Remove local optimizations #4799
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
OlivierBlanvillain
changed the title
Here are the results, each experiment consists of 40 forks of 10 warmups
and 20 measurements each (that's what it takes to get ~30ms precision!):
ConstantFold 12672.914 ± 29.044 ms/op
InlineLocalObjects 12742.463 ± 29.693 ms/op
DropNoEffects 12717.516 ± 28.266 ms/op
DropGoodCasts 12753.709 ± 31.020 ms/op
Jumpjump 12758.874 ± 28.979 ms/op
Devalify 12820.393 ± 29.601 ms/op
Valify 12732.851 ± 29.466 ms/op
InlineOptions 12705.638 ± 29.186 ms/op
RemoveUnnecessaryNullChecks 12740.366 ± 29.604 ms/op
InlineCaseIntrinsics 12675.068 ± 29.075 ms/op
Not Optimized 12731.068 ± 28.731 ms/op
All Optimizations 12762.053 ± 28.751 ms/op
Also here a count of how many time each optimization modifies a tree
when bootstrapping (running all of them together):
20688 Devalify
7165 DropNoEffects
2787 InlineCaseIntrinsics
1340 ConstantFold
728 DropGoodCasts
76 InlineOptions
49 InlineLocalObjects
16 RemoveUnnecessaryNullChecks
3 Jumpjump
Here are some additional numbers obtained using the Scala Native
benchmarks:
https://plot.ly/~olivierblanvillain/1/#plot
OlivierBlanvillain
force-pushed
the
remove-local-opts
branch
from
22d9046 to
131b8e9
Compare
| @@ -49,7 +49,6 @@ object TestConfiguration { | |||
|
|
|||
| val basicDefaultOptions = checkOptions ++ noCheckOptions ++ yCheckOptions | |||
| val defaultUnoptimised = TestFlags(classPath, runClassPath, basicDefaultOptions) | |||
There was a problem hiding this comment.
Rename defaultOptions?
allanrenucci
approved these changes
Jul 18, 2018
Also remove defaultUnoptimised
OlivierBlanvillain
force-pushed
the
remove-local-opts
branch
from
f5e63a7 to
a408a03
Compare
|
@liufengyun Can you update the benchmarks once this is merged? |
allanrenucci
added a commit
to lampepfl/dotty-ci
that referenced
this pull request
Jul 18, 2018
After scala/scala3#4799, a build spawns 2 jobs instead of 3. So we can increase the number of parallel builds.
This was referenced Jul 18, 2018
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR removes the local optimizations, as discussed in the last meeting. To summarize:
The performance improvements are too small (and inconsistent) compared to the additional compilation cost
With the plan to reuse the scalac backend in dotty, it makes little sense to have similar optimizations earlier in the pipeline
We are currently spending 1/3 of the CI time on testing this code