Kshitij Patil on Nostr: Hey npub1slp5q…e4xfq, thanks for this article ...
Hey npub1slp5q3n9nyjn3p28hnxmvj5x82w6yw03cvplvkjdrnnuqqw4et3ste4xfq (npub1slp…4xfq), thanks for this article
https://blog.p-y.wtf/statistically-rigorous-android-macrobenchmarks
It has really helped us standardising the benchmarking process. One question -
You mentioned to use full AOT compilation while benchmarking, but I noticed on perfetto while benchmarking the app startup path, there’s still a lot of JIT compilation after doing full AOT compilation. Is this expected or is a sign of doing something wrong?
https://blog.p-y.wtf/statistically-rigorous-android-macrobenchmarks
It has really helped us standardising the benchmarking process. One question -
You mentioned to use full AOT compilation while benchmarking, but I noticed on perfetto while benchmarking the app startup path, there’s still a lot of JIT compilation after doing full AOT compilation. Is this expected or is a sign of doing something wrong?