News - Bulldozer Benchmarks correct; but not definitive

[HEXUS]

Is Bulldozer's unique architecture skewing benchmarks?

Read more.

[Noxvayl]

It is good to finally have evidence in support of this theory. It bodes well for AMD's future because without software derived performance improvements I wasn't sure they would be able to keep up with Intel.

[Steve]

HW guys: "We've got this brilliant new architecture that takes a different approach to superscalar resource usage."

System guys: "But won't that affect performance for software and OSes that aren't optimised to the new micro-architecture?"

HW guys: "We'll leave that one to the compiler writers."

System guys: "Good call."

Poor compiler writers

It's always been the case, of course. Unfortunately, in this instance, it's lead to some performance figures that haven't been very marketing-friendly.

[TheAnimus]

But its one of the big advantages of the RISC system, that simpler design, leads for simpler compilers, simpler debugging.

If AMD want to design a new chip, which has an un-convential design, and as such different from everything before in terms of optomisations, THEY should be leading the way....

[Scribe]

HW guys: "We've got this brilliant new architecture that takes a different approach to superscalar resource usage."

System guys: "But won't that affect performance for software and OSes that aren't optimised to the new micro-architecture?"

HW guys: "We'll leave that one to the compiler writers."

System guys: "Good call."

Poor compiler writers

It's always been the case, of course. Unfortunately, in this instance, it's lead to some performance figures that haven't been very marketing-friendly.

Reminds me of the same troubles Sony had with the CELL CPU and developers doing nothing but moan for the first year.

[shaithis]

It's a chicken and egg scenario though isn't it?

And you could argue AMD lead the way....by releasing the chip

And yes, it reminds me very much of the PS2 and PS3 releases......although this one should be considerably easier to overcome!

[Steve]

But its one of the big advantages of the RISC system, that simpler design, leads for simpler compilers, simpler debugging.

Have you ever read an ARM ISA manual?

You're right, of course, but it's no silver bullet.

[Steve]

It reminds me very much of the PS2 and PS3 releases......although this one should be considerably easier to overcome!

...and the Sega Saturn. A dual processor system that nobody could program for. That did well, didn't it?

Not that I'm drawing that comparison here, but history is there to be learned from

[TheAnimus]

Have you ever read an ARM ISA manual?

You're right, of course, but it's no silver bullet.

Fraid so! I think its why I prefered messing around with PIC12/16/18, everything you need to know you can fit on one side of A4, pinouts, instructions, most of the special registers......

I wasn't trying to say its a silver bullet either, RISC's are always critisised for having a lack of pipeling and therefore wasted silicon.

But I guess all I was trying to say is AMD can't have a "we have built it, you shall optomise" version of field of dreams logic....

[Anusha]

What if Sandybridge runs faster on Windows 8 too? Without checking that, we cannot come to any conclusion.

[shaithis]

What if Sandybridge runs faster on Windows 8 too? Without checking that, we cannot come to any conclusion.

It doesn't.....it was tested at the same time as BD and actually gave a small decrease in the same tests.....although nothing more then most would expect considering Win8 is pre-beta.

[Scribe]

What if Sandybridge runs faster on Windows 8 too? Without checking that, we cannot come to any conclusion.

Sandybridge is formed on a well-established base architecture and already has full support from Windows 7.

[red1776]

"Posted by Anusha - Mon 31 Oct 2011 15:02
What if Sandybridge runs faster on Windows 8 too? Without checking that, we cannot come to any conclusion."

well that comment wasn't obvious at all.

[Lucio]

Definitely brave of AMD to release hardware that they know has issues with current generation OS's, but will hopefully give them the edge for future development. I can see them trying to leverage their way into the tablet/ultrabook market for Windows 8, particularly with their APU ranges built around this architecture

[DjSatansfury]

These things tend to be cyclical in nature. I just hope AMD (who have been in financial trouble in the past years) dont just totally give up the ghost. A market with Intel as the pure provider of Desktop CPU's would be bad :|

[Biscuit]

they wont give up, these CPUs will and have been selling, the market is just quite specific to people using programs which can actually utilize its abilities. I dont think AMDs position has moved at all from the release of this chip.