Author: Robert Hyatt
Date: 10:11:12 12/19/02
Go up one level in this thread
On December 19, 2002 at 12:43:13, Gian-Carlo Pascutto wrote: >It's not been able to compile Deep Sjeng with IPO since a long time, >bombing out with internal compiler errors. > >With Eugene's new tablebase code, it's not capable of compiling it with >profile feedback, also generating an internal compiler error. > >The latest version produces incorrect code for Deep Sjeng itself, >causing Deep Sjeng to crash. > >Enough for me that it's in the trashbin and *staying* there. > >-- Not my experience. It compiles all of Crafty just fine, and the profile-based optimizations seem to work well also. As I said, it has also compiled many serious applications run on my cluster with no problems at all. I should add I have only used version 6.0 and the new version 7.0. I did not try any earlier versions... The codes here are both integer-based and FP-based, and even combinations, so those don't seem to be an issue any longer... If you do unsafe aliasing and such, it might have a problem, but nobody here writes that kind of code... >GCP
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.