maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
315,155 Commits 23 Branches 0 Tags 2.2 GiB
C++ 39.1%
LLVM 35.7%
C 12.1%
Assembly 8.9%
Python 1.1%
Other 2.7%
Go to file
Sanjay Patel fb9a5307a9 [DAGCombiner] try repeated fdiv divisor transform before building estimate 
This was originally part of D61028, but it's an independent diff.

If we try the repeated divisor reciprocal transform before producing an estimate sequence,
then we have an opportunity to use scalar fdiv. On x86, the trade-off is 1 divss vs. 5
vector FP ops in the default estimate sequence. On recent chips (Skylake, Ryzen), the
full-precision division is only 3 cycle throughput, so that's probably the better perf
default option and avoids problems from x86's inaccurate estimates.

The last 2 tests show that users still have the option to override the defaults by using
the function attributes for reciprocal estimates, but those patterns are potentially made
faster by converting the vector ops (including ymm ops) to scalar math.

Differential Revision: https://reviews.llvm.org/D61149

llvm-svn: 359398
 		2019-04-28 12:23:43 +00:00
clang [PowerPC][Clang] Add tests for PowerPC MMX intrinsics 2019-04-28 06:27:33 +00:00
clang-tools-extra Fix typo in documentation. 2019-04-26 17:56:22 +00:00
compiler-rt [CMake] Don't modify `FUZZER_SUPPORTED_ARCH` is place. 2019-04-28 09:44:53 +00:00
debuginfo-tests Set config.lit_tools_dir, which is needed by lit.llvm.initialize. 2018-11-06 21:54:27 +00:00
libclc travis: Add LLVM-8 build 2019-03-27 21:28:31 +00:00
libcxx Add '_LIBCPP_ASSERT(ready())' to several match_results method that have this precondtion. Fix several tests which did not honor this precondition. Thanks to Andrey Maksimov for pointing this out. 2019-04-26 17:10:03 +00:00
libcxxabi Fix compilation error with -DLIBCXXABI_ENABLE_THREADS=OFF 2019-04-25 09:27:50 +00:00
libunwind [NFC] Fix typo in debug log 2019-04-22 15:40:50 +00:00
lld [LLD][ELF] - Do not remove empty sections referenced in LOADADDR/ADDR commands. 2019-04-26 06:59:30 +00:00
lldb Fix UNPREDICTABLE check in EmulateInstructionARM::EmulateADDRegShift 2019-04-27 18:15:35 +00:00
llgo IR: Support parsing numeric block ids, and emit them in textual output. 2019-03-22 18:27:13 +00:00
llvm [DAGCombiner] try repeated fdiv divisor transform before building estimate 2019-04-28 12:23:43 +00:00
openmp [OPENMP][NVPTX]Correctly handle L2 parallelism in SPMD mode. 2019-04-26 19:30:34 +00:00
parallel-libs Fix typos throughout the license files that somehow I and my reviewers 2019-01-21 09:52:34 +00:00
polly Apply include-what-you-use #include removal suggestions. NFC. 2019-03-28 20:19:49 +00:00
pstl [pstl] Make the default backend be the serial backend and always provide parallel policies 2019-04-24 20:12:36 +00:00
.arcconfig Update monorepo .arcconfig with new project callsign. 2019-01-31 14:34:59 +00:00
.clang-format Add .clang-tidy and .clang-format files to the toplevel of the 2019-01-29 16:43:16 +00:00
.clang-tidy Disable tidy checks with too many hits 2019-02-01 11:20:13 +00:00
.gitignore Add a reduced copy of the llvm .gitignore 2019-04-09 00:52:49 +00:00
README.md
…

README.md

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.