Bulldozer shows its muscle in server benchmarks
http://finance.yahoo.com/news/New-AMD-Opteron-TM-Processors-iw-4190901655.html?x=0&l=1
2x AMD Opteron™ processors Model 6282 SE (2.6GHz): SPECint®_rate score= 526.
2 x Intel Xeon processors Model X5690 (3.46GHz, 130watt) : SPECint®_rate score = 421.
Not surprisingly, the Bulldozer Opteron destroys Intel Xeon in other enterprise benchmarks, in some cases, 89% faster than the popular Xeons.
posted by Sharikou, Ph. D. @ 11/13/2011 10:14:00 PM
9 Comments:
Nice Try, and as a former AMD enthusiast who has been crapped on by stupid press releases, here is what I have to say:
First: you are quoting an AMD press release, A PRESS RELEASE.
Second: on benches where Intel DOES perform with HT on, they disabled it (look at bench data footer).
Third: cumulative measured bandwidth is better with the greatest amount of RAM. The AMD rig has 64GB while Intel has 24GB.
Fourth: Stop counting cores and just be objective by using SIMULTANEOUS NUMBER OF THREADS PROCESSED, if you determine that, you'd be surprised with the power scores!
Fifth: Is 12T vs 16T even considered fair?
Understandably, FP scores are higher since essentially, AMD's compiler takes advantage of BD's Flexible FPU which is admirable. But my problem is that where are the REAL WORLD software that use those features? More importantly, will there be ANY real world software for it???
Nice PRESS RELEASE, AMD, but I'd rather read unbiased REAL WORLD server benches as they come out.
Huh? Why didn't AMD compare BD against the Xeon E7 family at 16T vs. 16T??? And why disable Hyperthreading?
Yup. leave to a press release that compares Intel's previous generation. even my gaming Phenom IIs outperform the BD Opterons!
So frustrated with this! Major Fail. AMDs confusing Management on how they fucked up their engineers' to blame with the arch!
And now their Marketing is doing another BIG EPIC FAIL (read their fine print below!). OMG, fire those spinners and get the good engineers back!
Too bad most of the server vendors in the segment where the most profit lies are pretty much universally ignoring BD..
Typical Shiradouche and AMD marketing. Since they WILL fail, processor against processor, they have to tilt the scales by dropping the amount of RAM by LESS THAN HALF what the AMD box had, and disable HT.
Poor Shiradouche, not happy with just lying about his PhD, he has to copy Amd marketing materials which, quite honestly are just a bag of shit.
If you read the AMD Press release FINE PRINT, you WILL realize that even AMD is NOT confident with their product. It reads:
"Investors are cautioned that the forward-looking statements in this release are based on current beliefs, assumptions and expectations, speak only as of the date of this release and involve risks and uncertainties that could cause actual results to differ materially from current expectations."
In addition, the AMD spinnersstate (same paragraph):
"Forward-looking statements are commonly identified by words such as "would," "may," "expects," "believes," "plans," "intends," "projects," and other terms with similar meaning."
AMD classifies "forward looking" as being subjective, which defeats the objectivity of their BENTMARKS!
Linpack:
AMD using AMD Math Library (is there any realworld software that uses this?) using the BD opcode enhancements. And compared to Intel's Math Kernel Library version that does not use the newest AVX extensions as opposed to AMD's Core Math. And seriously, what does the artificial Linpack tell you? Give me a suite of multithreaded realworld FP renderers as benchmarks! BTW, HT on SB performs better on the newer Linpack, BD's real counterpart.
verdict: AMD bentmark win
- - - - - - - - - - - - - - - -
Memory bandwidth:
AMD system uses DDR3-1600 (64GB) vs. Intel's DDR3-1333 (24GB)
verdict: AMD bentmark win (DUH)
- - - - - - - - - - - - - - - -
Power per Core:
This is an obvious manipulation of terms when AMD's "2-core Module/CMT" can't even go way past Intel's L5630's HT core in bench suites! Thus the real measure should be actual simultaneous threads where intel is in fact just a few percentages higher, but AMD decides to leverage their faux Cores to gain better power scores even if the actual performance bench SUITES give Intel the bigger performance win.
Verdict: AMD bentmark win
- - - - - - - - - - - - - - - -
And really AMD, you compare your newest BD with the 4-yearl old Nehalems instead of Sandy Bridge?
Sharikou, stop copying and pasting...
He's just trolling at this point. I just don't like people that *LIE* about having credentials, like a ph.d. Or being a "published" journalist. If any respectable publication used his crap it would be to point out how completely retarded people can be.
Hi, thanks for writing such a nice blog on Earth movers Machine. We all like your blog information,
New Bulldozer
Post a Comment
<< Home