This is an archive of the discontinued LLVM Phabricator instance.

[TTI][AArch64][CostModel] Adjusting the cost of vector element insert/extract operations for ARM based Apple CPUs
DraftPublic

Authored by nilanjana_basu on May 26 2023, 7:04 PM.
This is a draft revision that has not yet been submitted for review.

Details

Reviewers
fhahn
Summary

This recovers an SLP regression revealed by D146033 while showing a maximum performance lift of 4% on some other benchmarks.

Diff Detail

Event Timeline

nilanjana_basu created this revision.May 26 2023, 7:04 PM