Linux Shootout: Opteron 150 vs. Xeon 3.6 Nocona
by Kristopher Kubicki on August 12, 2004 2:35 PM EST- Posted in
- Linux
TSCP
We apologize for the broken TSCP Makefile in the previous review which rendered our initial results inaccurate. Fortunately we posted the file so that others were able to detect the error and not find fault with the processors instead. The large issue many of our readers have brought to our attention are the severe difference in performance between various optimizations. Below you can see how various compile flags affected our benchmark scores.
The first benchmark is run with the optimization flags:
-O2 -funroll-loops -frerun-cse-after-loop
The next benchmark is run with the optimization flags:
-O3 funroll-loops -frerun-cse-after-loop
Finally, we have the architecture optimized flags as well:
(Intel) -O3 - march=nocona -funroll-loops -frerun-cse-after-loop
(AMD) -O3 - march=k8 -funroll-loops -frerun-cse-after-loop
You are reading these charts correctly, the O3 flag actually penalizes the AMD CPU. We also compiled the program with -O2 -march=k8 but we got virtually the same score with or without the march flag.
We were informed others have been capable of much faster nodes per second using GCC 3.4.1 and the flagset:
-O3 -march=athlon-xp -funroll-loops -fomit-frame-pointer -ffast-math -fbranch-probabilities
We did not have time to fully test GCC 3.4.1, although there is a strong likelihood that 3.4 encourages better optimizations (particularly on the x86_64 platforms).
Crafty
For good measure, we have included Crafty into our chess benchmarks section. Crafty was only built using the "make linux-amd64" target. From the Makefile, it seems as though the "AMD64" moniker is slightly inappropriate. The target claims:
# -INLINE_AMD Compiles with the Intel assembly code for FirstOne(), # LastOne() and PopCnt() for the AMD opteron, only tested # with the 64-bit opteron GCC compiler.
The benchmark was generated by running the "bench" command inside the program.
It is clear the difference between both processors is quite severe in this instance. Although it is difficult to pin an exact culprit, there are likely multiple arch optimizations were left untapped, and thus our reasoning for discouraging overusage of optimizations in general.
Facebook
Twitter
RSS
92 Comments
View All Comments
epicstruggle - Thursday, August 12, 2004 - link
"The 3.6F will still be marketed against the 3500+/3800+ dual channel AMD CPUs. The benchmarks were messed up in that article, not our processor choice. This was addressed at the beginning of both articles"First great redo. :) I dont care who wins or looses, but at least now the comparision is fair.
About the above quote, who is doing this marketing/comparison. Im assuming Intel? Doesnt it look suspect that they want to compare a 800+ dollar processor with one in the low 300s? Why fall for marketing pr/fud?
again, thanks for the info.
epic
dke - Thursday, August 12, 2004 - link
I enjoyed this article much more than the first one. This one is much more accurate in my opinion.I would like to point out that you can purchase an Opteron 150 (boxed) for $594.49 with free shipping at
http://www.chumbo.com/info.asp?s=030143803701&...
and not $850 like you quoted in your "Final Thoughts" page. Additionally, AMD's pricing page suggests that the Opteron 150 be priced at $637. Any store pricing the Opteron 150 at $800 or $850 will not make any sales, so, I don't think you can justify your statement, "Thus, it is priced around $850 at time of publication." That is the only thing I think you should change with this article. That sentence should be changed to, "Thus, it is priced (by AMD) at around $650 and can be purchased at around $600 at time of publication."
Other than that, I think you've written a wonderful article. I'd also like to thank you for doing this during your "vacation" time. I suppose that wasn't much of a vacation. I think what you did shows quite a bit of dedication to your work, and I (for one) appreciate it.
love4ever - Thursday, August 12, 2004 - link
thanks Kristopher.very nice review.
Carfax - Thursday, August 12, 2004 - link
Kris, that's fine man.. Good work on this review, and I look forward to seeing the 32 and 64 bit Nocona benches in a future review!KristopherKubicki - Thursday, August 12, 2004 - link
datacipher: The 3.6F will still be marketed against the 3500+/3800+ dual channel AMD CPUs. The benchmarks were messed up in that article, not our processor choice. This was addressed at the beginning of both articles/Kristopher
datacipher - Thursday, August 12, 2004 - link
Sorry for the empty posts, I just signed up to say this....I think I can speak from a different perspective as I am not a techhead. I do keep a casual eye on hardware and have done extensive programming so I would not classify myself as computer illiterate.
Still I WOULD not have noticeed the rediculous choice of cpu's used in the 1st article. I would have just skimmed the article and assumed that a reasonable choice of benchmark material was used. I would not have known about any of the flaws in the benchmarks. I WOULD have accepted and given weight to the conclusion...which was clearly unwarranted given this second review.
I would like to thank all the posters who criticized that review because without you watchdogs, I would have blindly accepted the article as I used to do with Anandtech which I always though was a reliable source.
Kristopher, it's good you posted a new article, but honestly, your first article was extremely misleading...almost fraudulent...it really changes the nature of Anandtech in my mind...I thought I could rely on you fellows...
If you had initially even written of your reasoning as to why you were using such an uncomparable processor and then properly framed your conclusions it would have been fine....but in the article itself you seemed to be saying that it was a reasonable comparision and the conclusions were not given proper reference.
I'm really disappointed.
KristopherKubicki - Thursday, August 12, 2004 - link
Carfax:I will address that in a future review. I just didnt have the time to do it all over again :'(
Kristopher
Carfax - Thursday, August 12, 2004 - link
This review is MUCH better.. But still, WHY aren't there reference 32 bit scores for the Noconca, so we can see how much of a difference between 32 and 64 bit performance there is?TotalImmortal - Thursday, August 12, 2004 - link
thanks to all you guys at anandtech, never let it be said that you don't listen to your readership!sprockkets - Thursday, August 12, 2004 - link
At the end of page 5, the last paragraph is worded weird. Page 6 graph at the top, "Opteron 250"?I realize that you probably were sleeping when you did this :).
The only thing else I don't like and I know it's out of your control is those stupid "get the facts" propaganda from Micro losers.