Summary: | inkscape unstable when using -O3 cflag on opteron | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Andrew Whalan <skozombie> |
Component: | [OLD] GNOME | Assignee: | Gentoo Graphics Project <graphics+disabled> |
Status: | RESOLVED INVALID | ||
Severity: | critical | CC: | jer, ralf, zypher |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Andrew Whalan
2005-03-06 00:40:54 UTC
let bugwranglers do the assignment (In reply to comment #0) > It seems -O3 (or one of the other optiosn noted) is an unstable cflag, perhaps > x86_64 specific? As someone who has worked on GCC itself, I *strongly* recommend against using -O3 and -funroll-loops as default CFLAGS. They work for the majority of applications, but not all. In a significant number of cases, -O2 produces *faster* code than -O3. See my article at: http://www.coyotegulch.com/products/acovea/aco5k8gcc40.html Which I suppose means I should put in another plug for package-specific flags... ;) using -O3 is simply not recommended for normal use by the gcc documentation. proposed to be closed. (In reply to comment #3) > using -O3 is simply not recommended for normal use by the gcc documentation. > proposed to be closed. Good idea. What version of inkscape was this anyway? I can't imagine it is still a problem now. The reason I reported this is I've seen many apps report that they filter cflags. I thought that by reporting this, inkscape could have forcefully -O3 filtered out on compile. I've personally given up on gentoo, but filtering cflags that are known to be unstable for some apps that may work for others would help the overall quality of a gentoo install. |