Remove optLevel from B3 
https://bugs.webkit.org/show_bug.cgi?id=194079
Summary Remove optLevel from B3
Robin Morisset
Reported 2019-01-30 22:29:10 PST
Now that the BBQ tier for WASM goes directly to Air, there is no more need for different optimization tiers inside B3. Both FTL and Warm OMG use B3 -O2.
Attachments
Add attachment proposed patch, testcase, etc.
Saam Barati
Comment 1 2019-01-30 23:39:30 PST
(In reply to Robin Morisset from comment #0) > Now that the BBQ tier for WASM goes directly to Air, there is no more need > for different optimization tiers inside B3. Both FTL and Warm OMG use B3 -O2. I think it might be worth keeping this running just in case we find a use for it in the future.
Robin Morisset
Comment 2 2019-02-01 12:05:34 PST
(In reply to Saam Barati from comment #1) > (In reply to Robin Morisset from comment #0) > > Now that the BBQ tier for WASM goes directly to Air, there is no more need > > for different optimization tiers inside B3. Both FTL and Warm OMG use B3 -O2. > > I think it might be worth keeping this running just in case we find a use > for it in the future. I am a bit wary of keeping dead code in the codebase, especially as it will no longer be tested going forward (unless we run extra copies of our tests with different optimization levels, but that sounds like it would heavily increase our already massive testing time). If we find a use for it later, would it be that much of a burden to reinstate it?
Saam Barati
Comment 3 2019-02-01 14:53:12 PST
(In reply to Robin Morisset from comment #2) > (In reply to Saam Barati from comment #1) > > (In reply to Robin Morisset from comment #0) > > > Now that the BBQ tier for WASM goes directly to Air, there is no more need > > > for different optimization tiers inside B3. Both FTL and Warm OMG use B3 -O2. > > > > I think it might be worth keeping this running just in case we find a use > > for it in the future. > > I am a bit wary of keeping dead code in the codebase, especially as it will > no longer be tested going forward (unless we run extra copies of our tests > with different optimization levels, but that sounds like it would heavily > increase our already massive testing time). We already do this > > If we find a use for it later, would it be that much of a burden to > reinstate it?
Note You need to log in before you can comment on or make changes to this bug.