Are Palominos faster than Thunderbird core?

[senior guy]

Anand's review:

<< This allows the Athlon MP to run code optimized for 3DNow! or SSE instruction sets although it doesn't necessarily mean that it can run SSE optimized code as fast as a Pentium III... >>


Insane3D's rendition:

<< AMD's implementation of SSE (3DNow! Professional) is just plain old slower than on Intel CPU's. >>


Insane, aren't you taking a lot of liberty in the way you interpreted Anand's SSE comments, or do you know (from any comparative testing) that the Palomino's SSE implementation is really &quot;just plain old slower&quot; than the PIII's (and if so, can you please provide a link to any such test)?

 

[Citadel535]

Actually it is good that it supports SSE but I would wait for revisions then to be on the safe side since you never know what problems would crop up.

 

[TravisBickle]

you of course know that SSE is mainly there to help the floating point unit like MMX was there to help the integer unit.
the Athlon has a brute of an FP unit. if you don't get as much % boost with SSE on one it doesn't mean the wiring is inferior. try benchmarking against a PIII and you'll get more sense in the comparison.

 

[senior guy]

The SIMD instructions of SSE apply to both integer and floating point. But I'm not questioning the Athlon's superior FPU (over the PIII's), I'm just curious about the basis for Insane3D's statement to the effect that the Palomino's SSE implementation is &quot;just plain old slower&quot; than the PIII's. While Anand merely suggested that the Palomino's implementation of SSE may not be as efficient as the PIII's SSE, I just haven't seen benchmark testing (from Anand or anyone else) to objectively show that's true.

I realize that for most users this may not be worth any more discussion time, but for those of us who use SSE-embeded software every day, it's a matter of significant importance.

 

[Quickfingerz]

There could be a chance that AMD's CPU's are &quot;faster&quot; in SSE implementation. This is however very unlikely.

 

[Insane3D]

&quot;Insane, aren't you taking a lot of liberty in the way you interpreted Anand's SSE comments, or do you know (from any comparative testing) that the Palomino's SSE implementation is really &quot;just plain old slower&quot; than the PIII's (and if so, can you please provide a link to any such test)? &quot;

I am sorry if it seemed that way. I just think to expect AMD's first stab at implementing an Intel technology on their CPU's to be comparable in performance to Intel's CPU's is unrealistic. I will admit that I really don't use any SSE enabled applications, so I may be way off base here. It just makes sense to me that Intel's implementation is most likely faster as it is their technology to begin with, and they have been using it in their CPU's far longer than AMD has. I have not seen any comparitive testing and yes, I am just assuming. If it turns out I am completely wrong, I have no problem admitting it..

 

[senior guy]

<< I am sorry if it seemed that way. I just think to expect AMD's first stab at implementing an Intel technology on their CPU's to be comparable in performance to Intel's CPU's is unrealistic. I will admit that I really don't use any SSE enabled applications, so I may be way off base here. It just makes sense to me that Intel's implementation is most likely faster as it is their technology to begin with, and they have been using it in their CPU's far longer than AMD has. I have not seen any comparitive testing and yes, I am just assuming. If it turns out I am completely wrong, I have no problem admitting it. >>


Now that's a far more reasonable way to put it than your &quot;just plain old slower&quot; remark, and more in keeping with what I've come to expect from your many informative posts. Thanks for clearing that up.

Since this is a matter of significant importance for myself, as well as to other potential Palomino users who frequently run apps with SSE-embedded instructions, it would be great if Anand would conduct a benchmark test to objectively determine the facts of the matter (rather than just speculate as to the outcome)!