In the past time for full build was a problem, I remember build
times longer than two hours. Given that many things changed
(machnes changed, Lisps changed) I tried to do objective
comparison. For this I compared recent trunk and FriCAS-1.0.0
on my machine (quad core Core 2 2.4 GHz).
The results are as follows:
FriCAS 1.0.0 -- parallel build using sbcl-1.0.16 fails after
real 44m47.097s
user 50m51.739s
sys 2m56.322s
FriCAS 1.0.0, serial sbcl-1.0.16:
real 63m2.557s
user 54m12.843s
sys 3m6.112s
FriCAS 1.0.0, serial gcl:
real 67m36.906s
user 56m0.507s
sys 5m46.800s
trunk, parallel, sbcl-1.0.16:
real 8m27.687s
user 20m3.082s
sys 2m6.359s
trunk, parallel, sbcl-1.0.54:
real 7m14.147s
user 16m50.953s
sys 1m53.220s
So, now using sbcl-1.0.54 build is 8.71 times faster. Part
of speedup comes from improvement to sbcl, but using sbcl-1.0.16
build is still 7.45 faster. Largest speedup comes from
parallel build, part from improvement to build process
(skiping unnecessary parts), a small part for using easier
to compile constructs in algebra, but large part comes from
improvements to Spad compiler.
In 1.0.0 algebra has 152828 wc lines, in trunk 188238 wc lines,
so Spad compiler has about 23% more work to do.
--
Waldek Hebisch
[email protected]
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/fricas-devel?hl=en.
