AMD's performance comparisons for the new Opteron 6100

I’ve had several requests for links to benchmark data on the new AMD and Intel processors that have launched over the past couple weeks. We added a little bit of information with our posting of SGI’s SPEC results for its forthcoming UV line running the Xeon 7500s. Yesterday one of AMD’s marketing drones pointed to some SPEC results for the 6100 from the AMD blogs.

You can get over 40 pages of benchmark data on the new Opterons at AMD’s performance portal, www.amd.com/opteronperformance (select the AMD Opteron 6100 Series Processor and then the power bands and workload you in which you are interested). Here are some highlights (pay attention to how hard they are hitting Intel’s products, calling them by name rather than the more nebulous “the competition”)

1. Servers using AMD Opteron processor Model 6174 (12-core 2.2GHz) have higher SPECint®_rate2006 and SPECfp_rate2006 performance (at a lower processor price) than servers using Intel Xeon processor Model X5680 (6-core 3.33GHz)…

2. Servers using AMD Opteron™ processors Model 6136 (8-core 2.4GHz) have higher SPECint_rate2006 and SPECfp_rate2006 performance (at a lower processor price) than servers using Intel Xeon processor Model X5570 (4-core 2.93GHz)….

3. Value Four-Socket Servers using four AMD Opteron processors Model 6136 (8-core 2.4GHz) have higher SPECint_rate2006 and SPECfp_rate2006 performance (at a lower total processor price) than servers using two Intel Xeon processors Model X5680 (6-core 3.33GHz).

There are many, many other comparisons both in this post and in the published benchmark data, though you have to read between the lines (and bring your salt grains) since these numbers are from the vendor, ableit at least some of this data is reported within the context of the SPEC process.

In particular you might be interested in the HPC Application Performance Summary chart for two socket servers from the performance pages; covered are LS-DYNA, GROMACS, and FLUENT.
AMD/Intel comparison

These comparisons were obviously run before yesterday’s launch of Intel’s 8-core platform, and I suspect that a head-to-head between Beckton and Magny Cours would not skew so heavily in AMD’s favor.

Trackbacks

  1. […] AMD’s performance comparisons for the new Opteron 6100 (insidehpc.com) Tags: 6-core processors, AMD Opteron, Quad-Core […]

Comments

  1. Hi John,

    FYI, the SPEC.org website now has some results, too – and I expect we’ll see Nehalem-EX / Beckton scores on their shortly. IBM listed some SPECfp_rate scores for its x5 systems using the Nehalem-EX CPUs, but they don’t yet have the full breakdown by benchmark – it’s been submitted, but not yet approved by SPEC. I expect we’ll see it within the next few days.

    For example:
    http://www.spec.org/cpu2006/results/res2010q1/cpu2006-20100315-10005.html

    I personally find these detailed results much more interesting, since I can look at MILC, WRF, etc., and see how those particular benchmarks differ, and ignore ones that I’m less interested in.

    Cheers,
    – Brian

  2. Mark Hahn says

    Beckton vs M-C? the impression I have is that Beckton is very much aimed at the mainframe market (defacto replacement of Itanium) – lots of RAS and other expensive features. giant chip on a conservative fab process, etc. is there some reason to compare Beckton with M-C?

  3. John West says

    We are in the process of looking at the merits and demerits of building large clusters out of all of the new crop of chips, so it is of interest to the day job to know how Beckton performs absolute value. Not everyone builds on price/performance…there are some applications where performance is important enough that a small price delta against Beckton may be worth it if the performance is good enough.