Nehalem EX Confusion

One of the reasons that the Xeon X7560 did not show its full potential at launch was a small error in the firmware of the Dell R810 testing platform. This caused the memory subsystem to underperform. As a result some of the bandwidth sensitive benchmarks, including many HPC applications, were not performing optimally. Intel claimed that a dual CPU config should be able to reach 39GB/s, and a quad CPU configuration should reach up to 70GB/s. We could not reach those stream numbers as we test with our somewhat older stream binary as described here. Using the same stream binary as before allows us to compare our findings with all our previous measurements.

We reran our stream benchmarks on the new QSSC-S4R server system.

Stream TRIAD on 64 bit Linux—maximum threads
* New measurements.

The new results tell us that available memory bandwidth is about 21% higher (29GB/s) than what we previously measured on the DELL R810 (24GB/s). That means that many benchmarks published at the launch of the Xeon 7500 and using the Dell R810 were too low, especially the HPC ones. The Xeon X7560 will not be able to beat the quad Opteron 6174 when it comes to raw bandwidth, but it is far from a bandwidth starved platform.

The 32-Core, 64-Thread Beast Stress Testing the High End
POST A COMMENT

51 Comments

View All Comments

  • blue_falcon - Tuesday, August 10, 2010 - link

    The R715 is an AMD box. Reply
  • webdev511 - Tuesday, August 10, 2010 - link

    Yes, and the R715 has 2x AMD Opteron™ 6176SE, 2.3GHz with 12 cores per socket with an approx price of $8,000 Reply
  • fic2 - Tuesday, August 10, 2010 - link

    4. Part of the Anandtech 13 year anniversary giveaway?!! ;o) Reply
  • mino - Wednesday, August 11, 2010 - link

    Big Thanks for that ! Reply
  • Etern205 - Tuesday, August 10, 2010 - link

    *stares at cpu graph*
    ~Drrroooollllliiiieeeeeeee~~~~
    Reply
  • yuhong - Tuesday, August 10, 2010 - link

    The incorrect references to Xeon 7200 should be Xeon 7100.
    "Other reasons include the fact that some decision makers never really bothered to read the benchmarks carefully"
    You didn't even need to do that. Knowing the difference between NetBurst vs Core 2 vs Nehalem would have made it obvious.
    Reply
  • ELC - Tuesday, August 10, 2010 - link

    Isn't the price of software licenses a major factor in the choice of optimum server size? Reply
  • webdev511 - Tuesday, August 10, 2010 - link

    So does the NUMA barrier.

    I'd go for less sockets with more cores any day of the week and as a result Intel= second string.
    Reply
  • Ratman6161 - Wednesday, August 11, 2010 - link

    For the software licensing reasons I mentioned above, there is a distinct advantage to fewer sockets with more cores. Reply
  • davegraham - Wednesday, August 11, 2010 - link

    so NUMA is an interesting one. Intel's QPI bus is actually quite good and worth spending some time to get to know.

    dave
    Reply

Log in

Don't have an account? Sign up now