memory module helps reduce motherboard cost and design complexity, but ultimately transfers the cost over to the memory modules. It also makes DDR5 dependent on the supply of PMIC chips.

In addition to higher bandwidth and improved power consumption, DDR5 will also offer higher capacity per memory module. Memory density and banks go hand-in-hand. When you increase density, you also need to increase the number of banks to accommodate the extra capacity. DDR5 features a 32-bank structure divided into eight groups. In comparison, DDR4's 16-bank system sports four groups. There are still four banks per group - that didn't change. The increase from 16 to 32 banks enables more pages to be opened consecutively. DDR5 also has the Same Bank Refresh function (SBRF), allowing it to refresh one bank per group instead of all banks.

On-die ECC (ODECC) is another critical feature of the DDR5 specification, but it shouldn't be confused with standard ECC. Manufacturers turn to smaller nodes to increase the density of memory chips, and on-die ECC is designed to correct potential errors within those chips to improve reliability. Unfortunately, the protection is limited to the memory arrays inside the chips — the data is on its own once it moves outside the DIMM. On-die ECC doesn't protect data in transit, which is why it isn't a proper ECC implementation.

One can question the utility of on-die ECC, since errors are more prevalent when data travels over the memory bus. Furthermore, on-die ECC requires additional capacity to store parity, which adds another cost to DDR5 (in addition to the PMIC). On-die ECC isn't a replacement for standard ECC, but customers will use both in unison in a server or enterprise environment.

G.Skill Trident Z5 RGB DDR5-6000 C36

Intel's 12th Gen Alder Lake processors are older now, but they allow you to compare DDR4 vs DDR5 because they are the first consumer chips to support DDR4 and DDR5, and the Alder Lake Refresh chips also take the same approach, so we paired the flagship Core i9-12900K processor with G.Skill's Trident Z5 DDR5-6000 32GB (F5-6000U3636E16GX2-TZ5RS) memory kit. It's a dual-channel memory kit with two 16GB DDR5 modules, each with a single rank.

By default, the Trident Z5 RGB memory modules operate at DDR4-4800 at 1.1V with JEDEC timings (40-40-40). The XMP 3.0 profile quickly brings the Trident Z5 RGB up to DDR5-6000 with timings set to 36-36-36-76 and a DRAM voltage of 1.3V. There are faster DDR5 kits, but the Trident Z5 comes with generous overclocking headroom. We didn't have any problems pushing it to higher data rates for our tests.

MSI MAG Z690 Tomahawk WiFi: Brothers From Another Mother

Throughout the DDR evolution, we've seen some manufacturers offer a combination of new and legacy RAM support on some motherboards. In the past, it wasn't unusual to find motherboards that supported both DDR and DDR2. We saw the same trends with DDR2 and DDR3, and even DDR3 and DDR4. However, we don't expect to see a hybrid DDR5 motherboard due to the migration of voltage regulation to the DIMM. It's just too complex for both technologies to coexist on a single motherboard.

We used MSI's MAG Z690 Tomahawk WiFi motherboard in both DDR4 and DDR5 flavors. They're essentially the same motherboard with different RAM slots, making them directly comparable. It helps us eliminate or reduce the performance delta to a minimum, rather than using two different motherboards from other brands or product tiers.

The MAG Z690 Tomahawk WiFi implements MSI's Memory Boost technology, which uses optimized memory circuits to deliver pure data signals. In addition, the DDR4 and DDR5 motherboards support DDR4-5200 and DDR5-6400, respectively, which is more than sufficient for our needs.

Test System and Methodology

To ensure a level playing field, both DDR4 and DDR5 memory modules must operate in a similar configuration with identical density and an equal number of memory ranks.

Modern DDR4 16GB memory modules can feature single-rank (1Rx8) or dual-rank (2Rx8) designs with 16-gigabit and 8-gigabit ICs, respectively. DDR5 16GB memory modules, in contrast, come only in a single-rank layout with 16-gigabit chips. Therefore, we compared the single-rank 32GB (2x16GB) DDR5 memory with an equivalent single-rank 32GB (2x16GB) DDR4 memory kit for an apples-to-apples comparison.

On the DDR4 end, we used Crucial's Ballistix Max RGB DDR4-4000 32GB (2x16GB) memory kit, consisting of two 16GB single-rank DDR4 modules, so there's no better point of comparison. The Ballistix Max RGB memory modules are suitable for DDR4-4000 at 1.35V and 18-19-19-39 timings.

We tested different standard data rates, including DDR4-2133, DDR4-3200, and DDR5-4800 at JEDEC timings. As you know, JEDEC utilizes three separate bins for each data rate (A, B, and C). For our tests, we chose the middle ground. That would be 15-15-15 for DDR4-2133, 22-22-22 for DDR4-3200, and 40-40-40 for DDR5-4800.

Additionally, we evaluated data rates at the tightest possible timings to compare early DDR5 with some of the best DDR4. Do note that the Ballistix memory kit uses Micron ICs, so we couldn't get the timings as low as we would like. The Trident Z5 RGB memory kit, which uses Samsung ICs, achieves the best timings at each data rate. It offers a good representation of the pinnacle of DDR5 performance at the moment.

As usual, we manually configured each data rate and its corresponding primary timings. The motherboard handled the secondary and tertiary timings. Alder Lake officially supports up to DDR4-3200 on Gear 1. However, our sample could hit DDR4-4000 in Gear 1. DDR5, on the other hand, defaults to Gear 2, translating to higher latency. It's foul, but that's the way Intel designed Alder Lake.

All About The Bandwidth

Compared to the baseline, DDR5-4800 C40 delivered 112% more bandwidth than DDR4-2133 C15 and 46% more than DDR4-3200 C22. Against DDR4-4000 C16, there was a 19% margin in favor of DDR5-4800 C40. DDR5-6400 C36 is the best configuration out of the lot — it was just 26.43 GBps shy of hitting the 100 GBps mark in Sandra 2021.

While bandwidth has improved tremendously with DDR5, latency has gotten worse. That's within expectations because DDR5 has looser timings. Even run-of-the-mill DDR4-2133 C15 was 5% faster than DDR5-4800 C40. The margin jumped to 17% with DDR4-3200 C22.

We had to dial the transfer rate up to DDR5-6400 C36 to match DDR4-3200 C22 latency. Meanwhile, DDR4-4000 C16 was 13% faster than DDR4-6400 C36.

DDR5 vs DDR4 Performance

In our overall performance measurement, DDR5-4800 C40 was 19% and 14% faster than DDR4-2133 C15 and DDR4-3200 C22, respectively. However, the performance delta decreased when we compared it to performance DDR4. DDR5-4800 C40 was only 9% faster than DDR4-3200 C15. When we went up to DDR4-4000 C16, the advantage of DDR5-4800 C40 dropped to 5%.

Looking at the best, DDR5-6400 C36 was 11% faster than DDR4-4000 C16. So naturally, you can minimize the difference even more if you use DDR4-4000 C14. But, unfortunately, our Ballistix memory kit was not up to the task.

Predictably, not every workload type will benefit from DDR5's higher bandwidth. Take your everyday tasks, for instance, with Microsoft Office. The delta between the best (DDR5-6000 C36) and the worst (DDR4-2133 C15) was only 4%.

The performance benefits vary even in more specialized workloads. For example, we compared the fastest DDR4 and DDR5 memory kits in Adobe Lightroom (DDR4-4000 C16 vs. DDR5-6400 C36) and found that DDR5 is 28% faster, but the gains in Adobe Photoshop (DDR4-3200 C15 vs. DDR5-6400 C36) and Adobe Premiere (DDR4-3600 C16 vs. DDR5-5600 C36) were less than 1% and 3%, respectively.

Compression jobs will benefit from DDR5. DDR5-4800 C40 was 46% faster than DDR4-4000 C16. However, higher data rates didn't exhibit the same performance gains. For example, DDR5-6400 C36 was 14% faster than DDR5-4800 C40.

DDR5 also excelled in y-cruncher— DDR5-6400 C36 finished the calculations 25% faster than DDR4-4000 C16.

DDR5 vs DDR4 Gaming Performance

Modern processors and memory are more than sufficient for gaming, and the graphics card is the bottleneck in most cases. The DDR5 results clearly showed that. The performance difference in gaming between the two extremes (DDR4-2133 C15 and DDR5-6400 C36) was only around 8%. However, once we stepped up to something better, like DDR4-3200 C15, DDR5-6400 C36 was only 2% faster.

Focusing on the best DDR4 and DDR5 performance, we saw a 3% difference in Assassin's Creed Valhalla, 2% in Far Cry 6, Tom Clancy's Ghost Recon Breakpoint, Watch Dogs: Legion, and Borderlands 3, and a 1% delta in Shadow of the Tomb Raider and Wolfenstein: Youngblood.

Our Key Takeaways

• DDR5 is fast, but only in some workloads. On the one hand, our tests revealed that specific tasks benefited substantially from DDR5, and you can expect double-figure performance gains. However, some workloads were unaffected by DDR5 or showed only a minimal performance improvement. Therefore, you should identify the type of workloads that you use on your system and decide whether DDR5 is worth the investment.

• Don't buy DDR5 for gaming. The performance uplift is there, but it doesn't warrant an upgrade. Yes, DDR5 helps improve your frame rates, but you also need to keep your expectations in check. So, unless you're a hardcore gamer who doesn't want to know you're leaving performance on the table, you shouldn't pick up DDR5.

• For now, DDR4 offers more bang for your buck. For comparison, the cheapest DDR5-4800 32GB C40 memory kit retails for $73, whereas a DDR4-3200 32GB C16 memory kit costs as little as $50, as it did four years ago. While the former offers 6% higher performance, it's also 46% more expensive than the latter. DDR5 pricing won't remain this high forever; however, the current situation has slapped significant markups on DDR5 memory.

• DDR5 has more future-proofing value. Intel and AMD have stopped supporting DDR4 on their latest processors. Upgrading to a DDR5 memory kit today means you can reuse it for future platforms. DDR5 is no longer wet behind the ears, so there are some pretty high-end memory kits out there.

• MORE: Best RAM 2023
• MORE: How to Test RAM: Making Sure Bad Memory Isn't Crashing Your PC
• MORE: DDR DRAM FAQs And Troubleshooting Guide
• MORE: All Memory Content

DDR5 memory was expected to reach price parity with DDR4 or at least hit sweet spot in terms of pricing this year. However, a report from DigiTimes citing memory module makers claims that DDR5 memory price drops are slowing down — and are expected to stop, soon — due to supply-side constraints. 

Now that both AMD and Intel are shipping processors supporting DDR5 in high volumes and leading producers of memory have started mass producing various DDR5 memory chips and modules, prices of DDR5 memory should comply with supply-demand balance — and drop, because demand for PCs among consumers is weak. 

But it looks like there are two constraints that keep DDR5 prices from dropping. First, DRAM chipmakers are reportedly delaying deliveries of DDR5 chips, and, second, there is a shortage of power management ICs (PMICs) for server-grade memory modules.

It is impossible to tell why companies such as Micron, Samsung, and SK Hynix have all decided to delay shipments of their DDR5 memory chips to module makers. But if there are not enough chips, module makers cannot build modules and price drops will slow down. It should be noted that the report did not name a single DRAM vendor.

TrendForce last week reported an issue with PMIC compatibility for server-grade DDR5 RDIMMs and said that DRAM suppliers and PMIC vendors were working together to resolve the problem. The report suggested that PMICs from Monolithic Power Systems (MPS) have not been affected, leading analysts to predict high demand for MPS PMICs and creating a supply bottleneck for server-grade DDR5 memory modules. The issue of PMIC compatibility is a potential hurdle for the adoption of DDR5 memory in servers, but the collaboration between DRAM suppliers and PMIC vendors indicates a commitment to resolving the issue and ensuring the smooth adoption of DDR5 memory in the server market eventually.

Given that the DigiTimes report mentions two isolated factors for slowing down DDR5 price drops — delayed DDR5 supplies by unnamed vendors and potential shortages of server-grade DDR5 PMICs by MPS — it's hard to draw conclusions about how significantly actual DDR5 prices will be affected. Consumers are unlikely to directly suffer from server-grade DDR5 modules getting more expensive, however, slowing down of DDR5 ramp will indeed affect the pace of DDR5 price drops.

Samsung was the only major memory maker that did not cut the output of DRAM and 3D NAND when demand for these commodity ICs dropped due to softening of PC and smartphone sales last year. But as the company's profits plummeted in Q1 2023, it decided to reduce memory production to balance the supply and demand situation on the market.

"We have cut short-term production plans, but as we project solid demand for the mid-to-long term, we will continue to invest in infrastructure to secure essential cleanrooms and to expand R&D investment to solidify tech leadership," a statement by Samsung published by Bloomberg reads.

Samsung is the world's largest supplier of DRAM and NAND, and memory sales significantly contribute to the company's earnings. The company commanded a 45.1% share of DRAM market and a 33.8% revenue share of NAND market in Q4 2022, according to TrendForce.

While the company formally stated that it would cut memory production, it never revealed how significantly it intends to reduce wafer starts and memory bits output. 

Most of Samsung's rivals reduced memory production on older technologies but continued to gradually ramp up memory production on newer fabrication processes. Typically, the latest nodes decrease chip costs and increase bit output per wafer, so in many cases, memory ramp on newer nodes more than offsets production cuts on older nodes regarding bits output. 

However, analysts believe that Samsung's intention to cut down memory production will affect the demand and supply balance in the market and will at least slow down the collapse of memory prices in Q2 2023.

"Counterpoint expects that the utilization rate reduction [will] slow down the decline of commodity memory prices," Brady Wang, a senior analyst, told Nikkei Asia. "Still, this oversupply is due to weakening demand and high inventory, so Samsung's production cut is not expected to stimulate sales. Therefore, Counterpoint believes that the oversupply situation will continue until the third quarter when the market starts to deplete inventory for the fourth quarter seasonal demand." 

Samsung's Q1 2023 revenue dropped to 63 trillion won ($48.877 billion), or 19% compared to the same quarter a year ago. The company's profit collapsed to 600 billion won ($450 million), or by 95% year-over-year, falling short of the 1.4 trillion won ($1.064 billion) average forecasted by analysts.

For its second quarter of fiscal 2023 Micron posted a year-over-year revenue drop of nearly 53% and said its earnings will decline further in the ongoing quarter as demand for 3D NAND and DRAM remains soft. The company had to write down inventory worth billions of dollars, and said it would maintain conservative spending but would have to fire more people than it planned to this year. However, Micron remains confident that demand for memory will skyrocket by 2025 as the world embraces more computing.

Worst Downturn in Years

Micron earned $3.693 billion in Q2 FY2023 (which ended on March 2, 2023) — down from $7.786 billion in the same period a year ago, but still within guidance set by the company last quarter. Micron posted a whopping $2.312 billion GAAP net loss as it had to write down $1.43 billion worth of 3D NAND and DRAM inventory and saw its gross margin collapse to -32.7% — down from 47.2% in Q2 FY2022. 

Despite the fact that its loss per share totaled $2.12, Micron has declared a quarterly dividend of $0.115 per share, payable in cash on April 25, 2023.

Low average selling prices (ASPs) due to oversupply remain the main challenge for 3D NAND and DRAM industries. To somehow slowdown bit output increase, Microsoft said it had further reduced DRAM and NAND wafer starts by ~25% YoY. 

"The semiconductor memory and storage industry is facing its worst downturn in the last 13 years, with an exceptionally weak pricing environment that is significantly impacting our financial performance," said Sanjay Mehrotra, chief executive of Micron. "We have taken substantial supply reduction and austerity measures, including executing a companywide reduction in force. We now believe that customer inventories have reduced in several end markets, and we see gradually improving supply-demand balance in the months ahead."

Due to uncertainties about demand for DRAM and flash memory in the short and mid-term, the company continues to conservatively invest in capacity expansion and intends to accelerate job cuts. Micron announced that it would maintain its FY2023 CapEx at approximately $7 billion (down over 40% from FY2022 and down over 50% for fab tools), which is at the lower end of its previously forecasted range. Micron will also cut its headcount by 15% this year — up from its earlier goal of 10%.

PC Business Remains a Concern

Micron's PC business is a cause for concern: the company has projected a mid-single-digit decline in PC unit volume in 2023, which will bring the volume back to pre-COVID levels. Although client customer inventories are still high, they have deteriorated significantly and Micron anticipates a surge in bit demand in 2H FY2023. 

Micron is significantly more optimistic about its datacenter business — it forecasts revenue growth in fiscal Q3  (its DC revenue bottomed in Q2 FY2023). The company projects its datacenter customer inventories to return to relatively healthy levels by the end of calendar year 2023. 

As for its mobile business, Micron seems to consider this a mixed bag. On one hand, smartphone shipments are expected to decline in 2023 compared to the prior year. On the other hand, while some smartphone suppliers' inventories have returned to normal levels, other OEMs' inventories remain high. Micron anticipates an improvement overall in mobile customer inventory throughout the remainder of calendar year 2023 — and predicts growth in mobile DRAM and NAND bit shipments in H2 FY2023 compared to H1 FY2023. 

By contrast, Micron's graphics and automotive memory businesses remain profitable and look like they'll stay that way going forward.  

On the graphics side, Micron says that customer inventory adjustments are proceeding smoothly and the company expects a stronger demand in H2 2023 (which is inline with yearly trends for sales of discrete graphics cards). The memory maker cites industry analysts who anticipate that graphics DRAM total addressable market (TAM) to grow at a compound annual rate higher than that of the general market due to its application in both the client and datacenter domains.  

Micron also anticipates a persistent rise in demand for auto memory in the second half of the year, thanks to the gradual reduction of non-memory supply constraints and a rise in memory content per vehicle.

Great Yields of 3D NAND and DRAM

DRAM accounted for 74% of Micron's revenue — and while the company attempted to reduce output of computer memory, its bit shipments increased in the mid-teens percent range quarter-over-quarter. 

Micron has achieved record-breaking yields for 1α DRAM, surpassing the yields of any other node in its history. Furthermore, the targeted yields for 1β (1-beta) manufacturing node — which increases DRAM bit density by 35% and improves power efficiency by 15% — have been achieved ahead of schedule and faster than any of Micron's previous nodes, the company said. Regardless, because average selling prices of DRAM dropped 20% QoQ, the DRAM business still generated a huge loss for the company.

NAND memory accounted for 24% of Micron's Q2 FY2023 revenue, and, since the company accelerated transition to its more sophisticated 176-layer and 232-layer nodes, its bit shipments also increased during the quarter — despite efforts to cut down flash memory output.

More than 90% of Micron's NAND bit production in Q2 FY2023 was represented by 176-layer and 232-layer production nodes, whereas 3D QLC NAND accounted for over 20% of the company's flash memory bit production and shipments. Micron says it has achieved record yields for 176-layer NAND — surpassing the yields of other nodes in Micron's history. Additionally, the targeted yields for 232-layer 3D NAND memory have been reached — not only ahead of schedule, but faster than any of the company's previous nodes.

High yields constrain Micron's losses somewhat. However, since 3D NAND and DRAM ASPs continue to drop, the company will continue to generate losses in the coming quarter (or even quarters).

2025 Set to Be Record Year

Micron now guides GAAP revenue of $3.7 billion ± $200 million for Q3 FY2023, which will represent an approximately 46% year-over-year drop, from $8.642 billion in Q3 FY2022. This is going to be the steepest YoY earnings drop since the dot com bust in 2001. The company forecasts a $1.79 ± $0.07 loss per share in the ongoing quarter. 

Micron remains optimistic about demand for DRAM and 3D NAND going forward as the industry adopts more memory and storage-intensive applications, such as ChatGPT-like large language models (LLMs). 

"While our industry faces significant near-term challenges, we believe that the memory and storage total addressable market (TAM) will grow to a new record in calendar 2025 and will continue to outpace the growth of the semiconductor industry thereafter," said Mehrotra. "Recent developments in artificial intelligence (AI) provide an exciting prelude to the transformational capabilities of large language models, or LLMs, such as ChatGPT, which require significant amounts of memory and storage to operate."

Average selling prices of dynamic random access memory (DRAM) dropped 20% quarter-over-quarter in Q1 2023 as buyers maintained conservative buying behavior, according to TrendForce. Memory quotes are set to drop even further — by 10% to 15% in Q2 2023 — with DDR5 prices declining more significantly than DDR4 prices. 

While Micron, Samsung, and SK Hynix have begun to adjust their DRAM output to decrease supply and at least maintain current price levels, makers of PCs, consumer electronics, servers, smartphones, and graphics cards still have plenty of DRAM ICs and are unlikely to accelerate their purchases before they see increased demand for their products. As a result, the efforts of DRAM makers have yet to make a noticeable impact on declining prices, according to the TrendForce report.

Buyers among PC OEMs and ODMs have reduced their purchase quantities of PC DRAMs significantly in the last three quarters, but they still have enough inventory of DDR4 and DDR5 SDRAM to last around 9 to 13 weeks. There's a chance that PC makers may take advantage of low prices and increase their DRAM purchases, but it's not clear if this would relieve the inventory overstock situation for suppliers. 

Despite reduced output, TrendForce estimates that the price of an 8GB DDR4 module will decline by over 10% in Q2 2023. In general, the average selling price of PC DRAM is set to decrease 10% to 15% QoQ in the first quarter and then by another 10% to 15% sequentially in Q2, according to analysts. 

Unit sales of discrete graphics cards tend to rebound sequentially in the first quarter and then drop in Q2, based on historical data from Jon Peddie Research. Yet, because shipments of desktop GPUs declined drastically in Q3 and Q4 2022, GPU makers have enough GDDR memory chips in stock — and so demand for such ICs is sluggish. TrendForce predicts that average selling price of a 16Gb GDDR6 IC will decline 10% to 15% in Q2 2023 due to constrained demand. 

Thanks to slowing demand for PCs, memory makers increased proportion of server memory in their product mix. However, this led to a significant server DRAM inventory pile-up in Q1 2023. Server makers and cloud service providers are also adjusting their inventories, which is why prices for server memory are going south. TrendForce now projects that the server DRAM average selling price will drop by 13% – 18% in Q2 2023. 

Although smartphone suppliers' DRAM inventories have dropped to a relatively healthy level, these brands are adopting a cautious approach to handset production.  This is likely to limit buyer demand for mobile DRAM in Q2 2023, according to TrendForce. 

Despite mobile DRAM production cuts, it will remain a challenge for suppliers to reverse their current overstock situation. Because every DRAM producer wants to sell every LPDDR IC it has at almost any price, this has a drastic impact on average selling prices. TrendForce believes that average prices of mobile DRAM will continue to drop in Q2 2023, though the decline will narrow to 10% – 15%.

As far as consumer electronics memory is concerned, supply continues to outpace demand even though manufacturers have cut down production of appropriate ICs. TrendForce analysts think the average selling price of consumer DRAM will fall 10 - 15% in Q2 2023.

Micron has introduced its new DDR5 memory modules featuring 24GB and 48GB capacity, reports News.Mynavi.jp (via @momomo_us). The modules are compatible with both AMD EXPO and Intel XMP 3.0 profiles for quickly setting them up and are designed for desktop PCs running AMD's Ryzen 7000-series as well as Intel's 12th and 13th Generations Core processors. 

Micron's new family of DDR5 memory modules feature data transfer rates of 5200 MT/s and 5600 MT/s as well as CL46 latency at 1.1V. The DDR5-5600 DIMMs come in traditional 8GB, 16GB, and 32GB versions, but the interesting capacities are the new 24GB and 48GB models. Those are probably based on 24Gb memory chips, whereas the former likely use 16Gb DRAM ICs. Meanwhile, DDR5-5200 modules are available only in 8GB, 16GB, and 32GB sizes.

Typically, 24GB and 48GB capacities are considered to be optimal for new-generation server platforms as they allow systems to precisely balance memory capacity and the number of cores, which ultimately means lower costs. Meanwhile, support for AMD EXPO and Intel XMP 3.0 profiles designed primarily for enthusiasts in mind indicate that these modules are indeed aimed at desktops. 

Using a couple of 24GB or two 48GB modules instead of a pair of 32GB and 64GB DIMMs allows to build PCs with 48GB or 96GB of dual-channel memory, which are cheaper than machines with 64GB or 128GB of RAM. Meanwhile, capacities like 48GB and 96GB are more optimal for modern CPUs with 16 or 24 cores — you get potentially 2GB or 3GB per core if you're doing VMs, for example.

Unfortunately, Micron has not disclosed recommended prices for its DDR5-5200 and DDR5-5600 modules. Keeping in mind that demand for PCs is projected to be low in Q1 and DDR5 SDRAM prices are set to decline by 18–23% in Q1 2023, expect the new memory modules to be relatively inexpensive despite their high capacity and increased performance.

Murmurings from Taiwanese industry sources indicate that the DRAM market continues to face challenging conditions as 2022 ends. DigiTimes reports that DRAM spot prices are down 40% this year. Specifically, spot prices for all sizes of DDR4 and DDR3 memory ICs have seen significant declines during 2022. Probably of most interest to PC enthusiasts and DIYers, though, is the observation that contract prices of 8GB DDR5 RAM modules have declined by 43% between the end of February and October this year.

Ideally, the whole of the PC industry would have moved over to DDR5 system memory in 2022, with the support of the newest platforms from AMD and Intel. However, those that bought into DDR5 builds had to pay quite a premium to add memory – in addition to the new CPU, motherboard, etc – especially early this year. Some motherboard makers thus cannily offered identical motherboards for the newest-gen Intel CPUs in either DDR4 or DDR5 supporting configurations.

So, the 43% drop in contract 8GB DDR5 module prices in eight months sounds impressive, but what about pricing for consumers? Let's take a peek at Amazon DDR5 price tracking figures for some of the biggest brand 16GB (2x matched 8GB modules) kits and see what has happened since spring 2022 (approx 9 months):

The above Amazon pricing snapshot comparisons show some worthwhile price declines, which means putting together a DDR5 system isn’t so painful to the pocket at the end of 2022. Alternative DDR4 16GB (2x 8GB) kits from the likes of Corsair and PNY cost about $50 to $60 today, so much of the DDR5 premium seems to have vanished. However, faster cutting-edge DDR5 kits released recently aren’t encapsulated by our three-quarters-of-the-year price comparisons.

Moving back to the topic of the DRAM market in general, the DigiTimes report says that it can’t currently see any signs of recovery for this semiconductor business segment. The biggest influences being felt right now are due to DRAM makers having excess inventory, and device makers being cautious with orders. Last month we reported on Micron attempting to cut DRAM (and NAND) output to buoy prices, but such unilateral action usually doesn’t seem to have much of an effect. In Q3, DRAM makers reported a 30% quarterly decline in revenues, but of course, it is a little too early to have/see any Q4 2022 figures.

NAND News Nugget

Before we go, the source article shared a little extra news on NAND flash pricing. It claims that TLC NAND wafer prices continue to trend downwards, by as much as 20% last quarter, and it is starting to be uneconomical to produce these chips. SLC NAND price trends have held up much better though.

The DRAM and NAND contact price action commented upon above is a double-edged sword for consumers. We would like to see cheaper RAM and flash storage in retail, but we also need companies to have a sustainable income from their production - or they could grind to a halt, worsening the cyclical extremes in pricing.

Micron this week announced drastic cost-cutting measures, which includes a 10% workforce reduction as well as further lowering of capital expenditures. As a result, the company will slowdown the ramp of new DRAM nodes, which will delay its introduction of 1γ (1-gamma) production nodes, that use extreme ultraviolet (EUV) lithography to 2025. Meanwhile, the company has begun sampling of 24Gb DDR5 memory devices for enterprise applications.

EUV Roll Out Delayed

Micron is the only large DRAM maker that does not use EUV lithography with its latest fabrication processes. The memory producer plans to use EUV for several layers in its 1γ manufacturing technology, which was set to be introduced sometime in 2024. Because Micron has to reduce spending on new equipment in fiscal years 2023 and 2024 as well as reduce DRAM bit shipments in the coming quarters, it will have to slowdown ramp up of DRAMs on its 1β and 1γ fabrication technologies. 

The company's latest 1β (1-beta) manufacturing node — which increases bit density by 35% and improves power efficiency by 15% — solely relies on deep ultraviolet (DUV lithography). By contrast, Samsung and SK Hynix are already using EUV scanners for several layers in their 4th Generation 10nm-class technologies (1α, 1-alpha) and plan to increase their usage with 5th Generation 10nm-class DRAM nodes. But the company is delaying its 1β to reduce bit production shipments as well as cut down its CapEx, which is why it will delay its 1γ introduction too. 

"Given our decision to slow the 1ß DRAM production ramp, we expect that our 1γ (1-gamma) introduction will now be in 2025," a statement by Micron reads. "Similarly, our next NAND node beyond 232-layer 3D NAND memory will be delayed to align to the new demand outlook and required supply growth." 

The delay of a EUV-based production process is a big deal since one EUV layer replaces several DUV masks, thus reducing cycle times, improving yields, and reducing costs. Keeping in mind that both Samsung and SK Hynix will use EUV extensively in the coming years, they might have an edge over Micron in terms of costs.

24Gb DDR5 IC Sampling

Advanced process technologies are particularly useful for complex high-capacity DRAM devices, such as Micron's 24Gb DDR5 IC. The chip, which is being qualified by Micron's partners, is set to be made using Micron's proven 1α node.  

Usage of 24Gb memory chips instead of 16Gb devices could increase memory module capacity by 50% without increasing the number of ICs per module. For mainstream and ultra-dense servers servers, that means 48GB, 96GB, 192GB or 384GB modules. Also, Micron could produce 24GB and 48GB DDR5 modules for client applications. 

DDR5 ramp in servers will only begin when AMD and Intel start to roll out their next-generation EPYC and Xeon Scalable processors with DDR5 support. Meanwhile, Micron expects server DDR5 bit shipments to crossover with DDR4 in mid-calendar 2024.

CapEx Reduced

As noted, one of the reasons behind the 1β and 1γ delays are CapEx cuts in fiscal 2023 (ends in September, 2023) and 2024 (ends in September, 2024). Micron cut its FY 2023 CapEx  to a range between $7.0 to $7.5 billion, that's down from the $8 billion target the company set several months ago and from $12 billion in FY 2022 (i.e., the CapEx is cut by approximately 40%). The company intends to reduce spending on wafer fab equipment (WFE) by 50% year on year, but construction spending will increase as the company is building up its new fab in Idaho. WFE spending will also be down in FY 2024 compared to what Micron originally planned.

Lay Offs Coming

Since the company expects meagre demand growth for both types of memory it produces — 10% in DRAM and around 20% in NAND — it needs to reduce its operating expenses too. As a result, it plans to reduce headcount by 10% throughout 2023 'through a combination of voluntary attrition and personnel reductions.' 

"For both years, demand in DRAM and NAND is well below historical trends and future expectations of growth, largely due to reductions in end demand in most markets, high inventories at customers, the impact of the macroeconomic environment, and the regional factors in Europe and China," Micron said in its statement.

Balancing Supply and Demand

While the output and cost cutting measures Micron has introduced look drastic, the company believes it needs to balance supply with demand as its profitability and long-term prosperity depends on it.  

"The industry remains in an oversupply situation, but customers are depleting inventories, and we expect them to be in a better position by the second quarter of the calendar year," said Mark Murphy, chief financial officer of Micron (via SeekingAlpha). "But profit is going to be challenged through the year, and that will challenge gross margins." 

"The rate and pace of the recovery in terms of profitability depends on how fast supply is brought into line," Sanjay Mehrotra, chief exec of Micron, told Bloomberg.

When it comes to memory modules, there is usually a tradeoff between performance and capacity. But the novel MCR DIMM concept introduced by SK hynix and Intel on Thursday promises to wed extreme performance and capacity. The memory maker promises that its MCR DIMMs will enable data transfer rates of over 8000 MT/s, while offering unprecedented capacities. 

SK hynix's Multiplexer Combined Ranks (MCR) DIMMs are dual-rank memory modules that make both ranks work simultaneously using a special buffer. Normally, modules with two physical ranks work like one module, therefore when the host CPU (or memory controller) fetches data from such a module, it can only fetch 64 bytes of data at a time. But the buffer developed by SK hynix, Intel, and Renesas allows two physical ranks to work like two modules in parallel, thus doubling performance by fetching 128 bytes of data from both ranks at the same time. 

The magic of the MCR technology is that both physical ranks (i.e., memory chips) of a dual-rank module continue to operate at more or less 'standard' clocks, which simplifies building high-capacity modules. Meanwhile, it is the Renesas multiplexer buffer that fetches 128 bytes of data from two modules and works at an 8000 MT/s or higher data transfer rate with the host CPU memory controller, again simplifying building high-speed modules. 

All buffers tend to increase latency and consume power, so these are the things that will have to be addressed both on the system and on the module level. Meanwhile, to make MCR DIMMs work smoothly, the technology has to be supported by the host CPU. So it is impossible to throw in an MCR DIMM into an existing machine hoping to get higher performance and higher capacity. 

"For a stable performance of MCR DIMM, smooth interactions between the data buffer and processor in and out of the module are essential," said Sungsoo Ryu, SK hynix's Head of DRAM Product Planning. 

Intel says that its future Xeon processors will support MCR DIMMs, but it does not disclose which CPUs will support the technology or when. Although the company has reportedly been working on this tech for a while. So one would hope the groundwork has been laid.

"The technology brought forward comes from years of collaborative research between Intel and key industry partners to produce significant increases in deliverable bandwidth for Intel Xeon processors," said Dimitrios Ziakas, Vice President of Memory and IO Technologies at Intel. "We look forward to bringing this technology to future Intel Xeon processors and supporting standardization and multigenerational development efforts across the industry." 

For now, MCR DIMM is largely a concept technology. SK hynix has confirmed that its MCR DIMMs can "operate at the data rate of minimum 8 GT/s." SK hynix has plans to bring the product to the market, but it does not say when this happens. Considering that the DRAM maker does not disclose any timeframes, it is hard to make predictions about the capacity of MCR DIMMs. But keeping in mind the capabilities of DDR5 SDRAM and the fact that MCR DIMMs are likely at least a couple of years away, we are probably talking about 1TB ~ 2TB modules. 

Due to rapidly deteriorating demand for commodity memory and dropping prices, Micron is cutting down wafer starts for 3D NAND and DRAM immediately by 20% compared to the previous quarter. The company now expects its 3D NAND bit output to grow insignificantly in the next calendar year, while its DRAM bit output will be reduced in 2023. 

In a bid to address slowing demand for 3D NAND and DRAM memory, Micron is reducing wafer starts by approximately 20% compared to the fourth quarter of fiscal 2022, which ended on Sept. 1, 2022. The reductions will be made across all technology nodes that Micron uses for high-volume production, so essentially the company is cutting down output of virtually all types of its products.  

Micron's ongoing first quarter of fiscal 2023 ends in early Dec., so cutting down wafer starts today will hardly have a meaningful impact on the company's results for the quarter or the market. Since production and testing/packaging cycles of both 3D NAND and DRAM are pretty lengthy, the market will feel the effect of Micron's cuts in a few weeks. Meanwhile, spot prices might react to Micron's announcement sooner. 

But the current wafer starts cut will have an effect on the company's output for the whole fiscal 2023 as Micron expects its DRAM bit output to drop and its 3D NAND bit output to grow in the 'single-digit percentage range.'  

Micron initiated production of 232-layer 3D NAND memory this summer and started making LPDDR5X memory on its 1β (1-beta) fabrication process earlier today. Both new production nodes will enable Micron to cut down its costs and increase bit output, but the company warned in late Sept. that it would slow ramp up of 232L 3D NAND and 1β DRAM production in a bid to limit bit output. Micron's 232-layer 3D NAND devices with a 2,400 MT/s interface are set to enable the fastest SSDs with a PCIe 5.0 x4 interface that will outshine currently available best SSDs with a 12.4 GB/s sequential read speed. 

Micron also disclosed in Sept. that its capital expenditures in fiscal 2023 would total approximately $8 billion, down 30% from fiscal 2022. The cuts will primarily concern procurement of new wafer fab equipment, which will slowdown adoption of the company's latest fabrication technologies. The company's construction CapEx was projected to more than double as the company is building up its new fab in Idaho. Today, the company said it was "working toward additional CapEx cuts" without elaborating. 

"Micron is taking bold and aggressive steps to reduce bit supply growth to limit the size of our inventory," said Sanjay Mehrotra, chief executive of Micron. "We will continue to monitor industry conditions and make further adjustments as needed. Despite the near-term cyclical challenges, we remain confident in the secular demand drivers for our markets, and in the long term, expect memory and storage revenue growth to outpace that of the rest of the semiconductor industry."

Since demand for computers and consumer electronics has weakened, memory sales have dropped as a result of inventory adjustments by PC manufacturers. Consequently, DRAM revenue has dropped by nearly 30% quarter-over-quarter in Q3 2022, an unprecedented turn of events. At the same time, Micron and SK Hynix slightly increased their market share in the third quarter, whereas Samsung and smaller suppliers dropped.

Sales of DRAM Drop by 28.9%

The DRAM industry earned $18.187 billion in revenue in Q3 2022, down 28.9% from $25.594 billion in Q2 2022, according to TrendForce. Contract prices for memory declined by 10% to 15% during the quarter, whereas spot prices experienced even deeper declines. Analysts from TrendForce claim that sales of all types of memory dropped, including DRAM for consumer electronics, PCs, and even servers, as producers adjusted their inventory due to slowing demand.

When it comes to the top memory supplier, Samsung retained its No.1 position as the global DRAM leader by revenue, though its sales declined by 33.5% quarter-over-quarter, whereas its market share dropped to 40.7% from 43.5% in Q2 2022. 

SK Hynix lost 25.2% of DRAM revenue QoQ but increased its market share to 28.8% in the third quarter. Micron's DRAM sales decreased by 23.3% sequentially in Q3 2022, but its share increased from 24.5% to 26.4%. In fact, Micron's DRAM declines were the lowest in the industry. 

By contrast, smaller makers like Nanya, Winbond, and PSMC experienced 37.4% - 40.8% declines in revenue and lost market share.

DRAM Makers to Cut Output

Among the ways for memory makers to stabilize the pricing is to limit DRAM bit output by either slowing down capacity expansion or slowing down the transition to more advanced fabrication technologies. Putting equipped fabs on hold is an extremely expensive measure that leads to giant losses as expensive fab equipment has to be depreciated, but decelerating the transition to new production nodes is a relatively easy way to limit DRAM bit output growth. 

Apparently, this is exactly what memory makers will do in the coming quarters. Samsung will deploy its new P3L fab next year and will raise its DRAM production capacity, but will slow its transition to newer nodes. SK Hynix will also slow down the transition to more advanced fabrication processes at different fabs. Micron, which just started production of LPDDR5X memory using its 1β node in Japan, will not aggressively ramp up production using this fabrication process. Furthermore, TrendForce believes that Micron could implement more aggressive ways of cutting down its DRAM output.  

Smaller makers of DRAM will follow the larger ones with their node transitions, but since they controlled around 4% of the market in Q3 2022, the impact of their actions on prices will be limited. 

Good for End Users?

Dropping memory prices are generally good for the end user (and, to some degree, PC makers), especially if you transition to DDR5 memory this year. But after DRAM makers cut their output, prices will inevitably rise when demand stabilizes or gets stronger, so perhaps it makes sense to plan your upgrade now.

Micron on Tuesday announced its next generation 1β (1-beta) fabrication technology for DRAM (dynamic random access memory). The new node will enable Micron to cut down costs of its DRAM while also increasing their power efficiency and performance. 1β (1-beta) will be the company's last DRAM production process that will rely on deep ultraviolet (DUV) lithography and will not use extreme ultraviolet (EUV) tools.

Micron's 1β fabrication process uses the company's 2nd generation high-K metal gate (HKMG) and is said to increase bit density of a 16Gb memory die by 35% as well as to improve power efficiency by 15% when compared to a similar DRAM device made on the company's 1α node. The new manufacturing technology will be particularly useful for high-capacity DDR5 and LPDDR5X chips for mobile, server, and desktop applications. 

Unlike its Samsung and SK Hynix rivals, Micron currently does not use EUV tools and therefore has to rely on various multi-patterning techniques to keep making DRAM cells smaller.

"We drove innovation from all fields, including cutting edge pattern multiplication techniques," said Thy Tran, vice president of DRAM process integration at Micron. "1β also comes with new processes, materials, and advanced equipment to advance our memory cell integration so that we can shrink the memory cell array. […] To maximize all the technology benefits and our design innovations, we are also aggressively scaling both the memory cell height in terms of size and also the rest of the circuitry in the die to save space and bring you the smallest die possible for a given density while optimizing for power and performance improvements."

The first product that Micron will make using its leading-edge node will be 16Gb LPDDR5X-8500 memory, but eventually the company will start using the node for other products. The 16Gb LPDDR5X chip is said to offer enhanced dynamic voltage and frequency scaling extensions core (eDVFSC) voltage control techniques for power savings.

LPDDR5X memory was designed not only for mobile applications like tablets and smartphones, but also to improve performance of various bandwidth-hungry applications like PC-class system-on-chips as well as artificial intelligence (AI) accelerators.

Micron's 16Gb LPDDR5X-8500 memory as well as its 1β fabrication process are ready for mass production, according to Micron. At present the company is shipping samples of its LPDDR5X-8500 DRAMs to interested parties and will initiate mass production of these ICs once they pass qualification procedures.

As demand for PCs and smartphones slumps, so does demand for 3D NAND and DRAM memory. As a result, companies like SK Hynix suffer from dropping revenue and dramatic declines of profits. According to reports via Reuters and Nikkei Asia, in a bid to balance the books, SK Hynix plans to halve its capital expenditures next year and focus on the manufacture of more expensive types of memory. The company also plans to assess the future of its fab in China.

SK Hynix Cuts CapEx Due to Major Profits Fall

Just like other semiconductor companies, the memory maker believes that demand for chips will be sluggish for several quarters and supply will exceed demand, which means tiny profits amid lowering revenue. To that end, SK Hynix will reduce its CapEx investment next year by more than 50% year-over-year. The company does not disclose how much will it spend on new fabs and tools in 2023, but only says that it would be 'at the upper range of 10 – 20 trillion won ($7 billion – $14 billion). 

SK Hynix has good reasons to cut down its expenses. This week the company posted a 10.983 trillion won ($7.741 billion) revenue, which was down 7% compared to the same quarter a year ago. A 7% YoY drop is hardly significant, but the company also reported operating profit of 1.656 trillion won ($1.167 billion) in operating profit (a 60% year-over-year decline), and net income of 1.103 trillion won ($777.396 million) in the second quarter of 2022 (a 67% YoY decrease).

Reduce Production, Focus on Premium Segments

In addition to reducing its CapEx, SK Hynix will increase production of its premium products, which includes 238-layer 3D NAND memory that can be used to power the fastest SSDs with a PCIe 5.0 x4 interface. Also, the company intends to boost production of memory using its 1a-nm fabrication technology in general as well as premium DDR5, LPDDR5, and HBM3 for high-end applications like servers, AI, and analytics accelerators. 

In addition, the company intends to lower production of commodity 3D NAND and DRAM memory in a bid to balance supply and demand on the market. Meanwhile, the company does not disclose how significantly it plans to cut its output.

Fab Dilemma

Back in September SK Hynix began to construct its $11 billion fab in South Korea that is set to come online in 2025. It remains to be seen whether the company will be able to build the new production facility on time with its reduced CapEx budget, but it looks like it might face a dilemma with its spendings on new manufacturing capacity. 

Building a brand new fab is by definition more expensive than upgrading existing production facilities. SK Hynix has fabs in China and South Korea. The Chinese fabs may be cheaper to operate (and the company needs to cut costs now) and can even be upgraded with new tools produced in the U.S. in the next 12 months as the company has a waiver to ship new equipment from America to China for a year. But if the company's partners fail to get an export license from the U.S. Department of Commerce to ship new tools and spare parts to SK Hynix's facilities in Wuxi and Dallian (which is formally controlled by SK Hynix's subsidiary Solidigm), these fabs will become obsolete in a couple of years (i.e., 2025 – 2026). 

Given the risks associated with its Chinese fabs, it might make sense for SK Hynix to focus on investments in its all-new fab in South Korea. At present the company is considering several contingency plans for the facilities, some of which include transferring equipment to South Korea, reports Nikkei. 

"As a contingency plan, we are considering selling the fab, selling the equipment or transferring the equipment to South Korea," said Kevin Noh, chief marketing officer for SK Hynix, at the earnings call this week. "It's a contingency plan. We want to [continue to] operate without facing this situation. We expect to extend it each year, but it is not for sure. It is very uncertain."

After the U.S. imposed sweeping sanctions against China's semiconductor industry that prohibit shipments of advanced chip production equipment to the People's Republic of China, South Korea-based Samsung and SK Hynix got a one-year waiver to keep upgrading their fabs in China.

Samsung's and SK Hynix's fabs in China produce a substantial portion of the global 3D NAND and DRAM supply, reports Nikkei. Samsung makes some 40% of its 3D NAND chips in China, whereas SK Hynix produces around 40% of its DRAMs in the People's Republic. Normally, both companies installed the latest production equipment at their fabs and adopted leading-edge fabrication technologies there to be competitive. 

Under the new rules U.S.-based companies cannot ship tools that can be used to make 3D NAND with 128 or more layers as well as DRAM using a 18nm or more sophisticated production technology. Most products of Samsung and SK Hynix are made on even more advanced nodes. 

Both Korean companies will still be able to add advanced American tools to their operations in China for a year, but after that their suppliers will have to either get a new export license from the U.S. Department of Commerce to sell equipment to Chinese subsidiaries of South Korean companies, or stick to a case-by-case review scenario when they will have to get a different license for each deal. 

A source familiar with Samsung's matters could not confirm what would happen after the year is out. 

Such case-by-case licensing represents huge risks both for Samsung and for SK Hynix as if they are not able to procure the right production tools, they will not be able to transit to new fabrication processes and therefore they will be less competitive against their rivals (Micron, Kioxia, Western Digital) as new nodes lower per-bit production costs and increase performance. 

Most of products made by Samsung and SK Hynix in China are sold to local makers of electronics, including big PC makers like Apple and Lenovo, so if the two South Korean companies have to leave China, this will hit global suppliers. 

In a bid to assist its two major semiconductor companies the Korean Ministry of Trade, Industry and Energy had talked with the U.S. Department of Commerce's Bureau of Industry and Security on behalf of Samsung and SK, according to Nikkei. While this certainly increases Samsung's and SK Hynix's abilities to get appropriate export permissions, this does not mean a 100% probability of success.  

As a result, both Samsung and SK Hynix are now facing a dilemma whether to keep investing in their Chinese operations or gradually withdraw from the country and shift production of memory in South Korea, Taiwan, Japan or the U.S.

Asus is prepping at least five motherboards based on Intel's range-topping Z790 chipset that will support DDR4 memory. In addition, the motherboards will support Intel's flagship 13th Generation Core 'Raptor Lake' processors and users will be able to keep their existing high-performance memory modules. 

Asus plans to release two TUF Gaming (TUF Gaming Z790-Plus D4 and TUF Gaming Z790-Plus WiFi D4) as well as three Prime-series (Prime Z790-P D4, Prime Z790M-Plus D4, and Prime Z790-P WiFi D4) motherboards powered by Intel's Z790 chipset that will support DDR4 memory, based on an EAC filing discovered by @momomo_us. Of course, filings with EAC do not always guarantee that one product will be released, but three out of five motherboards are on the list of South Korea's RRA regulator (1, 2, 3), so Asus is indeed prepping DDR4-supporting Intel Z790 platforms. 

At reasonable prices, Asustek's TUF Gaming motherboards cater to demanding gamers who want performance, reliability, and overclockability. While TUF Gaming mainboards sit below the company's ROG platforms in the company's hierarchy, these are still relatively advanced motherboards with premium features. 

The Asus Prime family sits below TUF Gaming but still offers premium features, enhanced voltage regulating modules (VRMs) to enable overclocking, and various tuning capabilities. So it is a natural fit for the Prime Z790 family to include advanced LGA1700 motherboards that support DDR4 memory sticks. 

Like it or not, DDR4 memory's days are numbered, so Intel prefers to talk about DDR5 SDRAM even though its current 600-series and upcoming 700-series platforms support DDR4 and DDR5. But motherboard makers tend to offer what their customers want, and many want to continue using good-old DDR4 for a while, which is why both Asus and MSI will provide Z790-based platforms with DDR4. 

DDR4 memory has been around for seven years, and producers of high-end memory modules have sold a boatload of enthusiast-grade DDR4 modules with data transfer rates in DDR5's territory (yet with lower latencies and better performance). Throwing those modules away (or selling them at a bargain price) does not make sense, so keeping them for a couple of years is undoubtedly reasonable. 

The biggest question about midrange Z790 motherboards for Intel's Raptor Lake processors is whether they will adequately support the rumored 350W mode enabling ultimate turbo frequencies and therefore requiring an advanced voltage regulating module to deliver immaculate power to the CPU. TUF Gaming and Prime are still premium motherboards, but they are not as good as the Asus ROG family. It will be exciting to see how different Intel Z790 platforms perform.

According to AMD, Intel, and the South Korean trade ministry, sales of processors and memory for client PCs are dropping faster and more significantly than expected. Media reports say that shipments of memory were down nearly 25 percent last month because of softening demand in Europe and the U.S., as well as lockdowns in China.

Just a little less than a month ago, IDC said that shipments of PCs will drop to 305.3 million units in 2022, down 12.8% from 348.8 million units in 2021. The revised PC shipments forecast for 2022 was primarily based on reports from PC makers and sales forecasts of companies like AMD and Intel. But the situation got worse than expected pretty quickly. 

Earlier this year, Intel expected sales of PCs to drop around 10% year-over-year and start to rebound already in the second half of 2022. AMD was a little more pessimistic and forecasted the PC market to be down 'midteens percent' in 2022 compared to the previous year. But executives from both companies told Stacy Rasgon, an analyst with Bernstein, that the market was trending worse than expected, reports Barrons. 

CPU sales are an important indicator of PC market health. The earnings of memory suppliers are another good indicator of what to expect from the PC market. Shipments of DRAM fell 24.7% in August compared to the same month a year ago because of lower demand for PCs in Europe and the U.S. and COVID lockdowns in China. By contrast, sales of South Korean DRAM in July dropped 7% YoY, reports DigiTimes citing data from Bloomberg and the South Korean trade ministry.  

This is bad news for South Korea as the country's tech exports account for one-third of the country's exports. DRAM accounts for about half of Korean memory chip sales, and 3D NAND accounts for another half. In general, South Korean tech export dropped 4.6% year-over-year in August. 

South Korea-based Samsung and SK Hynix control about two-thirds of the global 3D NAND and DRAM memory markets, so if they have problems selling their chips, it is a clear indicator that demand is dropping. What remains to be seen is whether the two companies will lower their 3D NAND and DRAM pricing or reduce output to keep prices at comfortable levels.  

Memory price reductions could be good for the ongoing transitions to DDR5 memory and for costs of SSDs featuring a PCIe 4.0 x4 and PCIe 5.0 x4 interfaces. But this will affect the abilities of companies like Samsung, SK Hynix, and Micron to invest in future products and process technology transitions, which is not particularly good for the long-term. 

Intel’s Z690 platform introduced DDR5 to the desktop market, but initial DDR5 modules were quite expensive and did not quite live up to expectations regarding performance. But as high-speed DDR5 kits get cheaper, the new type of memory gets more attractive, which is why we expected most of Intel’s Z790-based platforms to feature DDR5 rather than DDR4. We were wrong. MSI is prepping a host of Z790-based motherboards with good-old DDR4 memory.

MSI readies Z790 Edge DDR4, Z790 Tomahawk DDR4, and Z790 Pro DDR4 motherboards that will combine Intel’s newest chipset and DDR4 memory, according to manuals published by Overclock.net (and discovered by momomo_us). These main boards will support up to DDR4-5333+ memory modules in one DIMM per channel configuration and up to DDR4-4400+ sticks in two DIMMs per channel setup. However, a lot will depend on CPU capabilities and processor cooling systems.

MSI’s Edge, Tomahawk, and Pro motherboards are not precisely the company’s top-of-the-range offerings. So it raises the question of whether they will be good enough to squeeze all the juices out of Intel’s 13th Generation Core Raptor Lake processors as far as overclocking is concerned.

One of the key features of unlocked Raptor Lake CPUs will be their 350W mode enabling ultimate turbo frequencies and therefore requiring an advanced voltage regulating module to deliver immaculate power to the CPU and a sophisticated cooling system both for the CPU and its VRM. Budget motherboards are pretty good these days but are not as good as flagship models with their expensive VRMs and advanced cooling systems for power circuitry.

 Using good-old DDR4 with an all-new Intel platform sounds like a good idea, especially if you own those high-end DDR4-5333+ kits. With such memory, you are going to get outstanding out-of-box performance. But with cheaper motherboards, you may not be able to take full advantage of one of Raptor Lake’s key features on the desktop; whether the trade-off is worth something, everyone has to decide for themselves.