Andrey G. Grozin wrote:
> On Fri, 8 Jan 2010, Waldek Hebisch wrote:
> > The 'seem to have problems on some machines' means that build may
> > crash like above.  Port was done by Anatoly Raportirenko and
> > for him it worked on a few different machines (both 32-bit and
> > 64-bit).  For me it failed on 64-bit machine, but worked in 32-bit
> > chroot.  He also wrote that unicode cmucl have bug, so one have to
> > patch and rebuild it (I tried non-unicode version).
> Thanks, I'll contact Anatoly (I know him). (I was told that the unicode 
> bug in cmucl-20a was fixed soon after the release, and later snapshots 
> should be OK.)
> 
> Is there a testsuit for fricas?
> make check
> says that there is no target check. I'd like to run the testsuit with all 
> lisps. First, to see if any of them has some problems. Second, this would 
> be an interesting benchmark.
> 

make all-input

runs test.  Note that currently a few tests prints summary at the
end, for other one have to use textual comparison between runs:

fricas-1.0.8/src/scripts/norm-out build-tree1 build-tree2

The norm-out script is an utility which gets rid of most inessential
differences.  Remaining legal differences are due to floating point
(rounding) noise and Lisp diagnostics.

Concerning using tests as a benchmark: I have noticed that after
building from release tarball 'make all-input' is spending
surprisinly large amount of time in activities other than
running tests, so the result has little in common with actual
use of FriCAS (when you build from SVN checkout time outside
tests is reasonably small).  Also, most tests are short
running so for them we measure startup time of Lisp executables.
For benchmarking one should probably take on few long running
tests: mapleok, marcbench, r21bugsbig and easter.

BTW: More than year ago I posted my results of
speed comparison:

http://groups.google.com/group/fricas-devel/browse_thread/thread/fbd1c7c9796a7173?hl=en


-- 
                              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.


Reply via email to