LINPACK: Intel's Nehalem versus AMD Shanghai
by Johan De Gelas on November 28, 2008 12:00 AM EST- Posted in
- IT Computing general
A "beta BIOS update" broke compatibility with ESX, so we had to postpone our virtualization testing on our quad CPU AMD 8384 System.
So we started an in depth comparison of the 45 nm Opterons, Xeons and Core i7 CPUs. One of our benchmarks, the famous LINPACK (you can read all about it here) painted a pretty interesting performance picture.
We had to test with a matrix size of 18000 (2.5 GB of RAM necessary), as we only had 3 GB of DDR-3 on the Core i7 platform. That should not be a huge problem as we tested with only one CPU. We normally need about 4 GB for each quadcore CPU to reach the best performance.
We also used the 9.1 version of Intel's LINPACK, as we wanted the same binary on both platforms. As we have show before, this version of LINPACK performs best on both AMD and Intel platforms when the matrix size is low. The current 10.1 version does not work on AMD CPUs unfortunately.
We don't pretend that the comparison is completely fair: the Nehalem platform uses unbuffered RAM which has slightly lower latency and higher bandwidth than the Xeon "Nehalem" will get. But we had to satisfy our curiousity: how does the new "Shanghai" core compare to "Nehalem"?
Quite interesting, don't you think? Hyperthreading (SMT) gives the Nehalem core a significant advantage in most multi-threaded applications, but not in Linpack: it slows the CPU down by 10%. May we have found the first multi-threaded application that is slowed down by Hyperthreading on Nehalem? That should not spoil the fun for Intel though, as many other HPC benchmarks show a larger gap. AMD has the advantage of being first to the market, Nehalem based Xeons are still a few months away.
Also, the impact of the memory subsystem is limited, as a 50% increase in memory speed results in a meager 6% performance increase. The Math Kernel Libraries are so well optimized that the effect of memory speed is minimized. This in great contrast to other HPC applications where the tripple channel DDR-3 memory system of Nehalem really pays off. More later...
Facebook
Twitter
RSS
60 Comments
View All Comments
BlueBlazer - Saturday, November 29, 2008 - link
Go spam somewhere else.JohanAnandtech - Friday, November 28, 2008 - link
Before you got out and lynch Anand, know that this *it*.anandtech.com.Anand has never received the Shanghai systems, we have. (That is Jason/Ross and me, Johan). And what is so bad about Anand continously publishing reviews about Nehalem? The more info the better I say.
So different people, different benchmarks and generally we are slower. My Dunnington review was a little bit later than planned too.
Anyway, I got the Quad Socket system. So the original plan was to run ESX etc. on it, like we have done on Dunnington. However, the BIOS is beta, and is not able to get ESX installed and power measurements are also not accurate. So I had to leave all the benchmarking we have done so far and go to plan B.
mkruer - Friday, November 28, 2008 - link
To quote Dean Yeager: Your theories are the worst kind of popular tripe, your methods are sloppy, and your conclusions are highly questionable. You are a poor scientist, Dr. Venkman!But seriously. Anandtech has become the next Toms Hardware. (The irony is that Toms Hardware has gotten better) Don't believe me look at the lat ATI vs Nvidia review. ATi had a card that was 85-90% as fast as Nvidia and cost half as much. The conclusion... Nvidia was better.
BlueBlazer - Saturday, November 29, 2008 - link
If you have nothing useful to inquire, comment or contribute here then STFU! Please keep your rants to yourself and with your other retarded buddies in your own asylum.formulav8 - Sunday, November 30, 2008 - link
Grow-up, He posted nothing improper like you did.JohanAnandtech - Monday, December 1, 2008 - link
No he only accussed us of bias without giving any proof. Why is it so hard for some people to distinguish between a blog post - which is meant to give first impressions - and a full blown review?MonkeyPaw - Monday, December 1, 2008 - link
Don't worry, I've been reading anandtech for a while, and I don't see this big bias theory. You have products to compare, which means you have to compare products to rivals and then pick a winner. That's not always easy or popular. Regardless, true tech enthusiasts want as much (accurate) info as they can get as soon as possible, that way we can make our own conclusions and figure out what product best meets our needs. A good tech site worries more about the details and less about the winner. From what I can see, anandtech uses much more space talking details, and very little space saying "and the winner is..."Griswold - Saturday, November 29, 2008 - link
I hope you'll join them there - its definitely where you belong to.BlueBlazer - Saturday, November 29, 2008 - link
MKruer, a moderator of AMDZone bragging about his "exploits" here...http://www.amdzone.com/phpbb3/viewtopic.php?f=52&a...">http://www.amdzone.com/phpbb3/viewtopic.php?f=52&a...
Well I guess that proves how childish those braindeads are... that's why they need an asylum like AMDZone.
ruiner5000 - Monday, December 1, 2008 - link
Ahh, 11 years of fun. Of course we will be around for longer than Anandtech whether you get it or not.I for one miss the Ace's Hardware Johan, but I still love Johan.
Screw the haters Johan. These guys have no idea what goes into doing a review. Ignorance is bliss.
Old bluebawlz must have a Core i7 TLB bug jacket.