Intel Xeon 5570: Smashing SAP records.

[kl0012]

It might be too early to say but Intel's leap ahead with new Xeon over AMD in server area looks even more impressive then Core 2 storm some times ago.

http://www.anandtech.com/weblog/showpost.aspx?i=532

The SAP numbers are absolutely astonishing, as Intel's dual socket is able to outperform quad socket opteron machines. Based on the scaling of Barcelona, we speculate that a quad Shanghai at 2.7 GHz would obtain the performance of the Dual Xeon 5570 w/o HT.The new Xeon 5570 outperforms the "old" 5450 by 119%!!!

These numbers are so high, that we checked and checked again. The database used is the same (SQL Server 2005), so unless there is some incredible tuning parameter that HP and FS have discovered and that we have yet to hear about, that is not it.

At this point we have no idea how it is possible that a 3 GHz Nehalem outperforms the latest Opteron by a margin as high as 80% and more. But we can give it a try. In a previous server oriented article, we summed up a rough profile of SAP S&D:

• Very parallel resulting in excellent scaling
• Low to medium IPC, mostly due to “branchy” code
• Not really limited by memory bandwidth
• Likes large caches
• Sensitive to Sync (“cache coherency”) latency

One of the biggest bottlenecks for Intel has been the sync latency. It is possible that once the "sync" bottleneck was removed, the intel architecture is able to show it's real integer crunching power thanks to the out of order loads (memory disambiguation) and better branch prediction.Those are two areas where the opteron architecture is still weak.

[Katanai]

Damn, this could be the end for the green team. The only thing going for them was the Opteron...

[gosh]

At this point we have no idea how it is possible that a 3 GHz Nehalem outperforms the latest Opteron by a margin as high as 80% and more. But we can give it a try. In a previous server oriented article, we summed up a rough profile of SAP S&D:

Maybe they have used one intel compiler to compile the code that is used in the test

[gallag]

Maybe they have used one intel compiler to compile the code that is used in the test

Would SAP allow that?

[gosh]

Would SAP allow that?

I don't know, it was a bit of a joke but the compiler from Intel is very good (for Intel CPU's anyway) if I have understand it correctly. If you know that customers has mostly Intel computers then I think that it isn't that hard to see the logic in using the best compiler for that processor (if it is the best).
Maybe those who is using it have gone a bit dubious to it when there was information about the deoptimized code for AMD processors.

[Shintai]

I don't know, it was a bit of a joke but the compiler from Intel is very good (for Intel CPU's anyway) if I have understand it correctly. If you know that customers has mostly Intel computers then I think that it isn't that hard to see the logic in using the best compiler for that processor (if it is the best).
Maybe those who is using it have gone a bit dubious to it when there was information about the deoptimized code for AMD processors.

Maybe one that cripples core 2 CPUs! Since it also performs alot better than those. yep..that must be it!

[alfaunits]

Testing with the same binary is for the worse to seem better.
The end-use would use the best binary available, so testing binaries best suited for each platform on that platform is the right way.

Not that it would have anything to do with this test.
BTW, where is Shintai's thread??

[Tiridum]

reviews are nog longer allowed in the news section

[alfaunits]

This is not a review, but a "hold your walltes, SAPpers".

[anubis]

It aint such a bad news for AMD actually, because with AMD you can upgrade to latest shanghai core cpus and get the speed bump (great if switching from barcelona) with the additional decrease in power requirements (win-win in current financial situation) .
In INTELs case you have to buy everything, and you cant really be sure your brand-spanking-new systems are compatible with nextgen HW.

[Donnie27]

Update (a few hours later): It seems that the SAP page was wrong about HT. It reported 8 threads on 8 cores on the Fujitsu Siemens Primergy Server. The certification page says otherwise: 16 threads on 8 cores. So hyperthreading (SMT) plays probably an important role in this benchmark as the SAP application has very low IPC and is very parallel. So this completely annihilating performance comes from combining a wide superscalar CPU with an excellent Simultaneous Multithreading implementation. Hats off to the Intel engineers...

Hats off to the Intel engineers indeed But who's really surprised here?

[Jacky]

It aint such a bad news for AMD actually, because with AMD you can upgrade to latest shanghai core cpus and get the speed bump (great if switching from barcelona) with the additional decrease in power requirements (win-win in current financial situation) .

Actually it is extremely bad news for them. What have those performance numbers to do with upgradeability? Upgrading was always possible, it is an inherent feature of a die-shrink on an old platform, but now everyone is much less likely to upgrade. If Nehalem is that fast it's perf/$/W will clearly obliterate Shanghai, but it won't be that fast in every application as far as we know (e.g. Linpack).

[Vinas]

Yes, things looking increasingly bleak for the green team. Some of the FUD spread around here a few months ago said AMD would have a "big performance lead over intel". I was hopeful but it looks like those posts were unfounded??? Either way I hope AMD can respond with something competitive. If this is true they may even lose the supposed lead in the VM ESX market. Competition is good, but an 80% performance premium is not competition anymore.

[gosh]

Isn't the SAP test one of those tests that can differ very much depending on how the computer is configured? It also done by Anandtech and most that reads tests know that few (if any?) testers is neutral, Anandtech is one of those sites that is among the group of sites that is least neutral.

[gallag]

Isn't the SAP test one of those tests that can differ very much depending on how the computer is configured? It also done by Anandtech and most that reads tests know that few (if any?) testers is neutral, Anandtech is one of those sites that is among the group of sites that is least neutral.

The tests were not done by anandtech, They were done by HP.

[Ghostbuster]

It aint such a bad news for AMD actually, because with AMD you can upgrade to latest shanghai core cpus and get the speed bump (great if switching from barcelona) with the additional decrease in power requirements (win-win in current financial situation) .
In INTELs case you have to buy everything, and you cant really be sure your brand-spanking-new systems are compatible with nextgen HW.

That's the problem... Normally for server hardware, companies hardly or never upgrades the processor at all. The most they will upgrade is the HDDs or RAM. When they do really upgrade they simply get a brand new server.

The tests were not done by anandtech, They were done by HP.

Yups, I guess gosh never bothered to even check the link nor read it.

[savantu]

reviews are nog longer allowed in the news section

Where do you see a review ?

It is a score submitted by HP and Fujitsu.

[Hornet331]

Where do you see a review ?

It is a score submitted by HP and Fujitsu.

Indeed, if one score counts as a review all cpu/gpu and gamethreads have to vanish form the news section.

[Chad Boga]

Isn't the SAP test one of those tests that can differ very much depending on how the computer is configured? It also done by Anandtech and most that reads tests know that few (if any?) testers is neutral, Anandtech is one of those sites that is among the group of sites that is least neutral.

With Nehalem not having a FSB, what will be your next crazy angle of attack against Intel in order to try and boost AMD?

[Jacky]

Isn't the SAP test one of those tests that can differ very much depending on how the computer is configured?

I don't think so judging by the comments on realwordtech.com

It also done by Anandtech and most that reads tests know that few (if any?) testers is neutral, Anandtech is one of those sites that is among the group of sites that is least neutral.

Really? Proof or stop it, this is bordering on libel or plain FUD. I've looked hard but couldn't find any clues for unfair testing by Anandtech.

[Movieman]

With Nehalem not having a FSB, what will be your next crazy angle of attack against Intel in order to try and boost AMD?

Let's try and not bash the guy ok.
Performance speaks for it self so lets wait and watch.
Remember, Friends first.
We all love the hobby no matter if it's green or blue that grabs you personally.


HOLY CRAP BATMAN!
I just looked at that pic.
I have a dual socket X5470 Harpertown system and it s a beast and this doubles it?

[gosh]

The tests were not done by anandtech, They were done by HP.

Even worse

Intel has apparantely allowed HP and Fujitsu-Siemens to break the NDA on the Xeon 5570 processor for PR reasons as both companies have published SAP numbers on a Dual Xeon 5570.

HP are selling hardware

[Chad Boga]

Let's try and not bash the guy ok.
Performance speaks for it self so lets wait and watch.

Well for some, the "performance reported" is the result of a nefarious transfer of funds from Intel to <insert site here>, as evidenced by his comments on Anandtech.

[gallag]

Even worse

Intel has apparantely allowed HP and Fujitsu-Siemens to break the NDA on the Xeon 5570 processor for PR reasons as both companies have published SAP numbers on a Dual Xeon 5570.

HP are selling hardware

lol I just dont know what to say.

[Clairvoyant129]

Even worse

Intel has apparantely allowed HP and Fujitsu-Siemens to break the NDA on the Xeon 5570 processor for PR reasons as both companies have published SAP numbers on a Dual Xeon 5570.

HP are selling hardware

LOL keep trying gosh.