Sign in with
Sign up | Sign in
Your question
Closed

Intel Core i7-3930K And Core i7-3820: Sandy Bridge-E, Cheaper

Last response: in Reviews comments
Share
December 9, 2011 3:41:42 AM

This is a really excellent analysis. Clearly, I must be drinking at the wrong places because I never leave the pub with any hardware nicer than a hangover.
Score
11
December 9, 2011 3:54:07 AM

So nice overclocking at 4.5ghz. I can expect that the upcoming ivy bridge unlocked series may be stable atleast 4.2 will all 4 cores active. I can't wait till Q2 next year to see benchmarks .
Score
0
Related resources
December 9, 2011 4:10:12 AM

Possible TYPO in the bottom graph for Dirt 3 benchmark.
FX-8150 benchmark with no AA says "68.8" FPS. I think it's more like "48.8".
Score
0
December 9, 2011 4:22:54 AM

For gaming (the high end CPU intensive), is there any noticeable difference between the 2500k and the 3960X?
Score
-5
December 9, 2011 4:31:32 AM

JOSHSKORNFor gaming (the high end CPU intensive), is there any noticeable difference between the 2500k and the 3960X?


If by "noticeable" you mean "perceivable to mere mortals", then no.

If you can in fact notice the difference between 105 vs 110 FPS, then you are a god, and you deserve only the best.
Score
27
December 9, 2011 4:35:46 AM

I bought the i7-3930K with 32GB of DDR3 1600 RAM last week and assembled a couple days ago. I have two Kingston 120GB SSDs in RAID that have been benched on my system at a theoretical 1,100MB/S Read and 1,300MB/S Write. Coming from a Pentium D 3.0GHz, this was like night and day. My renders went from 40minutes to 1minute. I'm not overclocking purely for the fact that this thing's a beast already and I'm doing high-end 3D work using Maya, Photoshop, After Effects, Video, etc. Also - I like the peace and quiet.

Intel did an awesome job with the SBE line - despite the fact that we're missing some wanted/promised features (native support for USB and PCI-Express 3.0. I'm waiting out for the PCI 3.0 cards before I upgrade my graphics... curious if the Asus P9X79 Pro will hold it's promises.

Thanks Chris for reviewing this processor. I felt like I went out on a limb getting this processor over the Extreme, but the $600 was well worth it.
Score
14
December 9, 2011 4:41:40 AM

spunky,

Glad you're enjoying. You do, actually get PCIe 3.0 support, but no USB 3.0, unfortunately.

Dacatak,

Yup, typo--fixing now!
Score
3
December 9, 2011 4:42:27 AM

the only good reason to get X79 is the more ram .. u can get cheap 32G ram system , or go for 64G of ram and enjoy a ram disk

it is a good thing
Score
3
December 9, 2011 5:24:42 AM

The Overclocking Sandy Bridge-E On A Budget page states, "With all of that said, 4.5 GHz was rock-solid down at 3.61 V". I'm pretty sure you meant 1.36 V.
Score
2
December 9, 2011 5:28:47 AM

soccerdocksThe Overclocking Sandy Bridge-E On A Budget page states, "With all of that said, 4.5 GHz was rock-solid down at 3.61 V". I'm pretty sure you meant 1.36 V.


Indeed, fixed! At 3.6 V, we'd have dead Sandy. :) 
Score
7
December 9, 2011 5:37:27 AM

Hmm, 7% improvement over 2600K in Visual Studio isn't all that impressive... Perhaps 3930K isn't such a smart choice for a developer workstation after all.
Score
-1
December 9, 2011 6:21:00 AM

Its interesting there is no core/core and clock/clock comparison with the 4 core 3820 and 2600k. If there was it would highlight just how little the X79 platform offers when compared to Z68.

Intel has made sure reviewers dont highlight on this factor, and instead asks reviewers to focus on the 6 core performance.

Intel didnt release the 4 core 3820(at launch) for this reason, it makes it easy to compare to normal sandy bridge and would show that even with a socket that is double the size, and quad channel memory X79 doesnt give you any better performance than Z68.

I always buy the high-end but X79 is a big letdown, Intel knows it and they're trying to control the reviews so it doesnt look as bad as it is
Score
-2
Anonymous
December 9, 2011 6:37:16 AM

For games and people who search for price/performance, i7 2600K/2700K is and will remain the best solution. With the price difference between the old Sandy and the new ones, you could buy a better video card, another one, or a SSD that would boost you system better.
Score
4
December 9, 2011 6:54:07 AM

cactus45Its interesting there is no core/core and clock/clock comparison with the 4 core 3820 and 2600k. If there was it would highlight just how little the X79 platform offers when compared to Z68. Intel has made sure reviewers dont highlight on this factor, and instead asks reviewers to focus on the 6 core performance. Intel didnt release the 4 core 3820(at launch) for this reason, it makes it easy to compare to normal sandy bridge and would show that even with a socket that is double the size, and quad channel memory X79 doesnt give you any better performance than Z68. I always buy the high-end but X79 is a big letdown, Intel knows it and they're trying to control the reviews so it doesnt look as bad as it is


This shouldn't be necessary. Same architecture = same per-clock performance. If you need numbers, look at iTunes, WinZip, and Lame benchmark results. If you need yet additional proof, check out the original Sandy Bridge-E review, where I explicitly run the results you're saying don't get run.

Finally, as is mentioned in *this* story, the CPUs didn't come from Intel. -3930K came from Newegg and -3820, which isn't out yet, came from an unnamed other source.

Thanks,
Chris
Score
9
December 9, 2011 7:10:13 AM

Intel should have sell the 2011 CPU as 8 core instead of disable the 2 cores and sell at six core. 2011 cpu may be an enthusiast CPU, but it is still a high volume CPU compared to server cpu,

it cant be the yield in Intel fab are so bad that all 2011 CPU produce by Intel have only 6 working cores at best.
Score
2
December 9, 2011 7:19:51 AM

cactus45Its interesting there is no core/core and clock/clock comparison with the 4 core 3820 and 2600k. If there was it would highlight just how little the X79 platform offers when compared to Z68. Intel has made sure reviewers dont highlight on this factor, and instead asks reviewers to focus on the 6 core performance. Intel didnt release the 4 core 3820(at launch) for this reason, it makes it easy to compare to normal sandy bridge and would show that even with a socket that is double the size, and quad channel memory X79 doesnt give you any better performance than Z68. I always buy the high-end but X79 is a big letdown, Intel knows it and they're trying to control the reviews so it doesnt look as bad as it is

This is the same as LGA 1366 v. LGA 1155 once the later was released. 1366 offered higher memory bandwidth and more Pci-e lanes, but even most enthusiasts wouldn't get the higher end platform due to price for performance.

Most settled for the i5-750(or lower since you could overclock anything then) just like most are settling for the 2500k now.
Score
-3
December 9, 2011 7:47:11 AM

wow.. . . Great intel is so good
Score
-2
December 9, 2011 8:11:15 AM

Why would you use Crysis 2 as a CPU benching game? Use Starcraft 2 instead.
Score
17
December 9, 2011 8:42:21 AM

Why are the gaming benchmarks on ultra settings where the GPU is the bottleneck? This is a CPU benchmark so resolution and settings should be on the lowest possible.
Score
-11
Anonymous
December 9, 2011 9:46:12 AM

It's just a shame that you didn't overclock the 2600K & 2500K during this article to give it a better perspective.
Score
8
December 9, 2011 10:09:33 AM

Great article! Kep going!
Score
0
December 9, 2011 10:32:42 AM

gmcizzleWhy would you use Crysis 2 as a CPU benching game? Use Starcraft 2 instead.

FunSurferWhy are the gaming benchmarks on ultra settings where the GPU is the bottleneck? This is a CPU benchmark so resolution and settings should be on the lowest possible.


The World of Warcraft benchmark already tells the story. And it's important to be realistic about how you use these chips. No one buy a $600 CPU and play at 1024x768 resolution at low graphic settings.
Score
6
December 9, 2011 10:56:15 AM

Chris can you do a 4 way multigpu article using core i7 3960x and compare it to core i7 980x and the X58 .
Score
2
December 9, 2011 11:20:20 AM

There's some error on the raffle page about proving you are a human, even though there nether graphic nor field to fill on the subject. Firefox 8.0
Score
-1
December 9, 2011 11:26:14 AM

Why isn't Starcraft 2 tested against this CPU? That game is greatly dependent on the CPU especially 4v4 matches. I would love to see benchmarks especially since you need a decent fps to play competitively
Score
1
December 9, 2011 11:52:43 AM

Chris if you could lose one of those test processors in an envelope that just happens to have this POOR user's address on it I'd be really a happy camper.

Do it before crashman steals everything on your bench and goes on some liquid nitrogen overclocking spree ... you just have to know he is eyeing your new stuff off right now ...

Just do it !!

:) 
Score
4
December 9, 2011 12:04:40 PM

Great review, as always :) 

Have a suggestion for the games. America's Army 2.8.5's co-op maps are very CPU intensive from my experience (especially after the enemy bots spawn). You may consider using that as a benchmark too, maybe. I believe it's largely single threaded however, so i'm not sure there's much use in this particular review, but generally it may be of some use. Maybe you could replace crysis 2 with it, for example.

Cheers,
Ojas
Score
-1
December 9, 2011 12:09:42 PM

HaserathThis is the same as LGA 1366 v. LGA 1155 once the later was released. 1366 offered higher memory bandwidth and more Pci-e lanes, but even most enthusiasts wouldn't get the higher end platform due to price for performance.Most settled for the i5-750(or lower since you could overclock anything then) just like most are settling for the 2500k now.


This is an invalid comparison. You meant to say LGA 1366 v. LGA 1156. The i5-750 was never made for LGA 1155. I also seem to recall that the i7-920 was quite popular because of its price for performance offering despite it being on a more expensive platform.
Score
2
December 9, 2011 12:45:21 PM

envolva said:
The World of Warcraft benchmark already tells the story. And it's important to be realistic about how you use these chips. No one buy a $600 CPU and play at 1024x768 resolution at low graphic settings.


If I am a PC gamer, the whole point of this CPU benchmark is to tell me how much is the 3930K better than the 2600K in games, and I just can't see it in any of these ultra-settings benchmarks, including WoW.
Score
0
December 9, 2011 12:58:51 PM

And not to mention why not all the damn benchmark can test cpu with the same damn frequency.. Intel showed us that they are stable overclocked overs years so take all the cpu overclock them to the cpu that has the highest stock mhz and make benchmark so we get a mhz per mhz performance.
Score
1
December 9, 2011 1:04:21 PM

wiggo100It's just a shame that you didn't overclock the 2600K & 2500K during this article to give it a better perspective.


Thank you for pointing this out. Having an over clocked processor beat out a stock processor is tilting the playing field a bit. The extra money for *any* SB-E is a waste.
Score
0
December 9, 2011 1:04:44 PM

agnickolovHmm, 7% improvement over 2600K in Visual Studio isn't all that impressive... Perhaps 3930K isn't such a smart choice for a developer workstation after all.

I work on a project with about 3M lines of code. To get build times under an hour, we've got a distributed build network set up, and it STILL takes about 10-15 minutes to compile the damn thing, depending on how much needs to be rebuilt. 7% in a situation like that is huge, especially when you're building potentially dozens of times a day.

Those few minutes here and there add up quickly. Say I build every hour at 10 minutes a build. I'm spending 16.6% of my time staring at the compiler. If I cut 7% off my build time, I save about a minute every hour, 8 minutes a day, or about an hour a week. At $50 an hour (roughly), this processor literally pays for itself in about two months. If we weren't using distributed builds, it would probably pay for itself inside a week, saving nearly $3 in labor per build.
Score
1
December 9, 2011 1:07:47 PM

FunSurferIf I am a PC gamer, the whole point of this CPU benchmark is to tell me how much is the 3930K better than the 2600K in games, and I just can't see it in any of these ultra-settings benchmarks, including WoW.

I think you've got unrealistic expectations. Games need GPU power way more than CPU power. You're just not going to see a major difference in gaming by upgrading your processor. Sink the $400 you would have spent on this into another high end GPU, set up SLI/Crossfire and enjoy the increased performance that way.

Or, if that's not an option, pick up a nice SSD and watch your load times disappear. Even on a SATAII platform, I don't load longer than 2 seconds in Skyrim...ever. Really nice since the engine is incredibly dated and forces you to load constantly. I barely even notice them anymore.
Score
0
December 9, 2011 1:19:58 PM

Great article! It really goes to show that
1) 2500K has all the performance you need for games and home use
2) 2600K is still a beast when it comes to productivity applications that do not need excessive amounts of ram, while staying on a budget (would have been nice to see OC 2xxxK processors to compare with the OC 3xxx processors)
4) If you need massive amounts of Ram then the x79 platform is a wonderful way to go!

In one of the first OC charts there are two 4.5 GHz when I think the 2nd one is supposed to be a 4.6

When benchmarking Adobe Premiere can you run a double benchmark so we can see the difference between running with GPU rendering vs CPU rendering? I know it would take more time, but it would better reflect how the CPU processes information, and also give an idea of how well the CPU works with CUDA. Also it would show people the potential difference between running Premiere on a high end nVidia card compared to a low end nVidia, AMD, or onboard GPU.

Great article! It really helps know what works best per budget and use!
Score
3
December 9, 2011 1:32:37 PM

willardI work on a project with about 3M lines of code. To get build times under an hour, we've got a distributed build network set up, and it STILL takes about 10-15 minutes to compile the damn thing, depending on how much needs to be rebuilt. 7% in a situation like that is huge, especially when you're building potentially dozens of times a day.Those few minutes here and there add up quickly. Say I build every hour at 10 minutes a build. I'm spending 16.6% of my time staring at the compiler. If I cut 7% off my build time, I save about a minute every hour, 8 minutes a day, or about an hour a week. At $50 an hour (roughly), this processor literally pays for itself in about two months. If we weren't using distributed builds, it would probably pay for itself inside a week, saving nearly $3 in labor per build.

Exactly! So many gamers think that everything intel makes is for them, and then they complain that the new chips suck and are overpriced. But these chips are made for the professional market where a $1000 processor can save tens of thousands of dollars in a year in productivity costs. When you work in that kind of environment then the old processor that you own becomes more expensive than the new 'expensive' processor. And these are still a steal compared to a server style setup that you would traditionally use in such an environment because they do not require the redundancy in the build that a server does.
Score
3
December 9, 2011 1:39:00 PM

spunkyddogI bought the i7-3930K with 32GB of DDR3 1600 RAM last week and assembled a couple days ago. I have two Kingston 120GB SSDs in RAID that have been benched on my system at a theoretical 1,100MB/S Read and 1,300MB/S Write. Coming from a Pentium D 3.0GHz, this was like night and day. My renders went from 40minutes to 1minute. I'm not overclocking purely for the fact that this thing's a beast already and I'm doing high-end 3D work using Maya, Photoshop, After Effects, Video, etc. Also - I like the peace and quiet.Intel did an awesome job with the SBE line - despite the fact that we're missing some wanted/promised features (native support for USB and PCI-Express 3.0. I'm waiting out for the PCI 3.0 cards before I upgrade my graphics... curious if the Asus P9X79 Pro will hold it's promises.Thanks Chris for reviewing this processor. I felt like I went out on a limb getting this processor over the Extreme, but the $600 was well worth it.


Is it just me or is spunkyddog's comment really impressive? That's what I need - I have to work with some of that stuff as well and I'm currently stuck with an AMD PC from 2004!!! I could do all I really needed to do myself if I had that PC. I'm just waiting for AMD to (finally) come out with motherboards with PCI Express 3.0 support before I buy for my next build. I have to decide between the super overly priced Intel or the much cheaper AMD. If money weren't an issue I'd go with Intel but, money certainly IS an issue.
Score
-1
December 9, 2011 2:03:07 PM

Kudos for spending 600 bucks of your own money to get this review done. That's customer service :) 
Score
4
December 9, 2011 2:13:38 PM

Very nice article.

I just build up my new computer about a month ago with a P8P67 WS Revolution and 2700K.

When I see these benchmarks, I see that I did a really good choice with the 1155 platform. I still find the socket 2011 very overpriced. Sure, performances are there and are incredible, but when you can afford waiting 5 seconds more on your 2700K on which you spent 370 $ instead of 1000$ for a 3960X, I really don't see why I should have gone with the 2011 platform.

I still think the price of a 3930K is too high. And while the 3820 is priced reasonably, the motherboard and quad-channel RAM is still priced higher than a 1155 platform for the exact same performance (excepted for HDD encryption and memory bandwidth).

And this is not the PCi-Express 3.0 that would have lead me toward the LGA2011 platform. Don't think you'll see any performance improvement with a standard single-chip graphics card because it is in PCi-E 3.0. The first performance improvement that PCi-E 3.0 will bring us will be when nVIDIA will release dual-Kepler GPU boards. Then, we might see an improvment of a couple of percent over a PCI-E 2.0 configuration, but nothing to create a scandal ... Remember when we have passed from PCi-E 1.1 to 2.0...

Sure, LGA2011 is a very, very nice platform, superb performance, but really overpriced.

Nice review Tom's Hardware !!!
Score
-2
December 9, 2011 2:24:04 PM

Oh and just like it was said above : these CPUs are mainly made for business purpose, where a performance improvment can have terrific consequences on productivity of an enterprise. In that way, then, the LGA2011 is a superb platform.

For a home-user ... I doubt that it would be very usefull regarding the price, unless you work onto a specific task.
Score
0
December 9, 2011 2:27:20 PM

assassin123wow.. . . Great intel is so good

Yes if you have the Ca$h to pay for it !! It's not the case for everybody !
Score
-3
December 9, 2011 2:34:08 PM

Now I cat wait for the next SBM. I can't see them not using the described 3930K config (perhaps w/ different Mobo and more ram) for the high-end, and it's anybody's guess what they would use for the low-end what with APUs, Zambezi and soon IB being available.

Great article, Chris. I too would have liked to have seen an OC to OC comparison, which is what I'm hoping to see in the next SBM I guess.
Score
-1
December 9, 2011 2:36:13 PM

I'd rather get a 2500k and spend the rest on beer.
Score
3
December 9, 2011 3:49:13 PM

this would be nice for my new build!
Score
-1
December 9, 2011 3:54:45 PM

FunSurferWhy are the gaming benchmarks on ultra settings where the GPU is the bottleneck? This is a CPU benchmark so resolution and settings should be on the lowest possible.


Because that's how games are actually played. Tests at 640x480 are meaningless. If you want less quality, buy a console ;-)
Score
4
December 9, 2011 4:01:57 PM

Whoever designed the captcha at the end of the competition entry form should be shot. It says "enter text exactly as displayed" when what it means is "enter the sum of the two numbers"
Score
0
December 9, 2011 4:56:09 PM

hmp_gooseThere's some error on the raffle page about proving you are a human, even though there nether graphic nor field to fill on the subject. Firefox 8.0

Probably the browser. I've had trouble with Firefox lately and their persistent updating is getting annoying.

Try Chrome or IE. It worked fine for me on IE9 (with tracking protection on, surprisingly) so maybe it will work for you too.

Score
0
December 9, 2011 5:03:59 PM

cangeliniBecause that's how games are actually played. Tests at 640x480 are meaningless. If you want less quality, buy a console ;-)


Do you actually own a console? I have the poorer performing console (Xbox) and a PC (C2Q OC'd to 3.77Ghz, 8GB DDR2 800Mhz, and a superclocked GTX 275) and yeah, my PC looks much better... If you sit real close to the TV when playing the Xbox.

Which no one in their right mind does. The minimum distance from an HD TV when playing Xbox is like 4-5 feet. Minimum distance on a PC is like 15 inches.

The only problems I notice graphically when playing my Xbox is the lack of AA. Textures look fine, effectts look fine, and you cannot build a computer from the ground up for the same price as the Xbox that performs as well as the Xbox. The Xbox is $300. Let me repeat that; the Xbox is $300.

You build me a computer (Mouse, Keyboard, OS, case, GPU, CPU, RAM, Mobo, HDD/SSD, PSU, and Disc drive; Heatsink not necessary) for $300 dollars NEW that perfoms as well as the Xbox and I will concede that the PC is better.
Score
-3
December 9, 2011 5:21:31 PM

guardianangel42Do you actually own a console? I have the poorer performing console (Xbox) and a PC (C2Q OC'd to 3.77Ghz, 8GB DDR2 800Mhz, and a superclocked GTX 275) and yeah, my PC looks much better... If you sit real close to the TV when playing the Xbox.Which no one in their right mind does. The minimum distance from an HD TV when playing Xbox is like 4-5 feet. Minimum distance on a PC is like 15 inches. The only problems I notice graphically when playing my Xbox is the lack of AA. Textures look fine, effectts look fine, and you cannot build a computer from the ground up for the same price as the Xbox that performs as well as the Xbox. The Xbox is $300. Let me repeat that; the Xbox is $300.You build me a computer (Mouse, Keyboard, OS, case, GPU, CPU, RAM, Mobo, HDD/SSD, PSU, and Disc drive; Heatsink not necessary) for $300 dollars NEW that perfoms as well as the Xbox and I will concede that the PC is better.


Sure I own a console. But this isn't about whether the console is better than the PC, it's about testing PCs correctly. And the reason we don't dial everything down to 640x480/low-quality is because that's not how PC games are played. If I buy a $600 CPU and a couple of $500 graphics cards, I'm cranking everything *up*.
Score
2
December 9, 2011 5:21:51 PM

alchemy69Whoever designed the captcha at the end of the competition entry form should be shot. It says "enter text exactly as displayed" when what it means is "enter the sum of the two numbers"


Checking into this for you!
Score
1
!