Improve Singlepass performance on Aarch64 (popcnt, select)
Closed this issue · 1 comments
syrusakbary commented
In ARM, Wasmer (with singlepass compiler) is not generating optimal code for the following opcodes:
- i32PopCnt: https://github.com/wasmerio/wasmer/blob/master/lib/compiler-singlepass/src/machine_arm64.rs#L3070-L3115
- i64PopCnt: https://github.com/wasmerio/wasmer/blob/master/lib/compiler-singlepass/src/machine_arm64.rs#L5146-L5189
- Select
We should aim for the Aarch64 code to be as optimal to the x64