Replies to post #2707 on Advanced Micro Devices Inc (AMD)

[jhalada]

Elmer,

Despite claims that Hammer would beat anything Intel has to offer, Hammer gets it's but kicked by Itanium in TPCC and edges out Xeon (for now)

I don't think you are reading these correctly. You have to pick a processor, and than look at it across all benchmarks, and see which one is an overall winner. If you pick Opteron vs. Itanium, Opteron is the overall winner. If you pick Xeon whatever, Opteron is the overall winner.

The fact that all the Intel processors combined are able to steal a win here and there vs. 1 concrete Opteron processor doesn't mean that Opteron is not the overall winner. With your approach, you could even add Alpha to the mix and point to some benchmark where Alpha is still leading and say that Intel is ahead.

Joe

[economaniac]

Elmer, re Hammer gets it's but kicked by Itanium in TPCC and edges out Xeon (for now)

http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_8796_8800~69696,00.html

following your link, Hammer is 6 % faster than Xeon (106 to 100) while Itanium is ... 6.7% faster than Hammer (113 to 106)

Do you really think that extra .7% makes the difference between "edging out" and "kicking butt"? Is it just possible that you aren't being entirely objective in comparisons between AMD and Intel?

E

[Dan3]

Hammer is twice as fast as Xeon under server loads:

Our most important server test for comparing the Opteron to the Xeon in an application server scenario is our 32-bit "Nile" application server benchmark. The test is both CPU and disk-intensive, and it emulates a book-ordering transaction-processing environment modeled on Amazon.com. The test uses Oracle 9i as the back-end, running on a Xeon cluster server, and uses BEA WebLogic Server 7.0.2 application server software. The BEA application server software runs on the test equipment – in this case we loaded it on both the 2P Opteron and 2P Xeon systems, with Windows 2000 Server as the OS.

Results on the Nile benchmark showed the dual Opteron system outperforming the dual Xeon by a fairly wide margin. Across a 300 to 500 virtual user load, where transaction processing stabilized with both high disk and CPU utilization, the Xeon averaged 8.5 transactions per second, and the Opteron averaged 15.5 transactions per second, nearly double the Xeon. In the response time measurements, at the 200 user load, average transaction time (start to finish) was approximately 34 seconds on the Xeon and 30 seconds on Opteron, but moving to 300 users, Opteron stayed at 30 seconds, and Xeon moved to 50 seconds. At 400 users, Opteron was 35 seconds, and Xeon was near 80 seconds. And at 500 users Opteron was about 50 seconds, and Xeon was near 100 seconds.

http://www.extremetech.com/article2/0,3973,1038369,00.asp