Budget-oriented hardware doesn’t always receive its fair share of attention. In fact, most of the time, it's pretty difficult to get samples to review from companies like Intel and AMD, which don't want to see their lower-end hardware maligned. Thankfully, many of Intel’s newer Ivy Bridge-based models are fairly affordable, starting at a dirt-cheap $35. So, we stopped into our local Micro Center retail store and bought three CPUs to create today's comparison. And we chose to pit these newest Celeron, Pentium, and Core i3 chips against former favorites that no longer get dusted off for testing.
If you've spent much time reading Tom's Hardware, then you're probably very familiar with Intel's LGA 775 interface. In fact, if you built or bought a PC between 2006 and 2008, it probably sported a Core 2 Duo or Quad CPU that dropped into LGA 775. Perhaps, like me, you constructed a gaming rig to take on 2007's big hit, Crysis. This legendary Chuck Norris of PC games convinced me that my old single-core AMD Athlon 64 FX-55 gaming rig just wasn’t going to cut it any longer. Amidst the dual- versus quad-core debates, I jumped past Intel's Conroe-based Core 2 Duo and went straight to a Core 2 Quad Q6600 (Kentsfield) with a G0 stepping. Tamely overclocked to 3.0 GHz, this CPU is still a fixture in my home, even today.
Not long after, Intel adopted 45 nm manufacturing. Once Wofldale-based processors hit the mainstream scene, it was hard to argue against a highly-overclockable Core 2 Duo E8400 for gaming. On the flip side, many power users preferred the overall performance of newer quad-core models, such as the Yorkfield-based Core 2 Quad Q9550.
Unlike AMD’s Athlon II and Phenom II families, which are still available and show up in our stories, Intel's LGA 775 platform got shoved aside in favor of newer architectures, which became the first-, second-, and third-generation Core processors. Some of our readers noticed the absence of those LGA 775-based CPUs in our tests, though, and we agree that it's time to revisit their performance in a more modern benchmark suite. So, today's story is for anyone holding on to an older Core 2 platform (or anyone who enjoys comparisons to modern offerings five years later).
Wolfdale And Yorkfield Take On Ivy Bridge
Core 2 Duo E8400 and Core 2 Quad Q9550 delivered solid clock rates right out of the box, but they're also best remembered for their substantial overclocking headroom. It was easier to keep two cores cool, so the Duos tended to handle higher voltages better, scaling higher in the process. Chips like ours, based on the E0 stepping, hit frequencies 30 to 50% higher on air cooling. So, we weren’t about to shy away from overclocking in this piece. Of course, we’ve come to expect high efficiency and per-clock performance from Intel’s Ivy Bridge-based processors, but we also know this great performance steered Intel towards locking the multiplier ratios of its lower-end offerings, a slap in the face for many enthusiasts, and a huge blow to gamers on a tight budget. Might high frequencies propel Intel’s old Core 2 architecture up beyond the latest locked-down Ivy Bridge-based chips?
Because this piece centers on data, we’re eager to hit the performance charts. But first we should introduce today’s contenders. Our trio of dual-core chips includes the Celeron G1610, Pentium G2020, and Core 2 Duo E8400. The Core i3-3225, equipped with Intel’s Hyper-Threading technology, is the oddball of the bunch, sporting two physical cores capable of addressing two threads each. It goes up against the Core 2 Quad Q9550, armed with four physical cores. As a control, we also include the quad-core Core i5-3570K, a modern enthusiast favorite and the only chip we're testing considered a worthy upgrade to the Core 2 Quad.
| Cores / Threads | Intel LGA 1155 Interface | Intel LGA 775 Interface |
|---|---|---|
| 2 / 2 | Intel Celeron G1610 (Ivy Bridge), 22 nm, 2.6 GHz, 2 MB Shared L3, Launched Q1 2013, Box: $42 | Intel Core 2 Duo E8400 (Wolfdale), 45 nm, 3.0 GHz, 1,333 MT/s FSB, 6 MB L2 Cache, Launched Q1 2008, Box: $179 |
| Intel Pentium G2020 (Ivy Bridge), 22 nm, 2.9 GHz, 3 MB Shared L3, Launched Q1 2013, Box: $64 | ||
| 2 / 4 | Intel Core i3-3225 (Ivy Bridge), 22 nm, 3.3 GHz, 3 MB Shared L3, Hyper-Threading, Launched Q4 2012, Box: $134 | |
| 4 / 4 | Intel Core i5-3570K (Ivy Bridge), 22 nm, 3.4 GHz (3.8 GHz Turbo), 6 MB Shared L3, Launched Q2 2012, Box: $235 | Intel Core 2 Quad Q9550 (Yorkfield), 45 nm, 2.83 GHz, 1,333 MT/s FSB, 12 MB L2 Cache, Launched Q1 2008, Box: $287 |
If you want to refresh your knowledge of the architectures we're testing, have a look back at Wolfdale Shrinks Transistors, Grows Core 2 and Intel Core i7-3770K Review: A Small Step Up For Ivy Bridge.
| System Test Configurations | |
|---|---|
| LGA 1155 Platform | |
| CPU | Intel Celeron G1610 (Ivy Bridge), 22 nm, 2.6 GHz, 2 MB Shared L3 |
| Intel Pentium G2020 (Ivy Bridge), 22 nm, 2.9 GHz, 3 MB Shared L3 | |
| Intel Core i3-3225 (Ivy Bridge), 22 nm, 3.3 GHz, 3 MB Shared L3, Hyper-Threading technology | |
| Intel Core i5-3570K (Ivy Bridge), 22 nm, 3.4 GHz (3.8 GHz maximum Turbo Boost), 6 MB Shared L3 | |
| CPU Cooler | Intel Boxed Cooler |
| Motherboard | Intel DZ77GA-70K, Intel Z77 Express, BIOS 0049 (07-13-2012) |
| RAM | 8 GB (2 x 4 GB) Crucial PC3-12800 kit Celeron / Pentium: DDR3-1333, CL 8-8-8-24 at 1.5 V Core i3 / Core i5: DDR3-1600, CL 8-8-8-24 at 1.5 V |
| LGA 775 Platform | |
| Intel Core 2 Duo E8400 (Wolfdale), 45 nm, 3.0 GHz, 1,333 MT/s FSB, 6 MB L2 Cache Overclocked @ 4.0 GHz (445 x 9), 1,780 MT/s FSB, 1.272 V Idle/ 1.240 V Load Overclocked @ 4.5 GHz (500 x 9), 2,000 MT/s FSB, 1.416 V Idle, 1.384 V Load | |
| Intel Core 2 Quad Q9550 (Yorkfield), 45 nm, 2.83 GHz, 1,333 MT/s, 12 MB L2 Cache Overclocked @ 3.4 GHz (400 x 8.5), 1,600 MT/s, 1.240 V Idle/ 1.200 V Load Overclocked @ 3.7 GHz (435 x 8.5), 1,740 MT/s, 1.328 V Idle, 1.240 V Load | |
| CPU Cooler | Xigmatek HDT-S1283 120 mm air cooler |
| Motherboard | Asus Rampage Formula, Intel X48 Express, BIOS 0902 (04-28-09) |
| RAM | 4 GB (2 x 2 GB) Corsair PC2-8500 kit DDR2-1066, 5-5-5-15 at 2.1 V DDR2-890 & DDR2-1000 for E8400 Overclocking DDR2-1066 & DDR2-1045 tRD 07 for Q9550 Overclocking |
| Common | |
| Graphics | Sapphire Radeon HD 7970, 950 MHz GPU, GDDR5-5700 |
| System Drive | Samsung 840 Pro 256 GB, SATA 6Gb/s SSD |
| Power | Corsair Professional Series HX1050, 1050 W, 80 PLUS Silver |
| Software and Drivers | |
| Operating System | Windows 8 Professional x64 |
| Graphics Driver | AMD Catalyst 13.2 Beta 7 |
| Platform Driver | Intel 7-series Inf v. 9.3.1026 |
| Benchmark Configuration | |
|---|---|
| 3D Games | |
| Borderlands 2 | Version 1.0.28.69606, DirectX 9, Custom Run, Fraps Test Set 1: Medium Quality Settings, Low PhysX, 8x AF Test Set 2: Highest Quality Settings, Low PhysX, FXAA,16x AF |
| Crysis 3 | Version 1.0.0.2000, Direct X 11, Custom Run, 60-Sec. Fraps Test Set 1: Lowest Quality Settings, No AA, 1X AF Test Set 2: Medium Quality Preset, FXAA, 8x AF Test Set 3: Very High Quality Preset, 2x SMAA, 16x AF |
| The Elder Scrolls V: Skyrim | Version 1.8.151.0.7, Custom Run, 25-Sec. Fraps Test Set 1: High Preset, No AA, 8x AF, FXAA Enabled Test Set 2: Ultra Preset, 8x AA, 16x AF, FXAA Enabled |
| F1 2012 | Version 1.2, Direct X 11, Built-in Benchmark Test Set 1: High Quality, No AA Test Set 2: Ultra Quality, 8x AA |
| Far Cry 3 | V. 1.04, DirectX 11, 50-sec. Fraps "Amanaki Outpost" Test Set 1: High Quality, No AA, Standard ATC., SSAO Test Set 2: Ultra Quality, 2x MSAA, Enhanced ATC, HDAO |
| Hitman: Absolution | V. 1.0.446.0, DirectX 11, Built-in Benchmark Test Set 1: Medium Quality Preset, No MSAA, 2x AF Test Set 2: Ultra Quality Preset, 2x MSAA, 16x AF |
| StarCraft II: Heart of the Swarm | Version 2.0.6.25180, Custom Run "Harvest of Screams" Campaign Mission, 60-Sec. Fraps Test Set 1: High Preset, No AA, 8x AF, FXAA Enabled Test Set 2: Ultra Preset, 8x AA, 16x AF, FXAA Enabled |
| Tomb Raider | Version 1.00.722.3, Direct X 11, Custom Runs, "Chasm Monastery", "Mountain Village", 45-Sec. Fraps Test Set 1: High Quality Preset Test Set 2: Ultimate Quality Preset |
| Audio/Video Encoding | |
| HandBrake CLI | Version: 0.98, Video: Video from Canon Eos 7D (1920x1080, 25 frames) 1 Minutes 22 Seconds, Audio: PCM-S16, 48,000 Hz, Two-Channel, to Video: AVC1 Audio: AAC (High Profile) |
| iTunes | Version 10.4.1.10 x64: Audio CD (Terminator II SE), 53 minutes, default AAC format |
| Lame MP3 | Version 3.98.3: Audio CD "Terminator II SE", 53 min, convert WAV to MP3 audio format, Command: -b 160 --nores (160 Kb/s) |
| TotalCode Studio 2.5 | Version: 2.5.0.10677, MPEG2 to H.264, MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG2), Audio:MPEG2 (44.1 kHz, Two-Channel, 16-Bit, 224 Kb/s) Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV |
| Abobe Creative Suite | |
| Adobe After Effects CS6 | Version 11.0.0.378 x64: Create Video, Three Streams, 210 Frames, Render Multiple Frames Simultaneously |
| Adobe Photoshop CS6 | Version 13 x64: Filter 15.7 MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates |
| Adobe Premiere Pro CS6 | Version 6.0.0.0, 6.61 GB MXF Project to H.264 to H.264 Blu-ray, Output 1920x1080, Maximum Quality |
| Adobe Acrobat X Pro | Version 10.0.0.396: Print PDF from 115 Page PowerPoint, 128-bit RC4 Encyption |
| Productivity | |
| ABBYY FineReader | Version 10.0.102.95: Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages |
| Autodesk 3ds Max 2012 | Version 14.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080 |
| Blender | Version 2.64a, Cycles Engine, Syntax blender -b thg.blend -f 1, 1920x1080, 8x Anti-Aliasing, Render THG.blend frame 1 |
| Compression | |
| 7-Zip | Version 9.28, LZMA2, Syntax "a -t7z -r -m0=LZMA2 -mx=5" Benchmark: THG-Workload-2012 (1.3 GB) |
| WinRAR | Version 4.2, RAR, Syntax "winrar a -r -m3" Benchmark: THG-Workload-2012 (1.3 GB) |
| WinZip | Version 17.0 Pro, Syntax "-a -ez -p -r" Benchmark: THG-Workload-2012 (1.3 GB) |
| Synthetic Benchmarks and Settings | |
| 3DMark 11 | Version: 1.0.1, Performance Suite |
| PCMark 7 | Version: 1.0.4, System, Productivity, Hard Disk Drive benchmarks |
| SiSoftware Sandra 2013 | Version: 2013.01.19.11, Processor Arithmetic, Cryptography, Memory Bandwidth Benchmarks |
We're starting out with 3DMark 11, concentrating on the Physics (a processor-oriented metric) and Combined (which adds a graphics workload) tests. Because 3DMark employs one thread for each physical and logical core, the Core i3’s Hyper-Threading feature propels it far beyond the dual-core processors. Its higher clock rate also allows it hold off the stock Q9550. But once we remove that advantage through overclocking, the old Core 2 Quad jumps back into the lead.
As expected, our Core i5 control sits unchallenged at the top, benefiting from Ivy Bridge’s great per-clock performance, and Turbo Boost frequencies of 3.6 GHz for all four physical cores.


Sandra 2013 Arithmetic and Cryptography yield the same basic finishing order as 3DMark 11. So far, these synthetic tests give us an early indication of a few battles, just as we hoped. It appears, clock-for-clock, four physical Wolfdale cores have greater performance potential than Core i3’s dual-core/quad-thread design. Not all software is as well-threaded though, so we'll have to see how much Core i3 benefits from the newer platform and far greater memory bandwidth.
The dual-core battle is just as interesting. There’s no doubt that Ivy Bridge offers better per-clock performance. But is 4.0 GHz enough for Core 2 Duo E8400 to surpass the 2.9 GHz Pentium G2020? Meanwhile, there may be a heated battle at the bottom with the Celeron G1610 holding a slight lead over the stock E8400.
Sandra Arithmetic also touts the Core i5's great performance potential, while inclusion of AES-NI support distances the CPU even further in Cryptography.


Intel's Ivy Bridge architecture establishes an enormous advantage in memory bandwidth. Thankfully, the company enables official support for DDR3-1333 all the way down its product line, including Celerons based on the design. In contrast, Sandy Bridge-based 600-series Pentiums and Celerons are limited to DDR3-1066. Our pair of third-generation Core processors provides an additional bandwidth boost thanks to DDR3-1600 XMP memory settings.
Our LGA 775-based platform is limited to DDR2-1066, but we also saw very little additional bandwidth when we tried out a DDR3-capable board. Any performance advantage we observed showed up at more aggressive overclocks, not at the stock CPU or memory frequencies. Most owners of processors this old went the DDR2 route, since it was a lot more affordable than DDR3 back then.
Single-threaded iTunes and LAME give us a closer look at per-clock performance, and it appears the Core 2 architecture requires between 17-21% (500-700 MHz) higher frequency to match Ivy Bridge. The Celeron G1610 would be roughly equivalent to a 3.16 GHz E8500. Overclocked to 4.0 GHz, Core 2 Duo should outrun any current frequency-locked Ivy Bridge-based Pentium or Celeron, at least in these audio encoders.


However, Intel’s Ivy Bridge architecture shines brighter in HandBrake and Total Code Studio video encoding. The Celeron G1610 makes quick work of a stock Core 2 Duo E8400, while the Pentium G2020 is able to trades blows with the overclocked E8400, despite an 1,100 MHz handicap.


Core i3-3225's dual-core configuration is no match for Core 2 Quad Q9550's four physical cores in HandBrake, but its logical cores and/or memory bandwidth are put to better use in Rovi TotalCode Studio.
It appears that Photoshop CS6 doesn’t effectively utilize Core i3’s logical cores, as our Core i5 control, with its quartet of slightly higher-clocked physical cores, cuts image processing times in half. This spells an easy victory for the stock Core 2 Quad Q9550, though the same doesn’t hold true in Premiere and After Effects.




Acrobat X, being single-threaded, yields results consistent with iTunes and Lame, though with a larger advantage for the Ivy Bridge-based platform.
Core i3’s logical cores are also no match for Core 2 Quad’s physical cores in our threaded productivity applications. The Q9550 earns victories across the board, despite lower clock rates.



The current-generation frequency-locked dual-core Ivy Bridge-based chips trail far behind 2008’s enthusiast-tweaked Core 2 Duo based on the Wolfdale design. While these tasks beg for more processing cores, this thrashing is still a bit hard to swallow.
In WinZip, WinRAR, and 7-Zip file compression, the Core i5-3570K continues to reign supreme at the top, while Celeron G1610 and Core 2 Duo E8400 trade blows down at the bottom. Core i3-3225 squeaks out a narrow second-place finish in overall compression time.



We used Fraps to measure the frame rates for each of today’s games, and will focus purely on 1920x1080, a popular resolution well within the Radeon HD 7970’s capabilities. We’re most concerned about the highest playable settings, but also ran less demanding graphics details to better flesh out CPU limitations and scaling.
We're plotting Average and Minimum Frames Per Second (FPS) in bar charts, and Frame Rate Over Time in line graphs. But for this CPU-related story, we are not logging Frame Times. Nvidia’s Frame Capture Analysis Tools (FCAT) are in heavy rotation across two Tom's Hardware labs, and are an integral part of multi-GPU stories like AMD Radeon HD 7990: Eight Games And A Beastly Card For $1,000. It's not as imperative for single-GPU coverage though, since the data recorded by Fraps and FCAT come much closer to being the same. Nevertheless, I chose not to log frame times with Fraps until a few other issues can be investigated. More than anything, I wanted to be sure that any frame time variance recorded in Fraps was a direct a result of the processor, and would occur during normal gaming, rather than just overhead from the software collecting data.
Subjectively, I’ll come out and say that there wasn't a single case where acceptable performance was hurt by noticeably inconsistent frame times simply by swapping CPUs. I can’t rule out the likelihood that frame latency might have impacted the fluidity of already sub-par frame rates, though.
This phenomenon did occur in Battlefield 3, but since it was equally evident with every processor, and carried over to both platforms, it was most likely attributable to the graphics card and drivers. At the Ultra detail preset, frame rates never dropped below 60 FPS, and both averages and minimums were within 3 FPS of each other for all tested processors. Even so, the game still didn’t appear fully smooth. I chose not to use Battlefield 3 data in today's story simply because there was almost no CPU scaling in our normal 90-second single-player benchmark routine. And for a story like this, you're going to worry more about the multiplayer experience anyway.
But don’t worry. On the following pages, we’ll explore data for eight other games, including many of the latest heavy-hitters. If anything, we could argue this test suite is overly brutal, but the idea is to push each processor to its limit, gauging current, and, if possible, future demands.
We start with our Tom’s Hardware Borderlands 2 manual run, which takes approximately 105 seconds to complete.

I've never been a fan of charting absolute minimum frame rate figures alone, since those floors may not be apparent during game play. Rather, I'm more concerned with sustained minimums and perceived smoothness (though these don't necessarily guarantee a smooth and enjoyable experience, as we’ve seen from dual-GPU configurations). I see minimums as most useful if they don't drop below a defined level, as opposed to a single dip at one point in time.
Charting frame rates over time in a line chart helps quite a bit in this area. But it also exposes an anomaly in how Fraps measures minimum FPS. For the Celeron G1610, our log shows the lowest second measured 34 FPS, while all other seconds were well above 40 FPS. But because Fraps reports a minimum of 29 FPS, we have to assume that absolute minimums are either calculated based on frame times or measured at far more frequent intervals than once per second.
This isn’t a fluke, it isn’t limited to one processor or platform, and it isn't exclusive to just this story. In fact, if the reported minimums and frame rate over time charts appear to match, then it's probable that the minimum was adjusted to match the frame rate over time log. Usually, we see a difference of one to four frames per second. But occasionally, the delta is even higher. We aren't going to make that adjustment in today's numbers, but keep this in mind as you read through today's analysis, and any other Fraps-based performance reporting moving forward. Bottom line: don't get too hung-up on the minimum FPS bar without first looking over the matching line graphs. Charting both lets you decide which number is more important.

Each processor delivers a nice playable experience at Medium settings. However, a couple of those downward spikes near the beginning of the run corresponded to a brief, yet still noticeable stutter.

The Radeon HD 7970 has no problem handling the highest quality settings in this DirectX 9 title, and CPU scaling remains abundantly clear. Core i5-3570K extracts the most potential from our graphics hardware, but every single one of these processors is at least capable of delivering an acceptable Borderlands 2 experience.
In a game supposedly optimized for four cores, it is interesting that the stock-clocked Core 2 Quad Q9550 only delivers about the same performance as a Celeron G1610. Also note that the Core i3-3225 is punished with a reported minimum of 32 FPS, though its frame rate over time log never drops below 42 FPS. In fact, the single lowest second logged overall for any of these processors was 37 FPS, yet half of them have reported minimums below 34.

We use a brutal sequence depicting a worst-case scenario for how these processors handle Crysis 3. The game is so CPU-dependent that we'll start by dropping down to the lowest quality before moving on to more interesting settings.


While you will certainly see higher performance in other areas of this game, you can’t bypass this sequence while playing through the single-player campaign. At the least-demanding settings, not one of our dual-core processors could come close to matching the stock Core 2 Quad Q9550 for playability. We’re not going to condemn dual-core CPUs as unfit based on one game alone, but if this doesn’t improve with a patch of some sort, it's bad news.


Again we need to point out differences between Fraps' reported minimum frame rate and the logged data. In fact, we used this exact data as an example to seek feedback from AMD and Nvidia. While every one of these processors varied from one to three frames per second, the Core i3-3225 experiences a huge 11 FPS difference. The slowest second logged 30 frames, yet the minimum was reported to be 19.
Subjectively, both the Core i5 and Core i3 deliver a fairly smooth experience at these settings. The Q9550 was very playable once we overclocked it, but noticeably slower at stock speeds. Repeating this run two to three times on each dual-core CPU was an almost painful experience.


While I won’t say that the Core i3 or 3.4 GHz Core 2 Quad were completely unplayable, the Core i5-3570K is the only processor I can whole-heartedly recommend for Crysis 3. Very High details demand expensive graphics hardware, and big graphics warrant the balanced platform only a fast quad-core CPU (or higher) can deliver.
F1 2012 is one of Codemasters’ popular DirectX 11 racing titles based on its own Ego Engine 2.0.


This game appears well-optimized for quad-core processors, allowing the 2.83 GHz Core 2 Quad to easily best a 4.0 GHz Core 2 Duo. With that said, all of our CPUs remain capable enough of playing through this title fluidly. There were no noticeable stutters and, apart from one dip, frame rates remained above 40 FPS.


Subjected to the Ultra detail preset, our dual-core processors only deliver half of the Core i5-3570K's performance, at best. There’s certainly a good reason to buy a quad-core or Hyper-Threaded dual-core CPU for this title, but F1 2012 is still playable with a respectable dual-core processor.
Next, we look at another threaded DirectX 11 title, Ubisoft’s Far Cry 3, based on the Dunia Engine 2.


I find this game easily playable when frame rates are in the 30s. But if I was gaming on the Core 2 Duo E8400, I’d want to capitalize on the extra performance attainable through overclocking.
Having ventured deep into the single-player campaign on a $500 System Builder Marathon PC (sporting a Pentium G850 processor and Radeon HD 7850 graphics), I found that pairing to be a good match at 1920x1080 using Very High quality settings. In fact, during the following quarter, there was no noticeable performance boost stepping up to Core i5 until I began overclocking the Radeon HD 7850.


However, just as we see here, that dual-core Sandy Bridge-based chip fell short under the increased demands of Ultra quality details. Basically, you are wasting your money buying graphics hardware capable of maximum visuals in this game unless you balance your card out with a processor able to juggle four threads. In this case, both the Core i3-3225 and Core i5-3570K are strong enough to shift the demands back over to our beefy Radeon HD 7970, even with just 2x MSAA enabled.
Next up is IO Interactive’s Hitman: Absolution, a DirectX 11 title based on the Glacier 2 graphics engine. We’re using the built-in benchmark routine for a worst-case look at how the massive crowd technology can punish our crop of processors. Keep in mind, in-game frame rates will generally be much higher, since most of the levels are far more confined than the densely-packed Chinatown map area.


The dual-core processors yielded such low results that I had to run through every one of the graphics quality presets, finding only a 5-6 FPS difference in average performance between Lowest and Ultra. In short, the built-in benchmark brutally punishes our dual-core processors, making them appear obsolete (never mind the fact that each is far above the game’s minimum requirement). I doubt that any of our contenders would have a hard time playing through this title at reduced settings, but Hitman is certainly best enjoyed with a quad-threaded processor.


While the lasting appeal of Blizzard’s popular StarCraft franchise is no doubt found within the multiplayer experience, I find the single-player campaigns well-designed, and always a worthy starting point. Rather than using our existing Wings of Liberty multiplayer map, I jumped into the Heart of the Swarm expansion and discovered that the "Harvest of Screams” mission was the first one capable of taxing my Core i5 gaming rig. This 60-second benchmark takes place as Kerrigan leads approximately 150 Zerg forces in to destroy the mission’s final Protoss base.
I purposely delayed my attack a couple of extra minutes to build up more Zergling than the mission required, plus I kept the game camera zoomed out and centered over the action. As a result, frame rates drop substantially as more and more units come into view, joining the battle. This may be considered overly brutal for your own style of play. After all, Core 2 Duo E6600 is the recommended processor requirement. But without a doubt, too little processor performance requires that you make compromises, whether you alter your strategy, zoom the camera in on fewer units, or totally avoid large-scale multiplayer maps.


StarCraft II is CPU-intensive, but unfortunately isn’t optimized for quad-core processors. Largely dependent on the amount of cache you can throw at it, Ivy Bridge-based processors appear to scale roughly 500-800 MHz ahead of the Core 2 architecture, leaving the Core 2 Quad Q9550 and Core 2 Duo E8400 at the bottom of our stack.


Cranking up graphics and texture quality for this second set of results yields frame rates close to how we'd expect to play this game with beefy graphics. Even so, it's clearly still CPU-limited.
The Elder Scrolls V: Skyrim remains a popular title thanks to the modding community and official downloadable expansion packs. So, while it’s not necessarily a demanding title, we think it remains a valid benchmark. During playability analysis, I’ll often pick a fight within the City of Markarth to grab a worst-case look at system performance, but the Tom’s Hardware’s standard 25-second run through Riverwood is almost as demanding, and far easier to repeat amongst editors.


Despite official quad-core CPU recommendations, Skyrim doesn’t effectively utilize more than two cores. What’s important is that all of these processors are fully capable of playing at our benchmark settings.


Pointing out the obvious, no frame rates logged over time drop below 40 FPS using Ultra details. Yet Fraps reports a minimum of 38 FPS for the Core 2 Duo E8400.
Core i5-3570K leaves the other processors in its dust, though each one delivers a solid maxed-out Skyrim experience, which is important considering the muscular graphics pairing.
Tomb Raider, one of AMD’s latest Gaming Evolved titles, is in my opinion one of the biggest hits thus far in 2013. It takes powerful graphics hardware to deliver playable performance using Ultimate quality settings, which enable realistic TressFX hair. Having dumped quite a bit of time into both playing the game and analyzing its performance, I felt it was beneficial to test two different game levels for today’s story.
Our 45-second custom run of the “Chasm Monastery” level contains a short cinematic that punishes graphics processors far more forcefully than the built-in benchmark, although mostly when TressFX hair is enabled. This manual run-through is easily repeatable and great for testing graphics cards.
Of course, as with many games, the hardware demands fluctuate from one map to the next. Outdoor areas encountered in the “Mountain Village” level offer a far better look at the game’s CPU demands. You’ll see less of this one on Tom’s Hardware, as it requires more user control and also overwrites the saved game, requiring somewhat tedious save slot juggling before each run. But used together, these two benchmarks provide a worst-case look at both the game’s CPU and GPU requirements.


Only normal hair effects are enabled at the High quality preset, and that flat area in the middle of our line graphs, where the dual-core processors receive a performance boost, is the cinematic sequence. All of our tested dual-core processors fly through this part of the game, yielding similar performance.


However, frame rates plummet once we step outdoors and overlook “Mountain Village”. This performance hit is most noticeable on the dual-core processors, and at times the Core 2 Duo E8400 makes it difficult to precisely control Lara’s maneuvers.


TressFX hair enabled by the Ultimate quality preset completely changes the flat cinematic portion of our run, and the mighty Radeon HD 7970 drops to 30 FPS, no matter the processor pairing. Once the camera zooms off of Lara, frame rates spike before control is handed back to the user. Similar cinematic sequences are unavoidable, and a big part of the game, which is why we’re taking the time to demonstrate this behavior within a CPU shootout.


Without a doubt, it takes powerful graphics hardware to crank out Ultimate details, but parts of this game really smack the processor as well. Game play is adversely affected by our two slowest dual-core chips; the Core i3 and overclocked Q9550 are about the least I’d want when playing though these areas of the game. But it’s the Core i5-3570K that earns respect for delivering far more consistent frame rates.
Cranking Things Up a Notch
We curbed the limits of our Core 2 overclocking to best-represent the maximum frequencies value-seekers with this hardware will probably be using long-term. But the enthusiast in me just couldn’t stop with little or no voltage increase to these E0-stepping chips, knowing I may never again set up this aged platform for testing. Plus, I had already put a considerable amount of tweaking time into this very hardware a few years back, and still had notes handy detailing how high my chips would go, and the voltages required to get there.


At 1.45 V in the BIOS, 1.416 V at idle, or 1.384 V under load, this Core 2 Duo E8400 is stable at 4.5 GHz. It’s seen a fair amount of testing at these settings, but I’d drop voltages and top out in the 4.2-4.3 GHz range for daily use. At this high of a front-side bus frequency, I limited tweaking to CPU clock speed, and didn’t maximize the RAM configuration, running DDR2-1000 @ 5-5-5-15 timings. The Common Performance Level was left at default tRD 09. My success in RAM tweaking on this platform has come with 1 GB sticks of CAS 4 DDR2-800, and with this particular 4 GB kit, voltage requirements have far outweighed any gains.
I don’t like to push Core 2 Quad voltages as high, but would quickly reach the limits of this fairly modest Xigmatek HDT-S1283 air cooler anyway. The Q9550 is stable at 3.7 GHz set to 1.3625 V in the BIOS, resulting in 1.240 V under load. Again for daily use, I’d drop voltage a tad and be in the 3.6-3.65 GHz range, while keeping an eye on temperatures during those hot summer months. This front-side bus results in DDR2-1045 data rates, and with a little bump in voltages, I was able to further tweak the Common Performance Level to tRD 07, boosting memory bandwidth up to 8.36 GB/s.

To be honest, I originally had no intention on publishing this data, and only ran a few applications just for my own fun and knowledge. But the deeper I got into this, the more the information seemed worthy of sharing. Pushed to 4.5 GHz, the Core 2 Duo E8400 is able to match the single-threaded performance of Core i5-3570K at its 3.8 GHz Turbo Boost frequencies. While we know our K-series Core i5 itself has plenty of headroom for higher clock rates, we are still impressed by how well our four-year old E0 stepping Wolfdale-based scales. While the Core 2 Quad's gains are less impressive, we are eager to jump to the next page and see the impact during gaming.
Further Overclocking: 3D Game Performance
We’ll leave most of the gaming analysis up to you, but we do have a few observations to point out along the way. While absolute minimum FPS is included, remember that those data points alone might not be accurate due to Fraps' reporting mechanism.


We were able to push the Core 2 Duo E8400 high enough to play Crysis 3’s most demanding levels at the lowest details, matching the frame rates of a 2.83 GHz Core 2 Quad. Meanwhile, the tweaked Q9550 caught up to our Core i5 control, at least until we bumped things up to Very High quality settings.


Oddly, the Core 2 Quad Q9550 fails to pick up any additional performance in F1 2012, despite 300 MHz-higher clocks and a boost in memory bandwidth.


The overclocked Q9550 just about catches Core i3-3225 in Far Cry 3, and surpasses it in Hitman: Absolution.




This Core 2 Duo E8400 has enough headroom to catch Core i3-3225 in The Elder Scrolls V: Skyrim and StarCraft II: Heart of the Swarm. Interestingly, the old dual-core chip pulls into the lead once graphics quality is cranked up.



A maximum overclock on the E8400 was a necessity for smooth Ultimate-quality game play in the most CPU-demanding areas of Tomb Raider. Meanwhile, a tweaked Core 2 Quad surpasses Core i3-3225, although it still can't offer the sustained frame rates of a Core i5-3570K.

Power-saving features were enabled for each processor at stock frequencies, but disabled for higher and more stable overclocks. We chart peak input watts during active idle, full Prime95 load, and throughout our gaming tests.
Our bench rig is way overpowered with the 80 PLUS Silver-rated Corsair HX1050. Gaming on these single-GPU platforms, we can assume roughly 87-88% efficiency, meaning the unit’s peak output was only around 185-285 W.

Although Intel’s Core 2 architecture wasn’t all too shabby in terms of energy consumption and efficiency back in the day, we can certainly appreciate the increased performance per watt the Ivy Bridge design enables. Take a close look at the E8400 and Q9550. The added performance secured by boosting voltages came at a steep increase in power consumption.
Let’s summarize our performance testing, and try to pinpoint how these various chips stack up to one another. Of course, what matters more than the averages is how they perform in the apps most important to you.

Was anyone surprised by the potency of the little Celeron G1610? It beat out the Core 2 Duo E8400 by 8% in both games and applications, placing it directly between the Core 2 Duo E8500 and E8600! Our in-store purchase was a real bargain, but this helps explain why the street price is often inflated.
Equally impressive is how well the Pentium G2020 performs, outpacing the E8400 by 21%, despite a 100 MHz-lower clock rate. That’s a solid showing for a $50-65 chip, even if the lack of overclocking support means you'll never get anything extra out of it.
On the other hand, overclocking is why we came to know and love Intel's Wolfdale design. Pushed to 4.0 GHz, our Core 2 Duo E8400 out-gamed the Pentium G2020 by 1.9% in average FPS, and by 4.0% in absolute minimum FPS. Its lead in applications was slightly higher still. Had we instead tested a Pentium G2120, with a 6.9% speed bump, gaming would likely have swung back in favor of Ivy Bridge.
Core 2 Quad Q9550 out-gamed the Pentium G2020 by only 3.3% in both average and minimum frames per second. More than likely, a Pentium G2120 would match the Q9550, while the G2020 we tested could hang in there with a Core 2 Quad Q9400. Of course, things are rarely this simple, and the outcome will almost always depend on the specific game in question. The most well-threaded titles we tested favored the Core 2 Quad architecture.
So what happens if we pull out Skyrim and StarCraft 2, the Pentium processor's strongest showings, since those two titles gain little to no benefit from more than two cores? Looking at only the highest playable settings for our (overpowered) Radeon HD 7970 paired with either chip, the Core 2 Quad Q9550 only held a 9% lead, which is far less than we expected. And in older titles, the Pentium would likely come out on top. Of course, everything changes once we overclock, but with such a small advantage favoring the stock Q9550, I’m left wondering how my overclocked Kentsfield-based Q6600 at 3.0 GHz compares to the affordable and power-friendly Pentium G2020.
Thankfully, applications continue to be better-optimized for multiple cores, and not one of our dual-core chips gets an advantage in productivity-oriented titles. Older quad-core processors like the Core 2 Quad Q9550 still pack a decent punch, even besting the new Ivy Bridge-based Core i3-3225 in a few of our benchmarked applications. Once we factor in overclocking, the locked-down Core i3 falls behind in every threaded title.
But gaming introduces other platform demands, and doesn't always benefit from threading the same way as content creation titles do. The Core i3-3225 simply dominates in this discipline. Once we overclock to 3.7 GHz, though, our E0-stepping Core 2 Quad is able to compete against the Ivy Bridge-based Core i3, and it even wins in three out of eight gaming titles. In short, if you already have a high-clocked Core 2 Quad, then Core i3 is not worth upgrading to based on performance alone.

Having tested and analyzed data for processors launched five years apart, what exactly did we learn?
In the little-league battle for dual-core supremacy, the entry-level Ivy Bridge-based Celeron G1610 stands toe-to-toe with the infamous Core 2 Duo E8500 and E8600, in both games and applications. The Pentium G2020 is a solid step up from either, making it a rough equivalent (in games, at least) to a 3.8 or even 3.9 GHz Wolfdale-based chip. If you are using an older Conroe-based Core 2 Duo, which didn’t overclock as well and couldn't match the Wolfdale design on a per-clock basis, then you’ll be lucky to squeeze out the performance of an entry-level Celeron G1610. At the top of Intel’s current Pentium line-up, the G2130 should game almost as well as any Wolfdale-based processor on air cooling. We suspect our 4.5 GHz E8400 may have a slight edge, but it might not matter in the long run. All of these dual-core chips share the same weakness: they're limited to two threads at a time.
Can we still recommend dual-core processors for gaming? At the most, we'd do so with reservations, and only after considering the prominence of gaming in your life, build goals, and the availability of other options. Locked ratio multipliers and the ability to execute two threads concurrently are both major strikes against Ivy Bridge-based Pentiums. For this story, I purposely picked a brutal batch of games to expose any weaknesses these budget-oriented chips might have, today and moving forward. Games are increasingly optimized for quad-core processors, and the trend is towards needing more than two cores. However, it’s rare that an Ivy Bridge-based Pentium falls short of playability. Plus, they have low prices, little heat, and modest power consumption in their favor, not to mention solid per-clock performance and a respectable upgrade path.
Given the right purpose and budget, I could argue that there is no better gaming CPU than a G1610 or G2020. Perhaps your funds are limited, and a dual-core CPU can tide you over until you can afford a Core i5 or i7. Maybe you're building a mini-ITX HTPC for gaming, where power consumption and cooling are equally important. A dual-core Ivy Bridge-based CPU is also smart in a pure gaming machine, though mostly if your combined CPU and GPU budget is in the $130-$230 range. Below that, a Trinity-based APU might be a better choice. Above, and you start leaving too much performance on the table, risking unplayable performance in some titles, no matter how much money you sink into graphics.
The final hurdle for dual-core Ivy Bridge-based processors to overcome is competition from AMD’s aging architectures. Athlon II X4 and Phenom II X4 offerings are available at or below $100, pack four physical cores and can be overclocked to overcome limitations at stock clock rates. At 3 GHz, is an Athlon II X4 640 better suited to gaming than a lower-priced Pentium? Or would it fall short in exactly the same areas as a stock Core 2 Quad Q9550? We plan to put quad-core Athlon II, Phenom II, and FX processors, stock and overclocked, against this crop of Intel processors soon.
In the end, we're impressed by the staying power of Intel's Core 2 architecture, especially the 45 nm CPUs tested today. But we’re also saddened that Intel no longer sells budget-friendly processors to enthusiasts, like so many Celerons and Pentiums from the past. While the company clearly made big improvements to threaded performance, memory bandwidth, efficiency, and value-added features, it's a little disappointing that an overclocked Core 2 Duo from four or five years ago can match or beat today's best dual-core offerings. Of course, we're keeping in mind that the E8400 original sold for almost three times as much. But still, it would be fun to disable two of Core i5-3570K's cores and see how a K-series Pentium might have performed.
Finally, we found that an overclocked Core 2 Quad (Yorkfield) easily outmaneuvers the latest Core i3 in many video encoding and productivity applications. Once it's clocked high enough, it even matches the i3 in threaded games.
The real winner today has to be Intel's Core i5-3570K control processor. Out of the box, it's pretty much unrivaled by any of these lower-priced chips. And, sporting the unlocked multiplier enthusiasts covet, it'd let us tap into even greater levels of performance, if that were the point of this story.
