Is the core i7-975 really faster than the w3580?

Discussion in 'Mac Pro' started by Roman23, Jul 15, 2010.

  1. Roman23 macrumors 6502

    Jun 12, 2010
    Hello everyone,

    In a continuing effort to see the benefits of the core i7-975, I present to you a benchmark of both the w3580 and 975.. for some reason the i7-975 beats the w3580.. have a look for yourself.

    One possible theory: ECC memory.. this slows down the computer, as it has to check for errors constantly.. where as non-ecc doesn't.

    Could it be that this is the one denominator that makes the core i7 faster? Other i7's and w35xx processors are benchmarked too.. let me know the answer to this. - actually shows the i7 975 kicking the w3580's butt.. not much though, but still something to consider.. or is it? And again, is it due to the fact of ecc memory as many have suggested, or is it due to the fact that the xeons were always slower than their desktop brothers?

  2. Umbongo macrumors 601


    Sep 14, 2006
    Those numbers don't prove the Core i7's are faster. To make the comparison you need proper testing on the same system with and with out ECC when using the Xeon. There are a lot of things that can affect the scores that are submitted to sites like that, especially as it is an average of many results.
  3. Hellhammer Moderator


    Staff Member

    Dec 10, 2008
    I agree. Using a real world benchmark such as HandBrake would be a lot better than a synthetic CPU Mark in addition to your recommendation.
  4. nanofrog macrumors G4

    May 6, 2008
    In this comparison, the clock speed is identical, as is the architecture, save the fact ECC is Disabled in the i7-975.

    Keep in mind that ECC increases the timing values, and under certain conditions, will be seen under benchmarks (synthetic), and real world under SMP based applictations that can really utilize the memory system.

    Real world benchmarks would give you a clearer picture, and the Xeon would need non-ECC RAM in it as well (keeping all other things equal). There may yet be very small variances, but would show that not all CPU's are identical in performance, even if it's the same part number on the same system (testing random samples of the same CPU P/N, with nothing else changing in the system).

Share This Page