5 Replies Latest reply on Aug 1, 2012 10:46 AM by Jim-Lar

    Xenon E5-2687W vs i7-3960x


      I'm about to build a system for PP/AE CS6 and I’m considering to build it up on dual  E5-2687W processors.


      Something like this:


      Processor: 2x Xeon E5-2687W

      Motherboard: ASUS Z9PE-D8 WS, 2xSocket-2011

      GPU: ASUS GeForce GTX 680 2GB

      HDD: 4x SATA 3.0, 64MB Cache, 7200RPM, 3.5

      SSD: INTEL 520 SERIES 2.5" 480GB

      RAM: 8x8GB 1600mhz

      + the other necessaries


      Due to the expensive cpu-cost in a dual E5-2687W setup I have an idea to cut the initial cost by beginning with one cpu and keep the option to upscale to two cpu's if needed in the future.


      The question is how will a setup with only one E5-2687W cpu stack up to an i7-3960X setup simular to the one above? Strictly by performance in PP/AE CS6, not taking the price tag in consideration.


      According to this test the DUAL E5 seem to give a real advantage in most cases:
      http://www.overclock3d.net/reviews/cpu_mainboard/asus_z9_pe-d8_ws_dual_xeon_insanity_e5-26 60_e5-2687w/7

        • 1. Re: Xenon E5-2687W vs i7-3960x
          Harm Millaard Level 7

          With a single CPU, it will be about equal to a single 3960X at stock speed.

          With dual CPU's, it may be about equal to a single 3930 or 3960 overclocked to around 4.6 GHz.

          • 2. Re: Xenon E5-2687W vs i7-3960x
            Jim-Lar Level 1

            Thanks for the reply Harm.

            I have read a lot of your configurations and post and it have given me a good insight on what works for an editing system for CS6. Great work!


            How does it come that the benchmarks on the screens below, rates the dual E5-2687W almost dubble the performance due to a i73960k overclocked to 4.6ghz?


            Does not PP and AE benefit from the extra 10 cores?

            Does PP and AE depend more on clockspeed than more cores?





            (link to the benchmarks here: http://www.overclock3d.net/reviews/cpu_mainboard/asus_z9_pe-d8_ws_dual_xeon_insanity_e5-26 60_e5-2687w/7)


            It also seems that companies that make custom videoediting systems uses dual Xenong E5 processors for their fastest machines. For example ADK, How does that come?

            • 3. Re: Xenon E5-2687W vs i7-3960x
              Harm Millaard Level 7

              First of all, the benchmark you refer to are artificial benchmarks, Sandra, Aida and Cinebench and they do measure raw computing power, but not real-life performance. In the PPBM5 Benchmark we try to show what real-life performance is about, using real clips, a real project, in a real situation that editors encounter in their daily workflow.


              Notice that when you look at the current results (970 systems) there are only two dual CPU systems in the top-50, one overclocked to 4.0 GHz at rank #1 (something that Intel has disabled with the E5 series and at the huge price of $ 20K) and the other one at rank #16, all the rest are single CPU systems and when sorted on RPI (Relative Performance Index) the picture does not really change. There are 105 Xeon systems, of which 74 are dual CPU systems.


              I know that BFTB (bang-for-the-buck) is not relevant to your question, you have excluded that in your originial post, but for me it is a definite factor to consider. See Adobe Forums: Planning / building a new system. Part 1


              I just don't have sufficient data to say that the E5-2687W performs better or worse than a i7-3960X at stock speed, Lasvideo should submit his results ASAP. Of course number of cores play a role, as does the size of the L3 cache, but they may be offset by clock speed.

              • 4. Re: Xenon E5-2687W vs i7-3960x
                ECBowen Most Valuable Participant

                If you decide on starting with 1 CPU then you can only populate 4 banks of ram. The 2nd CPU is required to populate the other 4 banks since the memory controller is in the CPU's.



                • 5. Re: Xenon E5-2687W vs i7-3960x
                  Jim-Lar Level 1

                  Thanks again Harm...

                  for clarifying the real-life performance vs artificial benchmarks. Artificial ones should be banned


                  Wy I brought this up here was to get some insight from the community (and from Adobe) that actually is using the apps I use. From my understanding it seems that the performance you get out of one system can differ a lot depending on how the software is adapted to the systems benefits not how general benchmarks score. With all the different benchmarks and advertisement it gets kind of confusing...


                  The dilemma is that on the one hand there are tests like the PPBM5 Benchmark that shows how little if none you gain from double price systems like E5-2600 platforms.


                  On the other hand there is the big ones. For example HP and Nvidia where Adobe clearly states that the CS6 software are optimised for the E5-2600 processors and that the very expensive Nvidia GPU’s (Maximus conf.) are multiple times faster and lets you work in nearly realtime.


                  However I have not seen any test that compares these expensive systems head to head to the ones that give you the best bang-for-the-buck. Which are actually better systems for PP and AE AND how much better are the one from the other in real life situations?


                  This brings me to the BFTB relevancy. This is of course a very important part when investing new hardware. However our company only have only 2 editor/vfx guys and a lot of work banging on our door.

                  We are more looking for ROI (Return Of Investment). For us it is mainly about how much we can edit in ”real time”. Final output is rarely any problem because this can be scheduled at non working ours.

                  But the more we can do without waiting for calculations (for example rendering) in postproduction, the more work we can take on.

                  Even if the investment cost is the double it brings us greater ROI - if we can do significantly more in less time.

                  Then again there is always a threshold where the investment cost gets to high...


                  It would be very clarifying to see a comparison between the BFTB system and an expensive one.


                  Simply, I just want to get the most out of PP an AE and focus on getting creative, but that seems to be easier said than done...