Author: Robert Hyatt
Date: 06:15:35 01/25/02
Go up one level in this thread
On January 25, 2002 at 00:22:19, Eugene Nalimov wrote: >Vincent was talking about Intel C, not Visual C. > >Intel C is known for its instability. (Of course I cannot be very objective >here). > >Eugene OK. My error. The originals of this were very bad. It is _possible_ that later versions are better. however, SPECINT has nothing to do with it as the SPECINT version of Crafty has no threading support at all. SPEC has asked me about a multi-cpu version, but I have told them "no" because of the non-deterministic behavior which would make repeatable results impossible. > >On January 25, 2002 at 00:14:29, Robert Hyatt wrote: > >>On January 24, 2002 at 19:28:38, Vincent Diepeveen wrote: >> >>>On January 23, 2002 at 22:03:04, Sean Mintz wrote: >>> >>>>I have not yet written any part of it in cilk. I am still compiling as plain c >>>>with the intel c compiler. So cilk can't be blamed for -my- speed problems. >>> >>>for parallel programs i can't recommend intel C++ either. >>> >>>crafty doesn't count here. they of course tested that at the intel c++ >>>design desk till it worked. >>> >>>for me it simply crashes. >> >> >>This is incorrect. When I first "threaded" Crafty and got it working, >>Jason Deines helped me test it under windows. It worked just fine before >>Microsoft ever heard of Crafty... >> >>I've not seen the thread stuff in MSVC screw up as much as the gcc guys >>have managed to break GCC and GLIBC. In fact, posix threads has been broken >>for a year now, so that I had to write my own clone() interface to get crafty >>to run on the newest RedHat releases.
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.