opteron at toms

[dannybin1742]

have fun

 

[silent tone]

Performance is fair. Can performance gains be expected from updated compilers that can optimize for the extra registers, bigger TLB, 64 bit buses?

 

[Nemesis77]

Confirmed: Tom sucks! The Xeons have 512 megs of RAM while Opteron has 2 gigs of RAM!

BTW: What's wrong with Opterons memory-bandwidth? I mean, dual-channel P4 beats it, even though Opteron has dual-channel mem-controller as well! And the mem-bandwidth should go up as CPU's are added, thanks to NUMA. Yet, 2x Opterons (that have 4x 64bit mem-channels (2x for each CPU, each CPU can use other CPU's RAM as well)) are beaten by P4 that has 2x64 mem-channels. What's wrong?

 

[mamisano]

I love how they have Centrino ads inside an Opteron review :disgust:

 

[m0ti]

Check out the article at aces

They actually explain what kind of tests they're doing (hmm... alter table? yeah, how often do you do that? select... what kind? come on Tom!)

 

[mamisano]

Originally posted by: m0ti
Check out the article at aces

They actually explain what kind of tests they're doing (hmm... alter table? yeah, how often do you do that? select... what kind? come on Tom!)

Yeah, that is a pretty damned informative review. I have to give those guys props :beer:

 

[glugglug]

Originally posted by: Nemesis77

BTW: What's wrong with Opterons memory-bandwidth? I mean, dual-channel P4 beats it, even though Opteron has dual-channel mem-controller as well! And the mem-bandwidth should go up as CPU's are added, thanks to NUMA. Yet, 2x Opterons (that have 4x 64bit mem-channels (2x for each CPU, each CPU can use other CPU's RAM as well)) are beaten by P4 that has 2x64 mem-channels. What's wrong?

The P4s that are beating it have Dual DDR400 and 800MHz FSB.

The Hammers being tested have Dual DDR333, 333MHz FSB but effectively like 666 since the FSB is twice as wide.

If you look more closely there are P4s with DDR333 and DDR266 tested which scored lower.

 

[DivideBYZero]

Thats Toms review sucked. I prefer the HardOCP and Annand reviews. They makes comparisons based on the processors, not pie in the sky statements.

Sometimes I think Tom should go and work for one of these companies as he seems to know so much regarding the direction they should be taking and the message they should deliver. Oh, on the other hand, maybe he's just a hack.

 

[DAPUNISHER]

Tom bashing is fun Ja? :beer::evil:

 

[flashbacck]

tom's does suck. and it has nothing to do with him being german (if that is what you were implying), he would still suck if he were american or anything else

 

[DivideBYZero]

Originally posted by: DAPUNISHER
Tom bashing is fun Ja? :beer::evil:

No, he just needs clear and objective conclusions, plus his tests are looking a little shabby compared to Annand etc.