Atlas compile time mostly depends on whether atlas has pre-computed defaults for your cpu, not so much on the CPU speed. If it bothers you, use the SAGE_ATLAS_LIB environment variable to use a system/previously-compiled library
On Tuesday, February 4, 2014 8:05:36 PM UTC, Emmanuel Charpentier wrote: > > I tried installing "the classic way", i. e. from the tarball (as mirrored > at Paris-VI university). > > On a "big laptom" (Core i7, 16 GB RAM) under Debian testing, a parallel > compilation (export MAKE="make -j8 -l10" ; time make ssl) t took almost 4 > hours : > real 215m37.165s > user 369m27.220s > sys 10m40.716s > > of this time, installing pyopenssl took 2.6 seconds. > > Compiling sage 6.0 (installed by cloning the git directory) took about > 40-45 minutes a while back. !!! > > For what it's worth, ATLAS compilation seemed (at visual examination of > the compilation console) to be the bottleneck : this console "froze" > minutes at a time showing results of ATLAS optimizing tests. At the end of > the atlas log, I find this : > real 191m4.515s > user 177m51.532s > sys 5m27.992s > > Intrigued by this result, I launched a *serial* compilation on another > desktop machine (Xeon, 8 threads, 8 GB RAM). That went much faster : > real 173m23.190s > user 162m59.508s > sys 5m56.480s > > As for atlas : > real 11m4.562s > user 12m37.052s > sys 0m52.264s > > Tentative conclusion : as distributed in the tarball, parallel compilation > of atlas seems to have a (serious) problem. > > Should this be filed as a ticket ? > > Emmanuel Charpentier > > > Le jeudi 30 janvier 2014 22:08:14 UTC+1, Volker Braun a écrit : >> >> I'm pleased to announce the release of Sage-6.1! >> >> This means that now the "master" branch changed for the first time, and >> now points to the new stable release. The "develop" branch for now also >> points to the very same stable release, but will move on once the first >> beta version is ready. As always, you can also check out a version tag >> (namely "6.1") if you want a particular version from the git repo. >> >> Source tarball is here (to be uploaded to the mirrors): >> >> http://boxen.math.washington.edu/home/release/sage-6.1/sage-6.1.tar.gz >> >> md5sum = 75130374fcdb9395955db81f1ed28422 >> >> Binary builds will be posted when they are finished. Also, I hope that >> Jeroen will post an overview over the changes from 6.0 to 6.1. >> >> Changelog: >> >> $ git log --author 'Release Manager' --oneline 6.1 ^6.1.rc0 >> 022e2ce Trac #15719: Fix documentation bugs >> >> -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/sage-release. For more options, visit https://groups.google.com/groups/opt_out.
