Unofficial Benchmarks Show Intel Core i7-4960X Ivy Bridge-E Performing 10% Faster Than i7-3970X user “Toppc” has posted some initial benchmarks comparing a Core i7-3970X Sandy Bridge-E to an engineering sample of an i7 Ivy Bridge-E configured to match the upcoming Core i7-4960X processor.

Using Toppc's system configuration (tabulated below), the i7-4960X performed 5 percent to 10 percent better across the board than its Sandy Bridge-E predecessor. Specifically, the i7-4960X scored 10.94 points in CineBench vs. i7-3970X's 10.16 points, 561 vs. 533 points in CPU Mark, 38,644 vs. 35,804 in 3D Mark Vantage, and in WPrime it crunched 32 M in 4.601s vs. 5.01s for the i7-3970X. 

Swipe to scroll horizontally
Header Cell - Column 0 Sandy Bridge-EIvy Bridge-E
CPUCore i7-3970XCore i7-4960X Equivalent
CPU Clock3.50 GHz (Base) / 4.00 GHz (Turbo)3.50 GHz (Base) / 4.00 GHz (Turbo)
MotherboardMSI X79A-GD45 PlusMSI X79A-GD45 Plus
GPUGeForce GTX 480GeForce GTX 480

We can’t say we’re surprised to see these results and are certainly looking forward to Intel’s next generation of CPUs. Screenshots from the benchmarks are available below.

  • powerincarnate
    All that time for 10%, that sucks. It's been 2 years since Sandy Bridge E came out.
  • vmem
    hopefully they at least fix and update all the mobo issues with Sandy E. I want to upgrade from my OCed 2500K because I'm starting to do some processor heavy work, and it seems that neither 4770k nor 4960X are too interesting atm. should've just bought a 3960x or something
  • g-unit1111
    Sounds about exactly what I'd expect anymore from one generation to the next.
  • InvalidError
    10722041 said:
    All that time for 10%, that sucks. It's been 2 years since Sandy Bridge E came out.
    What else did you expect? SB to IB only had 5-10% improvement on mainstream LGA1155 chips so it follows naturally that the same would occur on LGA2011.

    Based on leaked Haswell numbers, you can expect Haswell-E to be yet another "only ~10%" over IB-E as well.

    The majority of cost-effective and power-efficient clock+IPC improvements are tapped out. We will most likely never see the days of 40%/year improvements ever again unless Intel and AMD decide to start a core count war... but such a war is pointless when almost no mainstream software is capable of making meaningful use of it.
  • oraygungor
    If you really compared it, it is not even 10%...
  • powerincarnate
    Well, By the time Ivy Bridge E comes out, it will be two year, that is a long time in processor world. I miss the days of Moore's law of getting like a doubling of performance every 18 months.
    I remember having a 500 mhz Pentium3 and then like 2 years later, got an Athlon XP 1600+ and that was like night and day then.
    I remember like 2 years later getting an Athlon 64 3200 and that was like night and day compared to the previous one.
    I remember going from That to Core 2 due E 6600, and again, that was like night and day.
    I remember going from that to Core 2 Quad Q9550, and that was a lot bettter, but more importantly, I didn't go that route, but I remember the folks who went from a Core 2 duo like I had and went to the Core I7 processors, and they had a huge jump.
    For me, I took the route of coure 2 quad first, and then went to the Sandy Bridge E 3820, but 2 years is a lot of time, and to only get 5-10% performance then, that sucks.
    For the GPU days, 2 years use to be two generation, so again, same huge improvements like for me it was going from Rage card to Radeon 64 DDR, then I had the Radeon 9700 pro,
    Then the 8800 GT (nVidia),
    then I was stuck because I still had an AGP motherboard,
    finally when I switched to the Core 2 Duo, I also got a Radeon 4870, and switched it shortly after for the 5850, and now the 7970.
    It seems AMD and Nvidia are both in a rut as well, because the next general is going to be late by 2 years, and I don't have high hopes for the old days of massive gains.
  • rolli59
    Small steps just means that our current hardware is going to be competitive longer.
  • InvalidError
    10722168 said:
    I miss the days of Moore's law of getting like a doubling of performance every 18 months.
    Moore's law was about transistor count, not performance. The combination of clock rate increases and extra transistors just happened to roughly double performance at a similar rate for about 15 years or so.

    The clock rate part of the performance gain equation hit a brick wall almost a decade ago with the market shift towards power-efficiency while the transistor count is mostly driven by IGP and cache these days because there are few other cost-effective uses for the growing transistor budget on mainstream CPUs.

    Unless mainstream software starts actually using multi-core CPUs, even transistor counts may end up brick-walled by lack of software to justify it in the mainstream market.
  • I know its a CPU benchmark but a GTX 480 in the test run, really? This won't be worth an upgrade over Ivy but for those of us that are on 2 year old cpus (2600k) like me it should be.
  • warezme
    Isn't 10% considered within the margin of error so , no change. So sad AMD lost their competitive edge cause now we're stuck with no CPU performance progress.