Author: José de Jesús García Ruvalcaba
Date: 10:09:45 04/09/99
Go up one level in this thread
On April 09, 1999 at 01:08:59, David Blackman wrote: >On April 08, 1999 at 13:16:53, José de Jesús García Ruvalcaba wrote: > >>On April 08, 1999 at 08:12:09, David Blackman wrote: >> >>>Do these cpu figures include both user and system cpu time for the program, or >>>just user time? If the program is burning lots of system time, maybe you have a >>>problem with un-aligned memory accesses being trapped and emulated in the >>>kernel. >> >> It is only user time. How do I check for un-aligned memory accesses? >>Thanks for your answer. > >If it is just normal C code, there shouldn't be any. But if you do stuff like > >char a[10000]; >int *b; >b=(int *)a; > I am not the author of Faile. I am a simple user. I will check the code, and I will ask the author, Adrien Regimbald, if he does this kind of stuff. >then it might happen. Actually, there are a few other things could cause kernel >trap emulation on Sparc. The architecture has changed a few times, especially >for things like integer multiply and divide. Try to tell the compiler exactly >what version of the Sparc architecture you have. It is definately bad if the >compiler is producing code for a newer version than what you have. > I have heard of the different Sparcs before, but I never thought that could cause any poor performance. I will ask exactly which one I am using, it is a very old machine so what you describe is very likely. >There are other possibilities too. If the program is reasonably small and well >written, i might be able to tell by looking at it. Is it available somewhere? It is small and seems quite well written to me. You can find it at: http://www.ualberta.ca/~adrien/chessprog.htm It is a nice program in an early development stage.
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.