Sign in with
Sign up | Sign in
Intel Core i7-3930K And Core i7-3820: Sandy Bridge-E, Cheaper
By ,
1. Core i7-3930K And -3820 Get Reviewed

Editor's Note: As you'll see by the end of this story, we liked the Core i7-3930K for its specific purpose quite a bit. It's a pleasure, then, to offer four of these CPUs to our readers. Of course, we realize that the platform is still pricey, so we also have a quartet of Intel DX79SI motherboards and as many 120 GB SSD 320 drives. Four lucky winners will walk away with a trio of parts to get them started on their next machine. Read to the end of this story for your chance to win!

We love to lust after the latest hardware, no matter the cost. But when it comes time to buy, sexy takes a back seat to sensible. That’s why a thousand-dollar processor like Intel’s Core i7-3960X doesn’t really add up. According to Intel’s official price list, you can get the Core i7-3930K for exactly $416 dollars less, sacrificing 3 MB of shared L3 cache and 100 MHz in the process. There’s an Epic Meal Time phrase I could use to illustrate the intelligence of that trade-off, but I’ll just leave it as: smart.

And so today’s story is brought to you by a couple of different ideas. First, we got our hands on the other two LGA 2011-based parts: Core i7-3930K and Core i7-3820. We’ll give you the performance data on those chips running at their stock settings.

The former launched alongside Core i7-3960X, although it looks like Intel took a look at the competitive landscape, realized there wasn’t anything to contend with Sandy Bridge-E, and tacked an extra $10 onto its -3960X and $33 to its -3930K. As a result, the $550 I cited in my launch coverage is now officially $583, but closer to $600 once the online guys add their mark-up.

The latter wasn’t as easy to track down. Although Core i7-3820’s specifications are known, it’s not officially available until 2012. But since we’ve heard that a lot of really sensitive stuff gets left behind at bars, we’ve been drinking ourselves silly, hoping to scoop up a lonely Sandy-E with only half of her faculties intact.

The darnest things wind up in bars...The darnest things wind up in bars...

Sandy Bridge-E finds itself in a tough position no matter which model you flag for its potential superiority over the other two, though. Stepping away from Intel’s flagship platform, you’re faced with Sandy Bridge, the mainstream architecture we’ve praised over the last 11 months for its stock performance, reasonable cost, efficiency, and overclockability. At $1000, $600, and an undisclosed third price (a bar fly suggested $285), it’s hard to see any conceivable way to have a serious discussion about Sandy Bridge-E’s value compared to unlocked alternatives like Core i5-2500K and Core i7-2600K.

And then there’s Ivy Bridge, expected to land in the first half of next year. Informed enthusiasts have to be wary of buying into Sandy Bridge-E knowing that Ivy Bridge could deliver better performance in basic desktop applications.

How much of a threat will the quad-core Ivy Bridge-based chips be to six-core Sandy Bridge-E processors? It’s too early to say for sure, since production-quality hardware is still months away. But we were able to scoop up an Ivy Bridge-based 55 W Core i3 at the very same watering hole and run a handful of tests. My numbers are quite a bit lower than what Intel suggests on these leaked slides though, so we’re going to wait and see how the early processors evolve before posting charts. Should enthusiasts worry that Ivy Bridge will usurp their Sandy Bridge-E-based systems? I don’t think so. It kind of sucks that the high-end crowd has to wait another year or so for Ivy Bridge-E, but these two segments want different things, and a 77 W quad-core CPU isn’t going to displace a 130 W hexa-core processor in those environments.

Besides, I’m not 100% convinced that the Ivy Bridge/6-series chipset compatibility story is going to end as harmoniously as many folks with P67 and Z68 boards hope.  

Deriving Value From The High-End?

I read all of the comments from my review of Core i7-3960X proclaiming Sandy Bridge-E fodder for chumps with too much spare change. And although I agree insofar as Intel’s flagship isn’t the right model to buy for maximizing bang for your buck, you simply have to concede that, in threaded applications, it’s faster than both the Core i7-990X it replaces and the four-core Core i7-2600K to which it’s so easily compared. As a result, there will be affluent early adopters and professionals eager to pay top dollar for the fastest single-processor system available.

Here’s the thing, though. Having the money to spend shouldn’t compel anyone to blow it. Now that the Sandy Bridge-E launch is history and prices are publically available on sites like Newegg and TigerDirect, we can add up the cost of the parts we used and consider scaling back to save some cash.

The Intel DX79SI motherboard we used for our launch piece (and continue to use today) just hit Newegg's virtual shelves at $280. Add to that $1050 for a Core i7-3960X and $800 for two 16 GB DDR3-1333 kits from Crucial, and you’re looking at a total of $2130 for the three platform-specific components. Now, it’s clear that the insane price of high-density memory is throwing that number off. So let’s avoid sensationalizing the bottom line and swap over to the 16 GB DDR3-2133 kit from G.Skill that I used to test memory scaling in the original story. At $180, we end up with a (relatively) more palatable $1530.

Can we pare back the parts list to make Sandy Bridge-E a better buy for a larger group of enthusiasts…without sacrificing performance? I’ll explore the benchmarks shortly, but here’s the parts list that I’m going to use on my quest:

Components From Today's Experiment
Price on Newegg
Intel Core i7-3930K
$600
ASRock X79 Extreme4-M
$225
G.Skill F3-12800CL9Q-8GBRL (4 x 2 GB DDR3-1600)
$55
Total:
$880


At $225, ASRock’s X79 Extreme4-M is the least-expensive (and conveniently microATX-sized) X79-based motherboard available. G.Skill’s 8 GB, DDR3-1600 Ripjaws kit isn’t particularly flashy, but it gets the job done at a capacity point we’d still consider beefy. And of course, the Core i7-3930K gets us as close to Intel’s flagship as possible without giving up any cores. The grand total? $885. That’s an almost $650 chunk off of the original setup after dropping the high-density memory kit.

Although you’re still paying nearly $400 more than the cheapest Z68-based board, the same 8 GB memory kit, and a Core i7-2700K, six cores simply aren’t available in the mainstream space. Suddenly, the prospects of a fast hexa-core beast look a lot rosier for the folks who skipped over Sandy Bridge entirely in anticipation of higher-end hardware.

2. Overclocking Sandy Bridge-E On A Budget

Overclocking Core i7-3930K

Now, you don’t swap out a $300 motherboard for a $225 platform and expect the same experience, so I initially tempered my expectations of overclocking on ASRock’s X79 Extreme4-M.

That was premature, though. Using the company’s latest firmware, I had no trouble booting up at 4.4, then 4.5, 4.6, and finally 4.7 GHz. Those last two frequencies were fine for benchmarking, but they gave out under IntelBurnTest, compelling me to settle in at a modest 4.5 GHz.

I could have gone higher. The chip’s power consumption crested at 183 W at 1.375 V, and ASRock indicated to me that its board should take 200 W or so. Given a 91°C ceiling and temperatures that topped out around 80°, thermals weren’t the problem. The Core i7-3930K wanted more voltage. But even the settings I used aren’t guaranteed to be safe over the long term, so it wasn’t worth it to me to nudge up to 1.4 V.

With all of that said, 4.5 GHz was rock-solid down at 1.361 V, so long as the most important variable was controlled: cooling. I used Intel’s RTS2011LC for my processor, which left the X79 Extreme4-M’s VRM vulnerable. Naturally, instability ran rampant even at lower power levels. Simply adding a fan blowing over the motherboard solved all heat-related issues.

As a general rule of thumb, it’s possible to get great clock rates from the Sandy Bridge-E-based chips we’ve tested, so long as you’re willing to put plenty of voltage through them and throw big cooling at the resulting heat. We’ve talked to system builders willing to go 4.4 GHz on shipping machines, so our 4.5 GHz overclock on a retail-purchased processor turns out to be pretty solid.

Overclocking Core i7-3820

The Core i7-3820 hits respectable frequencies as easily, but it requires a slightly different approach. Because it’s neither an X- nor a K-series SKU, the -3820 is constrained by “limited overclocking.” In short, it scales up to six 100 MHz bins beyond its maximum Turbo Boost clocks. With three or four cores active, it hits 4.3 GHz. When one or two cores are busy, it jumps to 4.4 GHz.

That leaves performance on the table, though, making it necessary to exploit the strap ratios incorporated into the X79 Express platform. ASRock’s X79 Extreme4-M doesn’t expose them explicitly, though we’ve asked the company to add the ratios, and it now plans to. However, manually specifying 125 MHz, for example, allows the PCI Express and DMI buses to remain within spec.

Interestingly, our -3820 didn’t want to run at 4.5 GHz, but it worked at 4.625 and 4.75 GHz using 37x and 38x multipliers. Still finicky, it wouldn’t complete the entire benchmark suite, even with a longevity-unfriendly 1.44 V driving it. But my expectations for this one weren’t high anyway. And if you need a quad-core chip, I don’t see any reason to buy a high-end platform (X79), quad-channel memory kit, and a locked processor when the Z68/Core i7-2600K combo is cheaper, still very capable, and equipped with Quick Sync support.

3. Test Setup And Benchmarks

Before we get into the testing, I need to clarify one point from my launch coverage. In that piece, I used Intel’s DX79SI motherboard, along with a 16 GB memory kit from G.Skill to measure memory bandwidth scaling. Except that the modules wouldn’t operate above DDR3-1600, forcing me to switch to an alternative platform.

In working with Intel afterward, it was determined that the board’s firmware had an XMP bug that kept profiles from loading, necessitating manual specification of looser timings at higher data rates. A subsequent beta update fixed this, putting the Intel board back into shape for testing here today.

The blue bars you see across the following pages represent all of the tested processors at their default settings. For all Sandy Bridge-E-based platforms, testing is performed on Intel’s DX79SI, just as it was in my launch piece, with 32 GB of Crucial memory stomping out any potential capacity limitations.

The two red bars are indicative of both overclocked, cost-optimized platforms. There, I’m using the least-expensive X79 Express-based motherboard and quad-channel memory kit currently available on Newegg: ASRock’s X79 Extreme4-M and G.Skill’s F3-12800CL9Q-8GBZL.

Test Hardware
Processors
Intel Core i7-3930K (Sandy Bridge-E) 3.2 GHz (32 * 100 MHz), LGA 2011, 12 MB Shared L3, Hyper-Threading enabled, Turbo Boost enabled, Power-savings enabled

Intel Core i7-3820 (Sandy Bridge-E) 3.6 GHz (36 * 100 MHz), LGA 2011, 10 MB Shared L3, Hyper-Threading enabled, Turbo Boost enabled, Power-savings enabled

Intel Core i7-3960X (Sandy Bridge-E) 3.3 GHz (33 * 100 MHz), LGA 2011, 15 MB Shared L3, Hyper-Threading enabled, Turbo Boost enabled, Power-savings enabled

Intel Core i7-990X (Gulftown) 3.43 GHz (26 * 133 MHz), LGA 1366, 12 MB Shared L3, Hyper-Threading enabled, Turbo Boost enabled, Power-savings enabled

AMD FX-8150 (Zambezi) 3.6 GHz (18 * 200 MHz), Socket AM3+, 8 MB Shared L3, Turbo Core enabled, Power-savings enabled

AMD Phenom II X4 980 BE (Deneb) 3.7 GHz (18.5 * 200 MHz), Socket AM3, 6 MB Shared L3, Power-savings enabled

AMD Phenom II X6 1100T (Thuban) 3.3 GHz (16.5 * 200 MHz), Socket AM3, 6 MB Shared L3, Turbo Core enabled, Power-savings enabled

Intel Core i7-2600K (Sandy Bridge) 3.4 GHz (34 * 100 MHz), LGA 1155, 8 MB Shared L3, Hyper-Threading enabled, Turbo Boost enabled, Power-savings enabled

Intel Core i5-2500K (Sandy Bridge) 3.3 GHz (33 * 100 MHz), LGA 1155, 6 MB Shared L3, Turbo Boost enabled, Power-savings enabled

Intel Core i7-920 (Bloomfield) 2.66 GHz (20 * 133 MHz), LGA 1366, 8 MB Shared L3, Hyper-Threading enabled, Turbo Boost enabled, Power-savings enabled
Motherboard
Intel DX79SI (LGA 2011) Intel X79 Express Chipset, BIOS SI.0280B

Asus Rampage IV Extreme (LGA 2011) Intel X79 Express Chipset, BIOS 0067

Asus Crosshair V Formula (Socket AM3+) AMD 990FX/SB950 Chipset, BIOS 0813

Asus Rampage III Formula (LGA 1366) Intel X58 Express, BIOS 0505

Asus Maximus IV Extreme (LGA 1155) Intel P67 Express, BIOS 0901
Memory
Crucial 32 GB (4 x 8 GB) DDR3-1333, MT16JTF1G64AZ-1G4D1 @ DDR3-1600 at 1.65 V on Socket AM3+ and LGA 2011, DDR-1333 at 1.65 V on LGA 1155

Crucial 24 GB (3 x 8 GB) DDR3-1333, MT16JTF1G64AZ-1G4D1 @ DDR3-1066 at 1.65 V on LGA 1366
Hard Drive
Intel SSD 510 250 GB, SATA 6 Gb/s
Graphics
Nvidia GeForce GTX 580 1.5 GB
Power Supply
Cooler Master UCP-1000 W
System Software And Drivers
Operating System
Windows 7 Ultimate 64-bit
DirectX
DirectX 11
Graphics DriverNvidia GeForce Release 280.26
Nvidia GeForce Release 285.62 for all SLI testing
3D Game Benchmarks And Settings
BenchmarkDetails
Crysis 2
Game Settings: Ultra Quality Settings, Anti-Aliasing: Disabled, V-sync: Disabled, High-Quality Textures: Enabled, DirectX 9 and DirectX 11, 1680x1050, 1920x1200, 2560x1600, Demo: Central Park
DiRT 3
Game Settings:  Ultra Quality Settings, Anti-Aliasing: Disabled and 8x AA, Anisotropic Filtering: Disabled, Sync Every Frame: No, 1680x1050, 1920x1080, 2560x1600, Demo: Built-in Game Demo
World of Warcraft: Cataclysm
Game Settings: Ultra Quality Settings, Anti-Aliasing: 1x AA and 8x AA, Anisotropic Filtering: 16x, Vertical Sync: Disabled, 1680x1050, 1920x1080, 2560x1600, Demo: Crushblow to The Krazzworks, DirectX 11
Audio Benchmarks and Settings
BenchmarkDetails
iTunesVersion: 10.4.10, 64-bit
Audio CD ("Terminator II" SE), 53 min., Convert to AAC audio format
Lame MP3Version 3.98.3
Audio CD "Terminator II SE", 53 min, convert WAV to MP3 audio format, Command: -b 160 --nores (160 Kb/s)
Video Benchmarks and Settings
BenchmarkDetails
HandBrake CLIVersion: 0.95
Video: Big Buck Bunny (720x480, 23.972 frames) 5 Minutes, Audio: Dolby Digital, 48 000 Hz, Six-Channel, English, to Video: AVC Audio: AC3 Audio2: AAC (High Profile)
MainConcept Reference v2.2
Version: 2.2.0.5440
MPEG-2 to H.264, MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG-2), Audio:
MPEG-2 (44.1 kHz, 2 Channel, 16-Bit, 224 Kb/s), Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV
x264 Software LibraryAMD-Supplied AVX- and XOP-Optimized builds, TechARP's x264 HD Benchmark 4.0, Modified to accommodate new versions of x264 and CPU-Z 1.58
Application Benchmarks and Settings
BenchmarkDetails
WinRARVersion 4.01
RAR, Syntax "winrar a -r -m3", Benchmark: 2010-THG-Workload
WinZip 14Version 14.0 Pro (8652)
WinZip Commandline Version 3, ZIPX, Syntax "-a -ez -p -r", Benchmark: 2010-THG-Workload
7-Zip
Version 9.20 (x64)
LZMA2, Syntax "a -t7z -r -m0=LZMA2 -mx=5", Benchmark: 2010-THG-Workload
Adobe Premiere Pro CS 5.5
Paladin Sequence to H.264 Blu-ray
Output 1920x1080, Maximum Quality, Mercury Playback Engine: Hardware Mode
Adobe After Effects CS 5.5
Create Video which includes 3 Streams
Frames: 210, Render Multiple Frames Simultaneously: on
BlenderVersion: 2.59
Syntax blender -b thg.blend -f 1, Resolution: 1920x1080, Anti-Aliasing: 8x, Render: THG.blend frame 1
Adobe Photoshop CS 5.1 (64-Bit)Version: 11
Filtering a 16 MB TIF (15 000x7266), Filters:, Radial Blur (Amount: 10, Method: zoom, Quality: good) Shape Blur (Radius: 46 px; custom shape: Trademark sysmbol) Median (Radius: 1px) Polar Coordinates (Rectangular to Polar)
ABBYY FineReaderVersion: 10 Professional Build (10.0.102.82)
Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages
3ds Max 2012
Render Space Flyby, 1440x1080, from Y: RAM Drive
Adobe Acrobat X Professional
PDF Document Creation (Print) from Microsoft PowerPoint 2010
SolidWorks 2010
PhotoView 360, 01-Lighter Explode.SLDASM Benchmark File, 1920x1080 Render, 1.44 Million Polygons, 256 AA Samples
Visual Studio 2010
Miranda IM Compile, Scripted
Synthetic Benchmarks and Settings
BenchmarkDetails
PCMark 7Version: 1.0.4
3DMark 11
Version 1.0.2
SiSoftware Sandra 2011Version: 17.80
Processor Arithmetic, Multimedia, Cryptography, Memory Bandwidth, .NET Arithmetic, .NET Multimedia
4. Benchmark Results: PCMark 7

Although Futuremark tells us that PCMark 7 is able to utilize at least 16 cores (and, given the outcome of the Computation sub-test, we believe the company), the main battery of tests really only seems to exploit four. As a result, the Core i7-3820, with 2.5 MB of shared L3 cache per core and a 3.8 GHz maximum Turbo Boost frequency, actually lands in first place at its default settings.

We’re not particularly impressed, though; the Core i7-2600K really isn’t very far behind—and on a much more affordable platform. AMD’s chips and the Core i7-920 are the only losers here.

Overclocked, the Core i7-3930K flexes its muscle more visibly. The 4.625 GHz Core i7-3820 trails a ways, but still easily bests all of the processors operating at their factory settings.

5. Benchmark Results: 3DMark 11

Although the Performance preset isn’t 3DMark 11's highest, it’s still very much a graphics-bound metric. Thus, the overall suite score is pretty boring to look at.

Not until you reach the physics measurements does this synthetic benchmark factor out graphics and emphasize each platform’s CPU. At that point, the conclusion is foregone. Six Sandy Bridge-based cores running at 4.5 GHz decimate the rest of the field.

The Core i7-3820 at 4.625 GHz slides in ahead of the Core i7-3930K at its factory settings, but the comparison is hardly a fair one. Perhaps more interesting is that the -3820 manages a notable lead against Core i7-2600K, which doesn’t enjoy as high of a base clock rate and consequently falls back a bit.

6. Benchmark Results: Sandra 2011

In a measure of pure arithmetic alacrity, the overclocked Core i7-3930K once again dominates. Two other hexa-core Sandy Bridge-E configurations follow before the overclocked quad-core -3820 is allowed to surface, just ahead of Core i7-990X. The stock version of the same chip comes after that, besting Intel’s Core i7-2600K.

Strange that the overclocked Core i7-3820 leads this test, right?

Not really. In my Sandy Bridge-E review, we figured out that the AES256 throughput of Intel’s AES-NI-equipped CPUs is tied directly to memory bandwidth. With only four cores mated to a quad-channel DDR3-1600 subsystem, the -3820 rises right to the top of our Cryptography results.

Of course, overall, the six-core models are more powerful, which is why you see them turn in better SHA256 hashing numbers. The big jump in memory bandwidth is also responsible for the big jump in AES256 bandwidth moving from Sandy Bridge to Sandy Bridge-E.

7. Benchmark Results: Content Creation

A pretty easy analysis, 3ds Max 2012 clearly benefits from as many cores as you throw at it. Overclocking is enough to push the 4.625 GHz Core i7-3820 up next to a stock Core i7-990X. However, it’s the 4.5 GHz Core i7-3930K that really redefines performance here.

The one-second difference between Core i7-3930K and -3960X confirms what we originally suspected: there’s little reason to buy the $1000-dollar chip from a stock performance perspective.

In that same vein, a Core i7-3820 really doesn’t seem far enough superior to Core i7-2600K to warrant its more expensive platform.

The same holds true in Photoshop, and all six-core processors outperform their quad-core competition, regardless of architecture or clock rate. An overclocked Core i7-3930K really shines here, and we again see a close finish between the Core i7-3960X and -3930K at their respective stock clocks.

Accelerated by a GeForce GTX 580, this workload doesn’t take nearly as long as it used to. However, the compute muscle leveled against the task by a Core i7-3930K running at 4.5 GHz is enough to cut 10 seconds from the same processor tackling it at its stock speed.

Notoriously memory-hungry, After Effects reminds us that the overclocked machines only include 8 GB of memory, while all of the configurations with blue bars boast 32 GB (except for the X58-based Core i7-920, limited to 24 GB by its triple-channel controller).

If you’re a video editor, this chart should be proof positive that lots of RAM needs to be a priority. After all, it’d be a shame to sink your cash into a $600 CPU, overclock it, and still get outperformed by a $200 Core i5-2500K at its stock settings (but with more memory).

Though clearly well-threaded, Blender also demonstrates an affinity for the Sandy Bridge architecture running at high frequencies.  

A mix of architectural improvements and unadulterated clock rate propel the overclocked Core i7-3930K up ahead of the pack. A stock Core i7-3960X follows, trailed only just slightly by the Core i7-3930K at its default configuration.

The overclocked Core i7-3820 comes in fourth, ahead of the Gulftown-based Core i7-990X. But the fact that a Core i7-3820 at its stock settings falls into sixth place shows that parallelism matters just as much, if not more, than operating frequency.

8. Benchmark Results: Productivity

This finishing order is starting to look repetitive, isn’t it? The overclocked Core i7-3930K clinches a commanding finish, but because ABBYY’s FineReader 10 is well parallelized, the other two six-core Sandy Bridge-E-based setups snag second and third place ahead of Intel’s Core i7-3820 overclocked to 4.625 GHz.

In a single-threaded test, architecture and clock rate rule. The 4.625 GHz chip takes first, followed by the 4.5 GHz contender. The rest of the field drops in behind (led by the quad-core Core i7-3820, almost humorously enough).

Corel recently launched a newer version of WinZip, and we’re working on automating it. In the meantime, WinZip 14 demonstrates the same single-threaded behavior we’ve seen so many times before. The line-up is quite similar to what we just saw in our Lame conversion.

WinRAR is a different sort of compression app. And while it definitely uses more than one thread, the quad-core Sandy Bridge-E chip still holds onto first place, followed by the 4.5 GHz Core i7-3930K.

The more thoroughly parallelized 7-Zip rewards Intel’s six-core processors with top finishes. The only deviation is the overclocked quad-core model, which uses its searing clock rate to secure a second-place berth.

This seemingly single-threaded metric leaves us with what we can now say is a pattern. Sandy Bridge is the favored architecture, and enthusiasts able to get it running over 4 GHz stand to realize impressive performance.

Overclocking trumps all in this compile workload. The other six-core chips end up landing pretty close to each other, as Intel’s Core i7-3820 trails the Gulftown-based Core i7-990X at its stock settings.

9. Benchmark Results: Media Encoding

In a single-threaded app like iTunes, cranking up the clock rate is an effective way to directly affect performance. Both overclocked chips offer an appreciable advantage over the stock processors, even if this is a fairly low-impact workload.

Optimized for multi-core CPUs, MainConcept unquestionably favors the overclocked Core i7-3930K.

An increase in clock rate allows the quad-core -3820 competing with the hexa-core -3930K running at its default settings, though the -3930K’s unlocked multiplier makes it easier to get a better-than 20% performance increase.

Similar to MainConcept, HandBrake is threaded, allowing six overclocked cores to spring into a commanding lead. Again, if you’re a video guy, an artist, or an engineer, there’s a legitimate reason to sink cash into the right hexa-core chip. Even the aged Core i7-990X fares well here.

10. Benchmark Results: Crysis 2

Minimal variation between platforms in this graphics-bound title is expected. There’s really no reason to highlight any particular result. When it comes to Crysis 2 (and a single-GPU setup), any of these processors are suitable.

11. Benchmark Results: DiRT 3

From the bottom of our platforms to top, there is a measurable delta in DiRT 3 at 1680x1050 and 1920x1080, particularly when AA is disabled.

The overclocked Core i7-3820 is missing because it wasn’t able to consistently complete runs in DiRT 3, despite our aggressive settings. Dropping the multiplier with a 125 MHz BCLK made the issues worse, and the only way to run completely stably was to settle on a 100 MHz BCLK and accept the -3820’s maximum ratio of 43x with all cores active.

12. Benchmark Results: World Of Warcraft

Perhaps the most processor-bound game in our suite, World of Warcraft has little trouble taking advantage of overclocked Sandy Bridge-based cores at 1680x1050 and 1920x1080. Of course, cranking up anti-aliasing shifts the game’s emphasis back onto graphics.

Again, we’re missing numbers for the -3820 because it simply wasn’t stable through testing. The quad-core part performs well at its stock clocks, though.

13. Core i7-3930K And -3820: Stock Versus Overclocked

On average, overclocking Intel’s Core i7-3930K in 16 different applications results in a 16% speed-up. But that’s including After Effects, which suffers a horrible -27.8% slow-down due to our decision to shift from 32 GB of DDR3 memory down to 8 GB. Factor that out, and the average springs up to 19%.

The -3820 also picks up 16% with After Effects considered, and 19% with the memory-hungry app disqualified. Unfortunately, getting there on our M0-stepping processor required some nasty voltage settings, and even then, DiRT 3 and World of Warcraft exhibited periodic stability issues that kept us from collecting benchmark results. Stepping down to 4.5 or 4.375 GHz using the 125 MHz strap didn’t help, either.

Although the Core i7-3820 does enjoy its high points against Gulftown and the Sandy Bridge-based Core i7-2600K, its pricey platform, locked multiplier, and quad-channel memory requirement don’t seem to deliver any more value than a Core i7-2600K on an inexpensive Z68-based board—and that platform gets you Quick Sync support.

14. Core i7-3930K and -2600K: Making The Tough Choice

Now, here’s where things get murkier. Add up all of the benchmark results, and the stock $600 Core i7-3930K only gives you a 15% average advantage over the stock Core i7-2600K.

Look at where the gains are lost and found, though. Lame, WinZip, Acrobat X, iTunes, and to some extent After Effects offer little or even negative scaling. Photoshop, 3ds Max, Premiere Pro, SolidWorks, FineReader, 7-Zip, MainConcept, and Handbrake all favor the Core i7-3930K by around 20% up to about 30%.

If the amount of money I make is affected by those performance numbers, or if I do a lot of video work, or if I’m even applying threaded filters in Photoshop, paying an extra $280 for the -3930K is worth it.

If you’re gaming with one (or even two) high-end graphics cards in your system, doing a lot of desktop productivity work, or simply on a more constrained budget, the -2600K is a better bet. Shoot, at that point, I’d still say step it on down to a -2500K.

15. Core i7-3930K: The Smart Sandy Bridge-E Choice

After my review of Intel’s Core i7-3960X, I heard second-hand comments from several sources at Intel wondering why everyone else seemed to love the chip, and yet I pretty much recommended against buying it.

Knowing that the Core i7-3930K was just 100 MHz off of its mark and down 3 MB of shared L3 cache, I just had to know how it compared. So, I went and dropped $600 on the thing at Newegg. As an enthusiast making an actual purchase, and now with both the -3930K and -3960X in my possession, I can unequivocally affirm what I suggested in Intel Core i7-3960X Review: Sandy Bridge-E And X79 Express: mainly, that smart enthusiasts who need the effective Sandy Bridge architecture and raw compute power of six cores will buy Core i7-3930K instead of-3960X.

We’ve seen a ton of variance in the overclocking headroom of C0, C1, M0, and retail Sandy Bridge-E-based chips, so it makes little sense to crank our engineering sample as high as it’ll go to compare against this store-bought -3930K. What I can say, though, is that at 4.5 GHz and 1.36 V, you can’t do much better from a single-CPU daily-driver platform.

And that’s why I’m pleased to hand out my first ever Best of Tom’s Hardware award to Intel’s Core i7-3930K. Obviously, this isn’t something you see very often. It’s a distinction reserved for the best of the best, price be damned. Except, in this case, the price (relative to the flagship that just launched ahead of it), actually isn’t obscene.

We’ve seen that it’s super-easy to build a very expensive Sandy Bridge-E-based machine, and prices on high-end X79 motherboards like the ones featured in Thomas’ recent Ultimate X79? Five $320+ LGA 2011 Motherboards, Reviewed only perpetuate that stereotype. But I was able to snag the cheapest LGA 2011-equipped board on Newegg, ASRock’s X79 Extreme4-M, and construct a fairly feature-rich machine capable of potent two-card graphics configurations. It’s not the most purpose-built overclocking motherboard. However, the company deserves big credit for its ability to create a platform up to my little experiment.

And while it’s tempting to think a quad-channel memory controller needs to be populated with 4 GB modules, at least, 8 GB is still ample for most folks. G.Skill’s 8 GB F3-12800CL9Q-8GBZL kit gave me the data rate I wanted (1600 MT/s) at a modest 1.5 V using XMP settings. And that’s priced at $55 bucks (again, I went with the cheapest kit on Newegg). Done deal. Of course, After Effects demonstrated to us that some workloads can utilize more memory if it’s available, so if you’re in that category of high-end buyer ready to run workstation-class apps (and not as concerned about building on Sandy Bridge-E for half the price of my launch review), then step it up to 16 GB at least.

Want A -3930K Of Your Own?

Would you like to win your own Intel Core i7-3930K? How about a DX79SI motherboard and SSD 320 drive? Click here to enter for your chance!

The contest opens on December 8, 2011 9:00 PM PST and closes on January 12, 2012 9:00 PM PST.

Four Winners Will Be Chosen Randomly.

Prizes (provided by Intel):
Four (4) prizes consisting of one (1) Core i7-3930K CPU, one (1) DX79SI motherboard, and one (1) 320 Series 120 GB SSD. Approximate Retail Value Each: $600+$280+$200=$1,080

DUE TO LEGAL REQUIREMENTS, THIS CONTEST IS LIMITED TO LEGAL RESIDENTS OF THE USA (EXCLUDING RI) AND 18 YEARS OF AGE OR OLDER.

UNLESS OTHERWISE NOTED, ALL PERSONAL INFORMATION WILL ONLY BE USED TO QUALIFY AND CONTACT THE WINNER.