llvm-project/llvm
Arnold Schwaighofer cae8735a54 Costmodel: Add support for horizontal vector reductions
Upcoming SLP vectorization improvements will want to be able to estimate costs
of horizontal reductions. Add infrastructure to support this.

We model reductions as a series of (shufflevector,add) tuples ultimately
followed by an extractelement. For example, for an add-reduction of <4 x float>
we could generate the following sequence:

 (v0, v1, v2, v3)
   \   \  /  /
     \  \  /
       +  +

 (v0+v2, v1+v3, undef, undef)
    \      /
 ((v0+v2) + (v1+v3), undef, undef)

 %rdx.shuf = shufflevector <4 x float> %rdx, <4 x float> undef,
                           <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
 %bin.rdx = fadd <4 x float> %rdx, %rdx.shuf
 %rdx.shuf7 = shufflevector <4 x float> %bin.rdx, <4 x float> undef,
                          <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 %bin.rdx8 = fadd <4 x float> %bin.rdx, %rdx.shuf7
 %r = extractelement <4 x float> %bin.rdx8, i32 0

This commit adds a cost model interface "getReductionCost(Opcode, Ty, Pairwise)"
that will allow clients to ask for the cost of such a reduction (as backends
might generate more efficient code than the cost of the individual instructions
summed up). This interface is excercised by the CostModel analysis pass which
looks for reduction patterns like the one above - starting at extractelements -
and if it sees a matching sequence will call the cost model interface.

We will also support a second form of pairwise reduction that is well supported
on common architectures (haddps, vpadd, faddp).

 (v0, v1, v2, v3)
  \   /    \  /
 (v0+v1, v2+v3, undef, undef)
    \     /
 ((v0+v1)+(v2+v3), undef, undef, undef)

  %rdx.shuf.0.0 = shufflevector <4 x float> %rdx, <4 x float> undef,
        <4 x i32> <i32 0, i32 2 , i32 undef, i32 undef>
  %rdx.shuf.0.1 = shufflevector <4 x float> %rdx, <4 x float> undef,
        <4 x i32> <i32 1, i32 3, i32 undef, i32 undef>
  %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
  %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
        <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
  %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
        <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
  %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
  %r = extractelement <4 x float> %bin.rdx.1, i32 0

llvm-svn: 190876
2013-09-17 18:06:50 +00:00
..
autoconf Fix for executing AutoRegen.sh. Revert a part of r187209. 2013-09-13 10:29:42 +00:00
bindings [python-bindings] Added support for getting/setting operands of values and getting the number of operands of a value. 2013-09-11 01:38:12 +00:00
cmake [CMake] Hack GetSVN.cmake to handle unusual terminals. 2013-09-16 21:38:01 +00:00
docs Implement function prefix data as an IR feature. 2013-09-16 01:08:15 +00:00
examples ExceptionDemo.cpp: Tweak a @param. [-Wdocumentation] 2013-07-29 11:03:50 +00:00
include Costmodel: Add support for horizontal vector reductions 2013-09-17 18:06:50 +00:00
lib Costmodel: Add support for horizontal vector reductions 2013-09-17 18:06:50 +00:00
projects Port the detection of zlib from the main autoconf system to the sample 2013-08-18 01:55:15 +00:00
runtime Bring back the build of libprofile_rt on Sparc. It is now working correctly. See: 2013-09-08 09:15:09 +00:00
test Costmodel: Add support for horizontal vector reductions 2013-09-17 18:06:50 +00:00
tools ELF: Add support for the exclude section bit for gas compat. 2013-09-15 19:53:20 +00:00
unittests Re-submit r190469: YAMLIO: Fix string quoting logic. 2013-09-11 04:00:08 +00:00
utils TableGen: fix constness of new comparison function. 2013-09-16 17:33:40 +00:00
.arcconfig
.clang-format Add a clang-format file so that the tool can automatically detect the 2013-09-02 07:19:04 +00:00
.gitignore
CMakeLists.txt [conf] Add config variable to disable crash related overrides. 2013-08-30 20:39:21 +00:00
CODE_OWNERS.TXT Add more owners to CODE_OWNERS.TXT (Kostya Serebryany: AddressSanitizer and ThreadSanitizer; Evgeniy Stepanov: MemorySanitizer) 2013-06-27 08:47:12 +00:00
CREDITS.TXT Test commit. 2013-08-16 18:09:06 +00:00
LICENSE.TXT Be more specific and capitalize filenames. 2013-05-21 21:22:34 +00:00
LLVMBuild.txt
Makefile Fix regular expression used by 'make update' to only look for 'I' and '?' at the start of svn info results and to check for spaces after 'I' instead of just after '?'. 2013-07-03 14:48:37 +00:00
Makefile.common
Makefile.config.in Add an autoconf option for turning on -gsplit-dwarf by default 2013-06-25 01:12:25 +00:00
Makefile.rules Makefile.rules: Avoid -fomit-frame-pointer also on cygwin due to PR14646. 2013-08-18 03:38:40 +00:00
README.txt
configure [conf] Add config variable to disable crash related overrides. 2013-08-30 20:39:21 +00:00
llvm.spec.in

README.txt

Low Level Virtual Machine (LLVM)
================================

This directory and its subdirectories contain source code for the Low Level
Virtual Machine, 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're writing a package for LLVM, see docs/Packaging.rst for our
suggestions.