Am 15.10.2014 um 02:08 schrieb Tobias Klausmann:
On 14.10.2014 19:29, Daniele wrote:
Il 14/10/2014 17:04, Jan Engelhardt ha scritto:
On Tuesday 2014-10-14 15:03, Michael Riess wrote:
Am 13.10.2014 um 09:55 schrieb Andreas Schwab:
Claudio Freire
writes: Problem is, AthlonXP is an i686, but obs builds an i686 package that doesn't run on i686.
No, it doesn't. None of the (two) i686 packages are using SSE2 without testing for the feature.
There are only 2 packages with SSE2 enabled? which are they?
If you look through the build logs, you will find some packages that have -msse on the command line even though they probably should not But -msse is ok, the evil is -msse2
Daniele.
If you build a i586 package i'd go with no. The original pentium did not have support for it. There was a nice table somewhere in this thread. Packages with -msse should be i686 only! I think we need even more split for the packages.
i586 / i686 / sse1 / sse2 for 32Bit Builds
no, i think i686 with sse2 and i586 with no sse etc. as baseline would be OK but i586 should be the standard installation cause it just runs on all machines! (yes it will run even on i686 and perhaps you wont even recognize the missing sse|2 ...) perhaps just a i686 without the optimize-flags would be the better way cause an e.g. an AthlonXP identifies as i686 and would install the wrong RPMs The longer i think about it: Are there any arguments or any messurements that proof in any way any REAL benefits of the optimize-flags(-msse2 -msse ...)? (yes i know 13.2 Release is near, it is critical to do changes but RC is for eliminating bugs and i think this is a bug) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org