What happened to the 5570 bars on the chart? As of 12/23/08 they disappeared (I remember seeing them in the original article when I viewed it.) Did anandtech get slapped under some non-disclosure of some kind? Shouldn't the article be updated or the graph yanked altogether???
"This kind of sounds like one of those 'Intel-approved' articles."
No it doesn't. It's merely pointing out the results that are out there. Intel is winning hands down in performance so its logical that the review sites would be drooling all over it.
There's nothing in the near term that tells AMD will bring sort of changes.
In one way, for some business situation, cost doesn't matter. But, this doesn't mean it should be wholly ignored. SO, given Intel tendencies to overprice, can we get a pretty, shiny graph of SAPS/dollar?
"Basically, there are two classes of SAP-SD 2-tier submissions - "fast" with response time around 1 second and "throughput-oriented" with response time around 1.6-2 seconds."
The difference between the two results your friend put are the response times are also ~2x the difference.
Although i do not know much about SAP benchmarks, i tend to agree with you. In 15 years working in IT, i have yet to see a speed bump on this kind. 119% is a lot of improvement. 20% was very nice already, 119% it is just too good to be true.
Several were announced in several fields, The Willamette Core was supposed to be a big bang, and advertised as it. And others of course.
Anyway, the same results could be achieved with a web server. You just need to know how to tinker.
But 119% ? The dice must be rigged anywhere in pipe. Even if they cherry picked what test they did.
I think HP benchmarks are well tuned both on AMD and Intel Platform.
Benchmark value of 2384 on HP servers are inline with AMD's claim (30-35% betwer than Barcelona)
I think that the Core architechture stars to shine with the add of the Memory controller on die. Having memory controller outside gives you flexibility in the mainboard/chipset selection, but you pay this with latency. Now the improvements in the Nehalem (wider execution units, HT, blah blah blah) plus the Ondie memory controller gives the CPU all the bandwidth neccesary to has the CPU very busy crunching integers.
It was well documented that adding the memory controller on die to any cpu boost the performance, so i think that this record was expected by intel engineers...but not with this huge margin.
Tell that to the guy who won the $207 million lotto this past weekend....
Not surprising really. Wolfdale dual-cores were always competitive against quad-core Phenoms... Now you have removed the one thing keeping Core processors from scaling as well as K10... ie the FSB.. Especially in a highly threaded application, as the writer mentions.. Shows how data starved Penryn really was!
I agree. Still this is a certified by SAP benchmark, and one that is mostly CPU limited. I don't see how you can "cheat" on this one. It is not like you can recompile the SAP code.
both systems have HT on, check the detailed scores.
to good to be thrue??? no, just obvious that HT is working fine on this SAP benchmark, count 70-80% off when you shut it down. Weather or not if that is required in real life SAP environments is yet to be shown.
Good point, I updated the blog post. Well, when I see a +100% boost over the previous generation we have to be prudent.
I don't think 70% is a result of HT. Doubling the cores gives you a 70% increase, and there is no way that HT can be as good as doubling the cores. I expect 40% to be more realistic. Still, it is incredible how a dual machine is capable of defeating a quad server which is only a few months older.
The numbers are certainly high, but I think it could be possible. Compared to Dual Opteron 8384, the new Xeon is about 67% faster clock-for-clock based on this benchmark, which isn't too out of whack considering all the changes made to eliminate the various system bottlenecks and HT, since Intel CPU itself (by that I mean capabilities of the CPU only, such as its wider execution core, etc) has always been more powerful than the AMD counterpart.
If this is indeed true, though, it'd mean that Intel will wipe out AMD from their coveted 4 and 8 socket server market even with the new Opteron based on K10.5 architecture. Very scary...
I have been a hardware journalist for 10 years now, and I never seen this. A new CPU + platform doubles the performance over a previous one without: 1) Using new instructions 2) a newer process technology 3) large jump in clockspeed or 4) running a very exotic benchmark that stresses only a very small part of the CPU.
Sufficient bandwidth of the POWER6 results very good scaling with SMT not only in SAP but Spec tests as well. Nehalem's increased bandwidth could be a reason for the good scaling with SMT in this case.
My guess would be that they screwed up the number of cores (dual = quad)... That would bring it down to expected gains and figures...
Altough if the performance is indeed correct... The i7 based serverchips will be the fastest cpu's in the servermarket for a very long time... And that might be a very bad thing for AMD and the microprocs industry in general.
Customers will expect it to be in parity with what is offered breitling watches by AT&T and other GSM carriers. That said, the Journal story said the upgraded breitling watches GSM iPhone is being made by Taiwanese contract manufacturer Hon Hai Precision Industry, which produced Apple's breitling watch previous breitling watches iPhones. And the CDMA iPhone model is being made by Pegatron Technology. http://www.watchvisa.com/breitling-watches.html
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
28 Comments
Back to Article
androticus - Tuesday, December 23, 2008 - link
What happened to the 5570 bars on the chart? As of 12/23/08 they disappeared (I remember seeing them in the original article when I viewed it.) Did anandtech get slapped under some non-disclosure of some kind? Shouldn't the article be updated or the graph yanked altogether???stimudent - Wednesday, December 17, 2008 - link
This kind of sounds like one of those 'Intel-approved' articles.IntelUser2000 - Friday, December 19, 2008 - link
"This kind of sounds like one of those 'Intel-approved' articles."No it doesn't. It's merely pointing out the results that are out there. Intel is winning hands down in performance so its logical that the review sites would be drooling all over it.
There's nothing in the near term that tells AMD will bring sort of changes.
alphadog - Wednesday, December 17, 2008 - link
In one way, for some business situation, cost doesn't matter. But, this doesn't mean it should be wholly ignored. SO, given Intel tendencies to overprice, can we get a pretty, shiny graph of SAPS/dollar?
ordoequester - Wednesday, December 17, 2008 - link
So whatA 2 Tier T2 from Sun gets 20.900 SAPS
And thats only a 1.4 Ghz 65nm Produkt
IntelUser2000 - Wednesday, December 17, 2008 - link
Here's the explanation from our experts at RealWorldTech:http://www.realworldtech.com/forums/index.cfm?acti...">http://www.realworldtech.com/forums/ind...=95155&a...
"Basically, there are two classes of SAP-SD 2-tier submissions - "fast" with response time around 1 second and "throughput-oriented" with response time around 1.6-2 seconds."
The difference between the two results your friend put are the response times are also ~2x the difference.
yasinag - Wednesday, December 17, 2008 - link
Could be true as SUN also activated Unicode (15% additional Load).In the past HP used to publish their disk setup and always used RAID0.
RadnorHarkonnen - Wednesday, December 17, 2008 - link
Although i do not know much about SAP benchmarks, i tend to agree with you. In 15 years working in IT, i have yet to see a speed bump on this kind. 119% is a lot of improvement. 20% was very nice already, 119% it is just too good to be true.Several were announced in several fields, The Willamette Core was supposed to be a big bang, and advertised as it. And others of course.
Anyway, the same results could be achieved with a web server. You just need to know how to tinker.
But 119% ? The dice must be rigged anywhere in pipe. Even if they cherry picked what test they did.
IntelUser2000 - Wednesday, December 17, 2008 - link
Looks like Nehalem is about to shake the server market...yasinag - Wednesday, December 17, 2008 - link
I think HP benchmarks are well tuned both on AMD and Intel Platform.Benchmark value of 2384 on HP servers are inline with AMD's claim (30-35% betwer than Barcelona)
Yasin
Pablitus - Tuesday, December 16, 2008 - link
I think that the Core architechture stars to shine with the add of the Memory controller on die. Having memory controller outside gives you flexibility in the mainboard/chipset selection, but you pay this with latency. Now the improvements in the Nehalem (wider execution units, HT, blah blah blah) plus the Ondie memory controller gives the CPU all the bandwidth neccesary to has the CPU very busy crunching integers.It was well documented that adding the memory controller on die to any cpu boost the performance, so i think that this record was expected by intel engineers...but not with this huge margin.
wpapolis - Tuesday, December 16, 2008 - link
Yes, indeed, "Head's off those Intel Engineers!"How dare they?
Bill
icrf - Tuesday, December 16, 2008 - link
Heads off sounds more like they're on the chopping block.zsdersw - Tuesday, December 16, 2008 - link
Don't you mean "hats off"? I don't think the Intel engineers should have their heads taken off for this stellar result :)JohanAnandtech - Tuesday, December 16, 2008 - link
ouch. Fixed :-).Trisagion - Tuesday, December 16, 2008 - link
If it's too good to be true, it probably is...BSMonitor - Tuesday, December 16, 2008 - link
Tell that to the guy who won the $207 million lotto this past weekend....Not surprising really. Wolfdale dual-cores were always competitive against quad-core Phenoms... Now you have removed the one thing keeping Core processors from scaling as well as K10... ie the FSB.. Especially in a highly threaded application, as the writer mentions.. Shows how data starved Penryn really was!
JohanAnandtech - Tuesday, December 16, 2008 - link
I agree. Still this is a certified by SAP benchmark, and one that is mostly CPU limited. I don't see how you can "cheat" on this one. It is not like you can recompile the SAP code.duploxxx - Tuesday, December 16, 2008 - link
both systems have HT on, check the detailed scores.to good to be thrue??? no, just obvious that HT is working fine on this SAP benchmark, count 70-80% off when you shut it down. Weather or not if that is required in real life SAP environments is yet to be shown.
JohanAnandtech - Tuesday, December 16, 2008 - link
Good point, I updated the blog post. Well, when I see a +100% boost over the previous generation we have to be prudent.I don't think 70% is a result of HT. Doubling the cores gives you a 70% increase, and there is no way that HT can be as good as doubling the cores. I expect 40% to be more realistic. Still, it is incredible how a dual machine is capable of defeating a quad server which is only a few months older.
amazi - Wednesday, December 17, 2008 - link
Now both web-page and pdf shows that FSC had HT on (16 threads). So you need to correct the chart.Wernte - Tuesday, December 16, 2008 - link
The numbers are certainly high, but I think it could be possible. Compared to Dual Opteron 8384, the new Xeon is about 67% faster clock-for-clock based on this benchmark, which isn't too out of whack considering all the changes made to eliminate the various system bottlenecks and HT, since Intel CPU itself (by that I mean capabilities of the CPU only, such as its wider execution core, etc) has always been more powerful than the AMD counterpart.If this is indeed true, though, it'd mean that Intel will wipe out AMD from their coveted 4 and 8 socket server market even with the new Opteron based on K10.5 architecture. Very scary...
JohanAnandtech - Tuesday, December 16, 2008 - link
I have been a hardware journalist for 10 years now, and I never seen this. A new CPU + platform doubles the performance over a previous one without: 1) Using new instructions 2) a newer process technology 3) large jump in clockspeed or 4) running a very exotic benchmark that stresses only a very small part of the CPU.TeXWiller - Tuesday, December 16, 2008 - link
Sufficient bandwidth of the POWER6 results very good scaling with SMT not only in SAP but Spec tests as well. Nehalem's increased bandwidth could be a reason for the good scaling with SMT in this case.Riek - Tuesday, December 16, 2008 - link
My guess would be that they screwed up the number of cores (dual = quad)... That would bring it down to expected gains and figures...Altough if the performance is indeed correct... The i7 based serverchips will be the fastest cpu's in the servermarket for a very long time... And that might be a very bad thing for AMD and the microprocs industry in general.
defter - Tuesday, December 16, 2008 - link
That's not possible, since quad socket Nehalem will not be available until H2 2009.Riek - Tuesday, December 16, 2008 - link
Since it appeared that HT was enabled i was not that wrong :')liuxue - Monday, April 12, 2010 - link
Customers will expect it to be in parity with what is offered breitling watches by AT&T and other GSM carriers. That said, the Journal story said the upgraded breitling watches GSM iPhone is being made by Taiwanese contract manufacturer Hon Hai Precision Industry, which produced Apple's breitling watch previous breitling watches iPhones. And the CDMA iPhone model is being made by Pegatron Technology.http://www.watchvisa.com/breitling-watches.html