Le 14/03/2011 07:39, Dr. David Kirkby a écrit :
On 03/13/11 03:34 PM, Julien PUYDT wrote:
running "make ptestlong" gave (in part -- the rest is in other mails) :
sage -t -long -force_lib devel/sage/sage/tests/cmdline.py # 2 doctests
failed
sage -t -long -force_lib devel/sage/sage/calculus/riemann.pyx # Time out
But when I ran the first of them again, I got :
$ ./sage -t -long -force_lib devel/sage/sage/tests/cmdline.py
sage -t -long -force_lib "devel/sage/sage/tests/cmdline.py"
[68.9 s]
----------------------------------------------------------------------
All tests passed!
Total time for all tests: 68.9 seconds
That's nothing unusual. I never 100% trust the parallel testing. I think
people are working on improving it, but if a test fails, and then
passes, I tend to ignore the failure. I'm not sure if its still the
case, but there have been issues where one test overwrites files used by
another test during parallel testing.
It's nice if it gets improved, because tests become worthless if they
can't be trusted!
In any case, is your ARM CPU a multi-core CPU? If not, then I doubt
there's much point running the tests in parallel.
It is a nvidia tegra 2, so yes, it's multi-core. A small multi, but a
multi ;-)
And for the other :
$ ./sage -t -long -force_lib devel/sage/sage/calculus/riemann.pyx
sage -t -long -force_lib "devel/sage/sage/calculus/riemann.pyx"
*** *** Error: TIMED OUT! PROCESS KILLED! *** ***
[1800.4 s]
I don't know the performance of your CPU, but I think ARM is basically a
mobile device, so won't be very fast. That test takes some time on my
machine - it is one of the longest tests, though it is not the longest.
Let's say its main feature is that it doesn't draw the battery too
much... though sage is usable on it.
drkirkby@hawk:~/sage-4.7.alpha1$ ./sage -t -long -force_lib
devel/sage/sage/calculus/riemann.pyx
sage -t -long -force_lib "devel/sage/sage/calculus/riemann.pyx"
[94.2 s]
----------------------------------------------------------------------
All tests passed!
Total time for all tests: 94.2 seconds
which is a quad core 3.33 GHz Xeon. I could well believe your mobile CPU
might be more than 19x slower than a fairly modern high-end Xeon.
That said, the following test took 345.3 s on my machine, which is the
longest of any test
sage -t -long -force_lib
devel/sage/sage/schemes/elliptic_curves/ell_rational_field.py
[345.3 s]
Running "make ptestlong" (with failing tests as discussed) gave :
Total time for all tests: 28274.1 seconds
(which means a little less than eight hours)
What should I do with those failures?
Try setting
SAGE_TIMEOUT_LONG=10000
and running the second test. That will give it 10,000 seconds to run.
sage -t -long -force_lib "devel/sage/sage/calculus/riemann.pyx"
[2627.0 s]
It works.
So that leaves me with just two kinds of issues, discussed in other
mails (precision and prickling).
Thanks,
Snark
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
