Author: Jeremiah Penery
Date: 21:42:13 02/10/03
Go up one level in this thread
On February 10, 2003 at 23:35:05, Matt Taylor wrote: >On February 10, 2003 at 15:09:18, Tom Kerrigan wrote: > >>On February 10, 2003 at 02:41:50, Matt Taylor wrote: >> >>>In compiler-generated code, my Athlon tends to retire closer to 2 instructions >>>per clock. I would assume that McKinley does better. The restrictions really >> >>Which tool are you using to measure that? > >It is an observation. Every bit of data I've ever seen points to much lower rates of instruction retiring in real-world code. A few programs I've come across that have performance counters give results on the order of < 1/cycle, in fact. Seti@Home is one such application, and unfortunately I don't remember others specifically. I can't make any claims for their accuracy, but I can't imagine why they would be horribly wrong. >Occasionally I get to look at GCC's code. The few times I have compared VC's >output with GCC's, it seems GCC always did a tad better. I did not admit that >until yesterday when I observed that GCC emits cmov in regular code when >optimizing for Pentium Pro/K6-2 or higher. I have never seen VC emit cmov; to >the best of my knowledge, it never does. VC (up to version 6, at least, perhaps v7 .net also) emits only instructions that can run on 386 processors.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.