This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

RE: [PATCH] RE: gcc parallel make check

From: "VandeVondele Joost" <joost dot vandevondele at mat dot ethz dot ch>
To: Jakub Jelinek <jakub at redhat dot com>, Yury Gribov <y dot gribov at samsung dot com>
Cc: "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>, "fortran at gcc dot gnu dot org" <fortran at gcc dot gnu dot org>, "gcc-patches at gcc dot gnu dot org" <gcc-patches at gcc dot gnu dot org>
Date: Tue, 9 Sep 2014 15:01:40 +0000
Subject: RE: [PATCH] RE: gcc parallel make check
Authentication-results: sourceware.org; auth=none
References: <20140905143740 dot GL17454 at tucnak dot redhat dot com> <908103EDB4893A42920B21D3568BFD93150F414C at MBX23 dot d dot ethz dot ch> <20140905145304 dot GM17454 at tucnak dot redhat dot com> <908103EDB4893A42920B21D3568BFD93150F7F45 at MBX23 dot d dot ethz dot ch> <540ED02A dot 9080002 at samsung dot com> <20140909101043 dot GQ17454 at tucnak dot redhat dot com> <908103EDB4893A42920B21D3568BFD93150F8014 at MBX23 dot d dot ethz dot ch> <20140909132414 dot GT17454 at tucnak dot redhat dot com> <908103EDB4893A42920B21D3568BFD93150F80BA at MBX23 dot d dot ethz dot ch> <540F0E3E dot 3090101 at samsung dot com>,<20140909143300 dot GV17454 at tucnak dot redhat dot com>

Now with gzipped figure.. why do these bounce ?

> But if there are jobs that just take 1s to complete, then clearly it doesn't
> make sense to split them off as separate job.  I think we don't need 100%
> even split, but at least roughly is highly desirable.

Let me add some data, attached is a graph (logscale y) showing the runtime of tests before and after my changes (including a new patch for c++). There is virtually no change for tests running shorter than 50s, only slowly running tests have been split.

Now, there are only very few slow tests remaining:

gcc_trunk/obj.new> find . -name "*.log" | xargs grep " completed in " | sort -n -k 5 | tail -n 10
./gcc/testsuite/gcc/gcc.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/gcc/testsuite/gcc.dg/torture/dg-torture.exp completed in 521 seconds
./x86_64-unknown-linux-gnu/libstdc++-v3/testsuite/libstdc++.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/libstdc++-v3/testsuite/libstdc++-dg/conformance.exp completed in 530 seconds
./x86_64-unknown-linux-gnu/libstdc++-v3/testsuite/libstdc++.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/libstdc++-v3/testsuite/libstdc++-dg/conformance.exp completed in 553 seconds
./x86_64-unknown-linux-gnu/libgomp/testsuite/libgomp.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/libgomp/testsuite/libgomp.fortran/fortran.exp completed in 561 seconds
./gcc/testsuite/gcc/gcc.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/gcc/testsuite/gcc.c-torture/compile/compile.exp completed in 625 seconds
./x86_64-unknown-linux-gnu/libstdc++-v3/testsuite/libstdc++.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/libstdc++-v3/testsuite/libstdc++-dg/conformance.exp completed in 683 seconds
./gcc/testsuite/g++/g++.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/gcc/testsuite/g++.dg/dg.exp completed in 702 seconds
./x86_64-unknown-linux-gnu/libstdc++-v3/testsuite/libstdc++.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/libstdc++-v3/testsuite/libstdc++-dg/conformance.exp completed in 726 seconds
./gcc/testsuite/gcc/gcc.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/gcc/testsuite/gcc.c-torture/execute/execute.exp completed in 752 seconds
./x86_64-unknown-linux-gnu/libstdc++-v3/testsuite/libstdc++.log:testcase /data/vjoost/gnu/gcc_trunk/gcc/libstdc++-v3/testsuite/libstdc++-dg/conformance.exp completed in 904 seconds

They, of course, limit the ultimate speedup.

Attachment: timings.png.gz
Description: timings.png.gz

References:
- Re: [PATCH] RE: gcc parallel make check
  - From: Jakub Jelinek
- Re: [PATCH] RE: gcc parallel make check
  - From: Jakub Jelinek
- RE: [PATCH] RE: gcc parallel make check
  - From: VandeVondele Joost
- Re: [PATCH] RE: gcc parallel make check
  - From: Yury Gribov
- Re: [PATCH] RE: gcc parallel make check
  - From: Jakub Jelinek
- RE: [PATCH] RE: gcc parallel make check
  - From: VandeVondele Joost
- Re: [PATCH] RE: gcc parallel make check
  - From: Jakub Jelinek
- RE: [PATCH] RE: gcc parallel make check
  - From: VandeVondele Joost
- Re: [PATCH] RE: gcc parallel make check
  - From: Yury Gribov
- Re: [PATCH] RE: gcc parallel make check
  - From: Jakub Jelinek

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]