[PATCH] D99723: Transforming memcpy to Tail predicated Loop

[Malhar via Phabricator via llvm-commits]

malharJ created this revision.
malharJ added reviewers: dmgreen, SjoerdMeijer.
Herald added subscribers: hiraditya, kristof.beyls.
malharJ requested review of this revision.
Herald added a project: LLVM.
Herald added a subscriber: llvm-commits.

This patch converts llvm.memcpy intrinsic into Tail Predicated
Hardware loops for a target that supports the Arm M-profile
Vector Extension (MVE).

The patch adds a custom inserter (if optimization level is not -O0)
at the LLVM Machine IR (MIR) level that converts the llvm.memcpy
intrinsic into MIR suitable to be optimized (by later passes) into a WLSTP loop.

Depends on D99649 <https://reviews.llvm.org/D99649>


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D99723

Files:
  llvm/lib/Target/ARM/ARMISelLowering.cpp
  llvm/lib/Target/ARM/ARMISelLowering.h
  llvm/lib/Target/ARM/ARMInstrMVE.td
  llvm/lib/Target/ARM/ARMSelectionDAGInfo.cpp
  llvm/lib/Target/ARM/ARMSubtarget.h
  llvm/test/CodeGen/Thumb2/LowOverheadLoops/memcall.ll
  llvm/test/CodeGen/Thumb2/mve_tp_loop.ll
  llvm/test/CodeGen/Thumb2/mve_tp_loop.mir

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D99723.334666.patch
Type: text/x-patch
Size: 31976 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210401/04019935/attachment.bin>