Provide aliasing hint to the compiler#2421
Conversation
Summary: This change results in massive speedups on ProtocolBench. Prod canaries also gave some promising signals. # x86 ProtocolBench Benchmarks not included below were neutral or only small wins. See pastes for full results. Before: full results (P1791766582) ``` CompactProtocol_write_MixedInt 114.09ns 8.77M CompactProtocol_write_LargeMixed 526.47ns 1.90M CompactProtocol_write_BigListBigInt 34.63us 28.87K CompactProtocol_write_BigListMixedInt 824.45us 1.21K ``` After: full results (P1791763018) Benchmarks not included below were neutral or only small wins. See pastes for full results. Before: ``` CompactProtocol_write_MixedInt 47.03ns 21.26M CompactProtocol_write_LargeMixed 491.08ns 2.04M CompactProtocol_write_BigListBigInt 20.74us 48.22K CompactProtocol_write_BigListMixedInt 330.13us 3.03K ``` # aarch64 ProtocolBench Benchmarks not included below were neutral or only small wins. See pastes for full results. Before: full results (P1791766333) ``` CompactProtocol_write_MixedInt 56.34ns 17.75M CompactProtocol_write_BigListByte 18.49us 54.08K CompactProtocol_write_BigListBigInt 24.08us 41.52K CompactProtocol_write_BigListFloat 18.45us 54.20K CompactProtocol_write_BigListDouble 19.07us 52.43K CompactProtocol_write_BigListMixed 185.41us 5.39K CompactProtocol_write_BigListMixedInt 479.46us 2.09K CompactProtocol_write_LargeSetInt 149.90ms 6.67 CompactProtocol_write_LargeUnorderedMapMixed 22.94ms 43.58 CompactProtocol_write_LargeSortedVecMapMixed 22.56ms 44.33 CompactProtocol_write_NestedMap 541.29us 1.85K CompactProtocol_write_SortedVecNestedMap 548.71us 1.82K ``` After: full results (P1791763404) ``` CompactProtocol_write_MixedInt 30.46ns 32.83M CompactProtocol_write_BigListByte 6.26us 159.84K CompactProtocol_write_BigListBigInt 19.06us 52.48K CompactProtocol_write_BigListFloat 6.25us 160.10K CompactProtocol_write_BigListDouble 6.56us 152.39K CompactProtocol_write_BigListMixed 173.73us 5.76K CompactProtocol_write_BigListMixedInt 229.13us 4.36K CompactProtocol_write_LargeSetInt 107.13ms 9.33 CompactProtocol_write_LargeUnorderedMapMixed 18.93ms 52.84 CompactProtocol_write_LargeSortedVecMapMixed 18.68ms 53.54 CompactProtocol_write_NestedMap 419.08us 2.39K CompactProtocol_write_SortedVecNestedMap 412.94us 2.42K ``` Differential Revision: D73367438
|
This pull request was exported from Phabricator. Differential Revision: D73367438 |
|
Hi @embg! Thank you for your pull request. We require contributors to sign our Contributor License Agreement, and yours needs attention. You currently have a record in our system, but the CLA is no longer valid, and will need to be resubmitted. ProcessIn order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA. Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks! |
2 participants
Summary:
This change results in massive speedups on ProtocolBench. Prod canaries also gave some promising signals.
x86 ProtocolBench
Benchmarks not included below were neutral or only small wins. See pastes for full results.
Before: full results (P1791766582)
After: full results (P1791763018)
aarch64 ProtocolBench
Benchmarks not included below were neutral or only small wins. See pastes for full results.
Before: full results (P1791766333)
After: full results (P1791763404)
Differential Revision: D73367438