Intel Core i7 3960X (Sandy Bridge E) Review: Keeping the High End Alive
by Anand Lal Shimpi on November 14, 2011 3:01 AM EST- Posted in
- CPUs
- Intel
- Core i7
- Sandy Bridge
- Sandy Bridge E
The Test
To keep the review length manageable we're presenting a subset of our results here. For all benchmark results and even more comparisons be sure to use our performance comparison tool: Bench.
| Motherboard: |
ASUS P8Z68-V Pro (Intel Z68) ASUS Crosshair V Formula (AMD 990FX) Intel DX79SI (Intel X79) |
| Hard Disk: |
Intel X25-M SSD (80GB) Crucial RealSSD C300 |
| Memory: | 4 x 4GB G.Skill Ripjaws X DDR3-1600 9-9-9-20 |
| Video Card: | ATI Radeon HD 5870 (Windows 7) |
| Video Drivers: | AMD Catalyst 11.10 Beta (Windows 7) |
| Desktop Resolution: | 1920 x 1200 |
| OS: | Windows 7 x64 |
Cache and Memory Bandwidth Performance
The biggest changes from the original Sandy Bridge are the increased L3 cache size and the quad-channel memory interface. We'll first look at the impact a 15MB L3 has on latency:
| Cache/Memory Latency Comparison | ||||||
| L1 | L2 | L3 | Main Memory | |||
| AMD FX-8150 (3.6GHz) | 4 | 21 | 65 | 195 | ||
| AMD Phenom II X4 975 BE (3.6GHz) | 3 | 15 | 59 | 182 | ||
| AMD Phenom II X6 1100T (3.3GHz) | 3 | 14 | 55 | 157 | ||
| Intel Core i5 2500K (3.3GHz) | 4 | 11 | 25 | 148 | ||
| Intel Core i7 3960X (3.3GHz) | 4 | 11 | 30 | 167 | ||
Cachemem shows us a 5 cycle increase in latency. Hits in L3 can take 20% longer to get to the core that requested the data, if this is correct. For small, lightly threaded applications, you may see a slight regression in performance compared to Sandy Bridge. More likely than not however, the ~2 - 2.5x increase in L3 cache size will more than make up for the added latency. Also note that despite the large cache and thanks to its ring bus, Sandy Bridge E's L3 is still lower latency than Gulftown's.
| Memory Bandwidth Comparison - Sandra 2012.01.18.10 | |||||
| Intel Core i7 3960X (Quad Channel, DDR3-1600) | Intel Core i7 2600K (Dual Channel, DDR3-1600) | Intel Core i7 990X (Triple Channel, DDR3-1333) | |||
| Aggregate Memory Bandwidth | 37.0 GB/s | 21.2 GB/s | 19.9 GB/s | ||
Memory bandwidth is also up significantly. Populating all four channels with DDR3-1600 memory, Sandy Bridge E delivered 37GB/s of bandwidth in Sandra's memory bandwidth test. Given the 51GB/s theoretical max of this configuration and a fairly standard 20% overhead, 37GB/s is just about what we want to see here.
Facebook
Twitter
RSS
163 Comments
View All Comments
ggathagan - Monday, November 14, 2011 - link
You assume correctly.As stated in the article, the 3820 is due out early next year and expected to run at about $300.
If you look at the 1st page of the article, you'll note that the 3820 is a little faster than the 2700K, with the same max in turbo mode and a larger L3 cache.
DanNeely - Monday, November 14, 2011 - link
Except that the 2700k is unlocked and the 3820 has a severe overclocking limit.theangryintern - Monday, November 14, 2011 - link
Thanks for the great review, Anand. I had been waiting for SNB-E to do an upgrade from my X58 Core i7, but now I'm thinking of saving some money and going with a regular Sandy Bridge, the gaming gains just aren't enough to justify the added expense.Makaveli - Monday, November 14, 2011 - link
That makes no sense there are no gaming games going from X58 to SB!You will still be GPU bottlenecked on most games!
And a whole new build will be an added expense for no gain in Games!
Beenthere - Monday, November 14, 2011 - link
Not much to see here except over-priced CPUs and mobos. Nice to see Intel fans smartening up and passing on these cash cows.Lazlo Panaflex - Monday, November 14, 2011 - link
Hi Anand, thanks for the review :)Apologize in advance if this was asked earlier, but what specifically is your criteria for determining a stable overclock? For example, do you run Prime95, large FFT's for a predetermined period of time, or perhaps IBT (Intel Burn Test) for a certain amount of runs? Or do you utilize some other tool? Just curious, since this question often pops up in the CPU forums, and everyone has their own opinion of what constitutes a stable overclock.
Regards,
LP
yankeeDDL - Monday, November 14, 2011 - link
LOL. Supply and the mand is the cause. Jacking up the price is the effect. You are a bit mixed up it seems.Also, in the consumer market having the fastest CPU (or GPU) that you sell for a ridiculous price doesn't mean that someone buys it.
How many people buy a HD6990? Few, and that's why supplies are so scarce.
The 3960X is a "show off" chip that claims performance crown. Anyone in the right state of mind will not buy it: it is just not worth the money it costs, unless you're in absolute need of few extra % of performance.
JlHADJOE - Tuesday, November 15, 2011 - link
We have chips priced at $1000 because the market has shown that it is willing to pay that amount to get the top-performing chip. It doesn't matter that AMD doesn't have an entry in that segment, because if it did, then we'd probably have AMD's FX-9300 or something priced at $900, while Intel sells their 3960X at $1100.This was the exact case when AMD was competitive, and their FX-57 was sold at $1100, vs Intel's Pentium EE which was going for $999. Was there a competitive AMD at the time? Yes. They were even in the lead. Were prices still jacked up? Yes.
The $1000 CPU will only go away if we, as consumers wise up and say we are not willing to pay that much money for a chip.
karkas - Monday, November 14, 2011 - link
Rapid STORY Technology?GTVic - Monday, November 14, 2011 - link
Lack of Quick Sync is not nearly the negative that the reviewer seems to think it is. It is not a well supported technology and not many people would use it on a day to day basis. This shouldn't even be mentioned in the article unless you also want to bring up support for Intel Viiv http://en.wikipedia.org/wiki/Intel_Viiv.