[PATCH] D89031: [SVE] Add support to vectorize_width loop pragma for scalable vectors

[David Sherwood via Phabricator via cfe-commits]

david-arm created this revision.
david-arm added reviewers: sdesmalen, ctetreau, fhahn, c-rhodes.
Herald added subscribers: cfe-commits, psnobl, tschuett.
Herald added a reviewer: efriedma.
Herald added a reviewer: aaron.ballman.
Herald added a project: clang.
david-arm requested review of this revision.

This patch adds support for an optional second parameter passed to
the vectorize_width pragma, which indicates if the user wishes
to use fixed width or scalable vectorization. For example the user
can now write something like:

  #pragma clang loop vectorize_width(4, fixed)

or

  #pragma clang loop vectorize_width(4, scalable)

I have added a new 'scalable_numeric' state to the LoopHintAttr class
to indicate whether the numeric vectorization width is scalable or
not. When generating IR we make use of the new format for the
llvm.loop.vectorize.width attribute that allows us to effectively pass
an ElementCount that contains the vectorization factor and a scalable
flag.

Tests were added to

  clang/test/CodeGenCXX/pragma-loop.cpp

for both the 'fixed' and 'scalable' optional parameter.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D89031

Files:
  clang/docs/LanguageExtensions.rst
  clang/include/clang/Basic/Attr.td
  clang/include/clang/Basic/DiagnosticParseKinds.td
  clang/lib/CodeGen/CGLoopInfo.cpp
  clang/lib/CodeGen/CGLoopInfo.h
  clang/lib/Parse/ParsePragma.cpp
  clang/lib/Sema/SemaStmtAttr.cpp
  clang/test/CodeGenCXX/pragma-loop.cpp

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D89031.296907.patch
Type: text/x-patch
Size: 12445 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20201008/abf55d80/attachment-0001.bin>