3X testsuite slow-down?

3X testsuite slow-down?

[Michael Snyder]

I just did a cvs update, and the time that it takes to run "make check"
jumped from 8 minutes to 24 minutes.

Unfortunately I can't pin down the date of my last cvs update before
this one any closer than June 23, but I'm pretty sure it was more
recent than that.

Can't say if this is associated with the recent changes in the
testsuite makefiles, but it's the first thing that comes to mind.

Did anybody validate those changes without using -j?

Michael

Re: 3X testsuite slow-down?

[Paul Pluzhnikov]

On Fri, Jul 3, 2009 at 5:08 PM, Michael Snyder<msnyder@vmware.com> wrote:
> I just did a cvs update, and the time that it takes to run "make check"
> jumped from 8 minutes to 24 minutes.

8 minutes sounds too fast.

FWIW, runtest on my machine, with Ada unsupported in my GCC, took 18
minutes on 20090616:

                === gdb Summary ===
# of expected passes            13248
# of unexpected failures        109
# of expected failures          42
# of known failures             114
# of unresolved testcases       3
# of untested testcases         10
# of unsupported tests          62

and took 17 minutes just now:

		=== gdb Summary ===
# of expected passes		13359
# of unexpected failures	103
# of expected failures		43
# of known failures		116
# of unresolved testcases	1
# of untested testcases		11
# of unsupported tests		63


Also, most of the slowdowns I observed come from tests failing with a
timeout, not from breakage of the test suite itself.

-- 
Paul Pluzhnikov

Re: 3X testsuite slow-down?

[Michael Snyder]

Paul Pluzhnikov wrote:
> On Fri, Jul 3, 2009 at 5:08 PM, Michael Snyder<msnyder@vmware.com> wrote:
>> I just did a cvs update, and the time that it takes to run "make check"
>> jumped from 8 minutes to 24 minutes.
> 
> 8 minutes sounds too fast.

It's a fast machine.  ;-)
8 minutes has been about standard for me for some time.

> FWIW, runtest on my machine, with Ada unsupported in my GCC, took 18
> minutes on 20090616:

Hmmm.... no one else is observing this problem?

Re: 3X testsuite slow-down?

[Tom Tromey]

>>>>> "Michael" == Michael Snyder <msnyder@vmware.com> writes:

Michael> Can't say if this is associated with the recent changes in the
Michael> testsuite makefiles, but it's the first thing that comes to mind.

Michael> Did anybody validate those changes without using -j?

Yeah.  And you can try it, too:

    cd $build/gdb/testsuite
    make check-single

That runs the old code.

When I do this, the old and new code (without -j) differ by a few seconds.

Tom

Re: 3X testsuite slow-down?

[Michael Snyder]

Tom Tromey wrote:
>>>>>> "Michael" == Michael Snyder <msnyder@vmware.com> writes:
> 
> Michael> Can't say if this is associated with the recent changes in the
> Michael> testsuite makefiles, but it's the first thing that comes to mind.
> 
> Michael> Did anybody validate those changes without using -j?
> 
> Yeah.  And you can try it, too:
> 
>     cd $build/gdb/testsuite
>     make check-single
> 
> That runs the old code.
> 
> When I do this, the old and new code (without -j) differ by a few seconds.

Hmm, same here.  Must have been something else.
Sorry for the false alarm...