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
mcturkey - Monday, November 14, 2011 - link
Glad I'm not the only one who was thinking that! My 486 66 only had 4MB as well.Anand Lal Shimpi - Monday, November 14, 2011 - link
My 386 started with 4MB, but I kept it for a very long time as upgrading was fairly expensive. I eventually threw a ton of memory at it as my last upgrade to the platform :)Take care,
Anand
BSMonitor - Monday, November 14, 2011 - link
How big was your 386's hard drive?How many times over could you store it's entire contents in 8 DIMM's of DDR3 memory, now ?? And for probably less cost!
Thought I saw a 16GB kit on newegg for $75? Lol!
just4U - Monday, November 14, 2011 - link
I had a roomie 81 meg harddrive in my 386/16khanov - Monday, November 14, 2011 - link
"With the socket the same, is the reviewed SB-E the same design as the new Xeons? Will there be 3D design like Ivy Bridge?"1. It is the exact same die as the new Xeons, although of course different parts are harvested for each market.
2. Yes there will be a 3D transistor design (according to rumors) but this will be Ivy Bridge-E and will not launch until at least late 2012.
gamoniac - Monday, November 14, 2011 - link
Anand,More and more power users are running VMs on their desktops or workstation. With most Intel and AMD CPU now support Intel-VT or AMD-V, I notice a lack of measurement in this department in pretty much all online reviews. When you update your test suite, could you possibly include some sort VM test? Note: If so, could you possibly run the VM test on SSD to eliminate HDD limitation?
Thanks for the great review and conclusion, as always.
Senti - Monday, November 14, 2011 - link
I'm amazed how much fuss QuickSync is still generating in reviews. Let's face it - it's fairly useless in current state. Cool words "GPU video transcoding" can only impress casual users, not someone who cares about quality in first place and speed only after that.With time it will be even more useless if like GPU video decoding it's unable to work with 10 bit and 422/444 content (very likely).
gunslinger5577 - Monday, November 14, 2011 - link
This review indicates no significant improvement with 2x 16x PCI-E lanes in SLI. However the ASUS X79 Pro MB review seems to indicate there is a measurable and significant at times advantage.fishbits - Monday, November 14, 2011 - link
Weird stuff. Why fret over on-die USB 3.0 when every mobo supports it? And mourning Quick Sync for a CPU that flies at encoding without it? Or when you'd already have an SNB with Quick Sync? Really unhappy with the new Porche's glove compartment...Love the CPU/Platform, but too pricey for how much I'd use it over what's currently available. Hoping against hope that mainstream Ivy offers 8 RAM slots, but not holding my breath.
DanNeely - Monday, November 14, 2011 - link
Each addon chip the mobo makers include increases the cost of the board (not just the chip itself, but the engineering time needed to integrate it, and potentially (if enough are chips are added) by adding an extra layer to the PCB).You also take a hit in the number of PCIe lanes available for expansion slots. With legacy PCI gone from the southbridge we're unlikely to see any 4x electrical slots coming off of it. Audio, ethernet, and firewire, will take 1 lane each; USB3 controllers will take 1lane/2 ports, probably 3lanes total/board leaving only 2 for expansion slots. THe main impact here is just not being able to go all USB3 for the legacy free gloss without a major squeeze elsewhere. Scientific customers doing stuff that actually needs the PCE 3.0 bandwidth without needing 2x width cards could end up being dinged since it means several fewer total lanes for them to hook stuff up to.