AMD Bulldozer - Compiler Impact on Performance

Discussion of chess software programming and technical issues.

Moderator: Ras

FlavusSnow
Posts: 89
Joined: Thu Apr 01, 2010 5:28 am
Location: Omaha, NE

AMD Bulldozer - Compiler Impact on Performance

Post by FlavusSnow »

The question is, will bulldozer's chess performance increase as compilers are improved to target the new architecture? Or is it simply a hardware issue (too little L1 cache with too high of latency) and there is no hope?
bob
Posts: 20943
Joined: Mon Feb 27, 2006 7:30 pm
Location: Birmingham, AL

Re: AMD Bulldozer - Compiler Impact on Performance

Post by bob »

FlavusSnow wrote:The question is, will bulldozer's chess performance increase as compilers are improved to target the new architecture? Or is it simply a hardware issue (too little L1 cache with too high of latency) and there is no hope?
Every processor has its quirks. A compiler can help here and there. It can hurt far more if it wants to of course... I don't think you will see any major improvements due to optimizing software tricks, however.
rbarreira
Posts: 900
Joined: Tue Apr 27, 2010 3:48 pm

Re: AMD Bulldozer - Compiler Impact on Performance

Post by rbarreira »

FlavusSnow wrote:The question is, will bulldozer's chess performance increase as compilers are improved to target the new architecture? Or is it simply a hardware issue (too little L1 cache with too high of latency) and there is no hope?
Who knows... Once I made an optimization to a program which resulted in a 80% performance boost on the Pentium 4, and just a 10-30% improvement on other CPUs (even though this optimization did not involve using any of the new instructions like SSE etc.).

This is of course highly dependent on the particular software. There are probably programs which can hardly run better on Bulldozer and others which may benefit greatly from specific optimization tricks.