RESOLVED FIXED 151411
js/regress/script-tests/ftl-object-sub.js needs more iterations for benchmarking.
https://bugs.webkit.org/show_bug.cgi?id=151411
Summary js/regress/script-tests/ftl-object-sub.js needs more iterations for benchmark...
Mark Lam
Reported 2015-11-18 15:08:20 PST
Previously, in r192392, I reduced the test iteration count to make the layout test happy and not time out. Unfortunately, this defeats the use of this test for benchmarking. I've changed it to now use different iteration counts depending on who is running it.
Attachments
proposed patch. (1.80 KB, patch)
2015-11-18 15:11 PST, Mark Lam
fpizlo: review-
patch 2: reworded comments, and change run config to runFTLNoCJIT only. (2.13 KB, patch)
2015-11-18 16:56 PST, Mark Lam
fpizlo: review+
Mark Lam
Comment 1 2015-11-18 15:11:00 PST
Created attachment 265785 [details] proposed patch.
Filip Pizlo
Comment 2 2015-11-18 15:12:29 PST
Comment on attachment 265785 [details] proposed patch. View in context: https://bugs.webkit.org/attachment.cgi?id=265785&action=review > LayoutTests/js/regress/script-tests/ftl-object-sub.js:33 > + // We're doing a benchmark run and need more iterations. > + iterations = 100000; > + expectedResult = 2000020; run-jsc-stress-tests will take this path, so you're not getting what you intended.
Mark Lam
Comment 3 2015-11-18 16:56:41 PST
Created attachment 265810 [details] patch 2: reworded comments, and change run config to runFTLNoCJIT only.
Mark Lam
Comment 4 2015-11-19 15:10:49 PST
Comment on attachment 265810 [details] patch 2: reworded comments, and change run config to runFTLNoCJIT only. Cancelling commit. Will land manually.
Mark Lam
Comment 5 2015-11-19 15:16:36 PST
Thanks for the review. Landed in r192663: <http://trac.webkit.org/r192663>.
Note You need to log in before you can comment on or make changes to this bug.