Collin Funk <collin.fu...@gmail.com> writes: >> Then, you might see whether setting ENABLE_DEBUGGING to 1 in the test's >> code reveals a pattern... > > Thanks for the advice. I'll do some experimenting and let you know what > I find.
>> $ time ./test-pthread-once1 $ time ./gltests/test-pthread-once1 real 0m0.002s user 0m0.001s sys 0m0.001s >> $ time ./test-pthread-once2 $ time ./gltests/test-pthread-once2 Starting test_once ... OK real 8m6.167s user 56m55.342s sys 7m44.460s >> $ time ./test-call_once1 $ time ./gltests/test-call_once1 real 0m0.002s user 0m0.001s sys 0m0.001s >> $ time ./test-call_once2 $ time ./gltests/test-call_once2 Starting test_once ...Alarm clock real 10m4.626s user 71m5.161s sys 9m31.302s When using '#define ENABLE_DEBUGGING 1': $ time ./gltests/test-pthread-once2 [...] Main thread after synchronizing for round 50000 OK real 4m20.487s user 28m53.093s sys 4m26.508s $ time ./gltests/test-call_once2 [...] Main thread after synchronizing for round 50000 OK real 4m33.948s user 30m29.030s sys 4m37.637s So I guess the printf'ing causes there to be less contention and therefore the test passes before the alarm? Collin
