intel paper on 64 bit technologie

P4Man

Distinguished
Feb 6, 2004
2,305
0
19,780
this is getting confirmed everywhere; it would be identical to AMD64 (with of course the exception it supports SSE3 and HT). I even read in a thread over at Aces, that there are some clear signs intel implemetented 3DNow! If true, that has to be the joke of the decade :)

Anyway, this is good news for everyone really, I'd hate to see the market get fragmented.

One cave-at; I'm pretty sure intel does not want to help AMD unleashing the potential of AMD64 on their chips, so I wouldn't be surprised to see them play compiler tricks, like only generating optimal 64 bit code when SSE3 is available, and resorting to less optimized, slower code if no SSE3 is detected. Expect impressive SPEC scores for intel's implementation, but be sure to compare AMD scores using the same compiler (or flags) before you judge.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 

c0d1f1ed

Distinguished
Apr 14, 2003
266
0
18,780
I even read in a thread over at Aces, that there are some clear signs intel implemetented 3DNow! If true, that has to be the joke of the decade :)
It's already revolutionary as it is. Intel comptelely controlled x86 until now. But there's no reason at all why they would implement 3DNow!. SSE is simply superiour, especially with SSE2 and SSE3 on top.
Anyway, this is good news for everyone really, I'd hate to see the market get fragmented.
Definitely good news for Athlon 64 owners! And I feel no hesitation any more to buy one myself. I still have this small feeling that they make x86 even messier. But now there's no way around it to accept it as the best compromise.
I'm pretty sure intel does not want to help AMD unleashing the potential of AMD64 on their chips, so I wouldn't be surprised to see them play compiler tricks, like only generating optimal 64 bit code when SSE3 is available, and resorting to less optimized, slower code if no SSE3 is detected.
That's quite obvious. AMD already benefits a lot from things Intel does. They really should write their own compiler if they want optimal benchmarks.
 

P4Man

Distinguished
Feb 6, 2004
2,305
0
19,780
> But there's no reason at all why they would implement
>3DNow!

I don't know, maybe it was required to achieve full binary compatibility with existing (and upcoming) AMD64 code ? After all, MSVC or GCC compiled AMD64 code could well use some 3Dnow instructions. i'm not 100% certain SSE is always faster on A64, or can do everything 3Dnow can (even though it can things 3Dnow can't). And its not like it would have been all that hard for intel to add it either.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 

c0d1f1ed

Distinguished
Apr 14, 2003
266
0
18,780
I don't know, maybe it was required to achieve full binary compatibility with existing (and upcoming) AMD64 code ? After all, MSVC or GCC compiled AMD64 code could well use some 3Dnow instructions. i'm not 100% certain SSE is always faster on A64, or can do everything 3Dnow can (even though it can things 3Dnow can't). And its not like it would have been all that hard for intel to add it either.
I can't completely exclude the possibility, but it's highly doubtable.

MSVC and GCC know that Althon 64 supports SSE so there's no reason to choose 3DNow! instructions. Note that 3DNow! uses the MMX registers, which in turn use the FPU registers, so it only makes things more complicated for the compiler to optimize. SSE has separate registers and it wouldn't be intelligent to leave them empty... Just checked it; even the AMD manual for Athlon 64 advises to use SSE in every situation. There's not even a chaper about 3DNow! in the optimization guide.
 

Mephistopheles

Distinguished
Feb 10, 2003
2,444
0
19,780
Well, the issue with Intel is that they usually provide great software/compiler support for their processors. They provide excellent compilers for their processors, unlike AMD. And in doing so and controlling 85% of the market, it's almost as if a snap of their fingers could make everyone go for 64-bit extensions optimizations quickly. So most definitely, A64 will benefit from Intel's adoption of its technology immensely, because software for it will get written much, much faster because of Intel.

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
 

Latest posts