Sign in with
Sign up | Sign in

Benchmark Results: Content Creation

Intel Core i7-3960X (Sandy Bridge-E) And X79 Platform Preview
By

It's always interesting to get hands-on time with unreleased hardware. We were recently able to benchmark Intel's upcoming Core i7-3960X CPU, comparing it to Core i7-990X, Core i7-2600K, and AMD's Phenom II X6. Will you be in line for Sandy Bridge-E?

Our threaded filter test favours the Core i7-990X over Intel’s Core i7-2600K. But when you fold in Sandy Bridge’s IPC advantages to a six-core architecture, the result is even better (if only by three seconds, in this case).

Normally this benchmark would be a drawn-out workstation-class test capable of making clear the benefits of one processor over another. However, Adobe is slowly extending hardware support for its Mercury Playback Engine to more Nvidia graphics cards.

As a result, our GeForce GTX 580 kicks in, cutting what used to be a 15-minute benchmark on Intel’s Core i7-990X down to 38 seconds. CPUs still influence the overall result, though, and the field finishes just as we’d expect.

The only real loser in After Effects is AMD’s Phenom II X6, which lags behind the rest of the closely-grouped field.

Architectural improvements help Sandy Bridge keep up with Gulftown in our Blender test. Sandy Bridge-E combines the best aspects of both to deliver a new top score.

Core i7-3960X walks away with the highest Cinebench CPU score. That much was expected.

It’s interesting, though, that Sandy Bridge and Sandy Bridge-E both facilitate better OpenGL-based rendering than Core i7-990X, particularly because Maxon specifically aims to isolate graphics card performance in this part of the metric:

“The graphics benchmark in Cinebench is designed to minimize the influence of other system components. All geometry, shaders, and textures are stored on the graphics card prior to measurement, and no code is loaded during the measurement process. This minimizes the system influence, but unfortunately cannot eliminate it entirely.”

Display all 11 comments.
This thread is closed for comments
  • 0 Hide
    AdrianPerry , 12 September 2011 16:15
    It's a shame Deus Ex: Human Revolution wasn't included in the benchmark testing since we know CPU plays a very big role in the FPS the game can deliver. Other than that, interesting article :) 
  • 0 Hide
    mathew7 , 12 September 2011 18:34
    On the "Hardware Setup and Benchmarks" page, please correct the "Active cores" row with the rest of the table. It should be 1/2 active cores have 6 TB bins and 5/6 have 3 TB bins. After all, it's with 1 active core that 3.9GHz is allowed.
  • 0 Hide
    CPU666d1 , 12 September 2011 21:24
    Bad luck I can't afford to buy any of these Sandy-E chipsets yet. The Benchmarks ain't too bad either.
  • 1 Hide
    ps3hacker12 , 12 September 2011 22:02
    too bad its likely to cost three to four times what the current sandy bridge setup costs (a 2600K for example...)
  • 0 Hide
    icehot , 15 September 2011 16:47
    Hrm, so basically a little bit of improvement, but mostly extra pci-express lanes, and 2 extra cores, along with additional power draw... Think i'll stick to my 2600k until ivybridge comes out and see what that brings.
  • -1 Hide
    mathew7 , 15 September 2011 18:00
    icehotHrm, so basically a little bit of improvement, but mostly extra pci-express lanes, and 2 extra cores, along with additional power draw... Think i'll stick to my 2600k until ivybridge comes out and see what that brings.


    You forgot about the additional 2 chanels of RAM. Consider 4 identical modules. With LGA1155, the additional 2 modules (3 and 4) offer only additional RAM, but do not affect performance. On LGA2011, the additional 2 modules also increase memory access speed (think about a 4-lane highway vs 2-lane).
  • 0 Hide
    icehot , 15 September 2011 19:02
    mathew7You forgot about the additional 2 chanels of RAM. Consider 4 identical modules. With LGA1155, the additional 2 modules (3 and 4) offer only additional RAM, but do not affect performance. On LGA2011, the additional 2 modules also increase memory access speed (think about a 4-lane highway vs 2-lane).


    That's cool, except I have 2 sticks of 4gb's, I still can't see a need to go beyond 8gb (at home at least). Despite this though, looking at the benchmarks is incredibly disappointing, clock for clock it's generally the same as the 2600k, the only improvement is if something requires 6 cores... For gaming, we all know the additional pci express lanes even in crossfire makes very little difference, maybe a bit more in quadfire, but still, not sure if it's worth it
  • 0 Hide
    Gonemad , 21 September 2011 04:25
    I am more concerned if I should start my milling machine for a new copper brick matching the new socket and lineup. We are back at 130W TDP guys. The last generation on that TDP was less than forgiving on air cooling. Or, should it be no concern? Still, that i7-2600k is still crushing the bang-per-buck charts. The new chip should be smashing the previous ones with over 20% difference; instead they are trading punches here and there.
    From my POV, the 2600k is still the best upgrade right now.
  • 0 Hide
    bobwya , 26 September 2011 23:14
    mathew7You forgot about the additional 2 chanels of RAM. Consider 4 identical modules. With LGA1155, the additional 2 modules (3 and 4) offer only additional RAM, but do not affect performance. On LGA2011, the additional 2 modules also increase memory access speed (think about a 4-lane highway vs 2-lane).


    Hey dude,

    Hate to break it to you but if you don't need all that memory bandwidth then it's like buying a Ferrari to pootle about town - at 30mph.

    Don't forget that for regular Desktop CPU stuff (web browsing, gaming, etc.) it's memory latency that counts - not bandwidth...

    CPU Architecture #101

  • 0 Hide
    Anonymous , 1 November 2011 22:42
    Large amounts of RAM is great for HD Video editing and even more so for the big sample libraries people who make music on their PC use (eg. NI Kontakt, Vienna Symphonic Library, Synthogy Ivory).

    It would be great if all future CPU benchmarks included the DAWbench benchmark. Especially those that feature products not targeted at gamers but at creative people.
  • 0 Hide
    mathew7 , 17 December 2011 07:13
    Quote:
    Hey dude,

    Hate to break it to you but if you don't need all that memory bandwidth then it's like buying a Ferrari to pootle about town - at 30mph.

    Don't forget that for regular Desktop CPU stuff (web browsing, gaming, etc.) it's memory latency that counts - not bandwidth...

    CPU Architecture #101

    Except that the examples you gave, even memory latency does not affect it much. It's hidden pretty good behing cache.
    But turn to compiling a big project on an SSD(like Android 4 which they say requires 16GB or RAM), and the additional 2 channels give a nice boost. Also video compression scales good with bandwidth.


    And those that have the money and go for "Extreme", those would drive a Ferrary in town at 30mph (ok, maybe 60 if traffic is light).