[PATCH] D113350: [X86][Costmodel] `getReplicationShuffleCost()`: implement cost model for 32/64 bit-wide elements with AVX512F

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri created this revision.
lebedev.ri added reviewers: RKSimon, spatel.
lebedev.ri added a project: LLVM.
Herald added subscribers: pengfei, hiraditya.
lebedev.ri requested review of this revision.

This models lowering to `vpermd`/`vpermq`/`vpermps`/`vpermpd`,
that take a single input vector and a single index vector,
and are cross-lane. So far i haven't seen evidence that 
replication ever results in demanding more than a single 
input vector per output vector.

This results in *shockingly* lesser costs :)


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D113350

Files:
  llvm/lib/Target/X86/X86TargetTransformInfo.cpp
  llvm/lib/Target/X86/X86TargetTransformInfo.h
  llvm/test/Analysis/CostModel/X86/shuffle-replication-i32.ll
  llvm/test/Analysis/CostModel/X86/shuffle-replication-i64.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D113350.385293.patch
Type: text/x-patch
Size: 78552 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20211106/5e8a66f8/attachment-0001.bin>