This pragma forces compiler to align loop headers within the
function by architecture specific boundary: 32 bytes on X64
and ARM64 (with the exception of Apple Silicon, which explicitly
discourages aligning branch targets in the optimization manual).
Current implementation is rather naive and does not do any
attempt to decide whether aligning is actually profitable
based on loop body itself.
I have found this pragma to be helpful both to stabilize
benchmark results and achieve better performance
for tight loops on Intel hardware.
Issue https://github.com/dart-lang/sdk/issues/55522
TEST=vm/dart/align_loops_test
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-product-x64-try,vm-aot-linux-debug-x64c-try,vm-aot-linux-debug-x64-try,vm-aot-linux-release-arm64-try,vm-aot-linux-release-simarm_x64-try,vm-aot-linux-release-x64-try,vm-aot-mac-product-arm64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-obfuscate-linux-release-x64-try,vm-aot-optimization-level-linux-release-x64-try,vm-aot-win-release-x64-try,vm-aot-win-debug-arm64-try,vm-aot-win-debug-x64-try,vm-aot-win-debug-x64c-try,vm-aot-win-product-x64-try,vm-aot-win-release-arm64-try,vm-aot-linux-debug-simarm_x64-try,vm-aot-linux-debug-simriscv64-try,vm-aot-dwarf-linux-product-x64-try,vm-aot-android-release-arm64c-try,vm-aot-android-release-arm_x64-try
Change-Id: Ic22fb90d85e7fdebeeaa3908a43328c59436ab58
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/364121
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
Vyacheslav Egorov
2a2781edff