Author: Robert Hyatt
Date: 07:03:26 11/15/02
Go up one level in this thread
On November 15, 2002 at 07:40:25, David Rasmussen wrote: >On November 14, 2002 at 13:06:17, Robert Hyatt wrote: > >> >>You keep saying that. I keep asking for concrete examples of where the compiler >>produces incorrect code. To date you have not posted one single example. >>Several >>here at UAB have tested this thing carefully and we could find no bugs in some >>major codes we run, both integer and floating point codes in fact. >> >>Intel is faster than gcc 3.2 for me, by a significant margin. And they produce >>_exactly_ >>the same node counts for searches of a billion nodes or more. >> > >Have you tried the -fprofile-arcs and the -fbranch-probabilities options with >gcc 3.2 ? They gave my bitboard-based engine 30%, which makes it as fast as with >MSVC and Intel. > >/David Yes... that was the first thing I checked out...
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.