1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 03:02:36 +01:00
Go to file
Sanjay Patel 50d6ec057c [x86] avoid 256-bit andnp that requires insert/extract with AVX1 (PR37449)
This is the final (I hope!) problem pattern mentioned in PR37749:
https://bugs.llvm.org/show_bug.cgi?id=37749

We are trying to avoid an AVX1 sinkhole caused by having 256-bit bitwise logic ops but no other 256-bit integer ops. 
We've already solved the simple logic ops, but 'andn' is an x86 special. I looked at alternative solutions like 
extending the generic DAG combine or trying to wait until the ANDNP node is created, but those are bigger patches 
that can over-reach. Ie, splitting to 128-bit does not look like a win in most cases with >1 256-bit op.

The pattern matching is cluttered with bitcasts because of our i64 element canonicalization. For the affected test, 
we have this vector-type-legalized sequence:

        t29: v8i32 = concat_vectors t27, t28
      t30: v4i64 = bitcast t29
        t18: v8i32 = BUILD_VECTOR Constant:i32<-1>, Constant:i32<-1>, ...
      t31: v4i64 = bitcast t18
    t32: v4i64 = xor t30, t31
      t9: v8i32 = BUILD_VECTOR Constant:i32<255>, Constant:i32<255>, ...
    t34: v4i64 = bitcast t9
  t35: v4i64 = and t32, t34
t36: v8i32 = bitcast t35
      t37: v4i32 = extract_subvector t36, Constant:i64<0>
      t38: v4i32 = extract_subvector t36, Constant:i64<4>

Differential Revision: https://reviews.llvm.org/D52318

llvm-svn: 343008
2018-09-25 19:09:34 +00:00
benchmarks Pull google/benchmark library to the LLVM tree 2018-08-28 09:42:41 +00:00
bindings [LLVM-C][OCaml] Add UnifyFunctionExitNodes pass to C and OCaml APIs 2018-09-18 13:36:03 +00:00
cmake Move individual benchmark targets into the Utils folder in IDEs. 2018-09-21 23:01:32 +00:00
docs [llvm-exegesis] Fix doc in r342947. 2018-09-25 07:48:38 +00:00
examples [ORC] Update JITCompileCallbackManager to support multi-threaded code. 2018-05-30 01:57:45 +00:00
include Revert r342637 "[ADT] Try again to use the same version of llvm::Optional on all compilers" 2018-09-25 12:08:56 +00:00
lib [x86] avoid 256-bit andnp that requires insert/extract with AVX1 (PR37449) 2018-09-25 19:09:34 +00:00
projects
resources
runtimes Revert "[CMake] Pass Clang defaults to runtimes builds" 2018-07-13 20:01:55 +00:00
test [x86] avoid 256-bit andnp that requires insert/extract with AVX1 (PR37449) 2018-09-25 19:09:34 +00:00
tools [llvm-exegesis] Serializes registers initial values. 2018-09-25 15:15:54 +00:00
unittests [llvm-exegesis] Serializes registers initial values. 2018-09-25 15:15:54 +00:00
utils [globalisel][tblgen] Table optimization should consider the C++ code in C++ predicates 2018-09-25 17:59:02 +00:00
.arcconfig
.clang-format
.clang-tidy
.gitattributes [DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label. 2018-05-09 02:40:45 +00:00
.gitignore
CMakeLists.txt [WebAssembly] Revert r342701, "Add WebAssembly to LLVM_ALL_TARGTS." 2018-09-21 03:24:42 +00:00
CODE_OWNERS.TXT Add owner for llvm-objcopy 2018-08-09 22:05:19 +00:00
configure
CREDITS.TXT Update my information in the CREDITS file. 2018-06-15 20:02:11 +00:00
LICENSE.TXT Update copyright year to 2018. 2018-06-18 12:22:17 +00:00
llvm.spec.in
LLVMBuild.txt
README.txt Test commit. 2018-09-22 01:01:03 +00:00
RELEASE_TESTERS.TXT Remove myself from the release testers list. (NFC) 2018-06-20 21:25:50 +00:00

The LLVM Compiler Infrastructure
================================

This directory and its subdirectories contain source code for LLVM,
a toolkit for the construction of highly optimized compilers,
optimizers, and runtime environments.

LLVM is open source software. You may freely distribute it under the terms of
the license agreement found in LICENSE.txt.

Please see the documentation provided in docs/ for further
assistance with LLVM, and in particular docs/GettingStarted.rst for getting
started with LLVM and docs/README.txt for an overview of LLVM's
documentation setup.

If you are writing a package for LLVM, see docs/Packaging.rst for our
suggestions.