Minh Nguyen wrote: > Hi David, > > On Thu, Jul 23, 2009 at 7:29 PM, Dr. David > Kirkby<david.kir...@onetel.net> wrote: > > <SNIP> > >>> Now onto ATLAS. >> I won't expect to hear from you for another 8 hours or so then! >> >> I'm trying to add support for the later processors in ATLAS's code. Then >> hopefully I can reduce the time it takes to install, once I have the >> tuning data for the T2+ processor and I have ATLAS understanding what it >> is. Currently it checks for various UltraSPARC CPUs, but the T2+ is not >> one of them. > > The ATLAS SPKG atlas-3.8.3.p6 is successfully compiled on t2! >
I thought ATLAS would build! I believe if you should be able to get up to the point where the sage-4.1.1.spkg gets going. I think the installation will fail there due to two separate bugs. At least now the impact of not fully tuning the ATLAS code will only be seen on sun4v machines, not all of them. Perhaps you can review the atlas patch, which is about as simple as they go. http://sagetrac.org/sage_trac/ticket/6558 Since polybori-0.5rc.p9 built for you, perhaps you can look at: http://sagetrac.org/sage_trac/ticket/6437 Perhaps you could try the revised MPFR patch, http://sagetrac.org/sage_trac/ticket/6453 Note however, the version of gcc you used (4.2.4), did not actually exhibit the problem, which is why I changed the default gcc. But I have confirmed it passes with the latest gcc now. It is NOT a bug in gcc 4.4.0, but Sun have admitted this is a bug in Solaris. The fix allows MPFR to build on any machine, with any version of gcc and pass all tests with no impact on performance except on sun4v machines. I noticed my patch to singular worked for you http://sagetrac.org/sage_trac/ticket/6563 There is probably a better solution than that one. The singular installation appears to be broken by a patch added to Sage to patches/kernel.Makefile.in.diff. I don't know if you can shed any light onto what that patch was supposed to fix, as it seems have changed 'cp' for the 'install-sh' program and broke the installation in the process. I seem to be slacking at the minute, as there are only 4 Solaris patches needing review!! Dave --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
