Revision 1c5d7339e72b64ee9289c3645eb651cc611c18b1 authored by Jameson Nash on 22 February 2017, 19:58:46 UTC, committed by Jameson Nash on 22 February 2017, 20:01:19 UTC
The powi intrinsic optimization over calling powf is that it is inaccurate. We don't need that. When it is equally accurate (e.g. tiny constant powers), LLVM will already recognize and optimize any call to a function named `powf`, and produce the same speedup. fix #19872
1 parent 5a1f971
| File | Mode | Size |
|---|---|---|
| checksums | ||
| patches | ||
| tools | ||
| valgrind | ||
| .gitignore | -rw-r--r-- | 26 bytes |
| Makefile | -rw-r--r-- | 5.0 KB |
| NATIVE.cmake | -rw-r--r-- | 192 bytes |
| SuiteSparse_wrapper.c | -rw-r--r-- | 1.4 KB |
| Versions.make | -rw-r--r-- | 267 bytes |
| arpack.mk | -rw-r--r-- | 4.5 KB |
| blas.mk | -rw-r--r-- | 7.3 KB |
| curl.mk | -rw-r--r-- | 2.4 KB |
| dsfmt.mk | -rw-r--r-- | 2.2 KB |
| fftw.mk | -rw-r--r-- | 5.2 KB |
| gfortblas.alias | -rw-r--r-- | 706 bytes |
| gfortblas.c | -rw-r--r-- | 4.4 KB |
| gmp.mk | -rw-r--r-- | 2.3 KB |
| libdSFMT.def | -rw-r--r-- | 778 bytes |
| libgit2.mk | -rw-r--r-- | 5.3 KB |
| libgit2.version | -rw-r--r-- | 77 bytes |
| libssh2.mk | -rw-r--r-- | 2.7 KB |
| libssh2.version | -rw-r--r-- | 83 bytes |
| libuv.mk | -rw-r--r-- | 1.9 KB |
| libuv.version | -rw-r--r-- | 79 bytes |
| llvm-ver.make | -rw-r--r-- | 484 bytes |
| llvm.mk | -rw-r--r-- | 22.1 KB |
| mbedtls.mk | -rw-r--r-- | 3.6 KB |
| mpfr.mk | -rw-r--r-- | 2.3 KB |
| objconv.mk | -rw-r--r-- | 958 bytes |
| openblas.version | -rw-r--r-- | 79 bytes |
| openlibm.mk | -rw-r--r-- | 1.1 KB |
| openlibm.version | -rw-r--r-- | 78 bytes |
| openspecfun.mk | -rw-r--r-- | 1.6 KB |
| openspecfun.version | -rw-r--r-- | 84 bytes |
| patchelf.mk | -rw-r--r-- | 1.9 KB |
| pcre.mk | -rw-r--r-- | 1.9 KB |
| suitesparse.mk | -rw-r--r-- | 7.2 KB |
| unwind.mk | -rw-r--r-- | 5.6 KB |
| utf8proc.mk | -rw-r--r-- | 1.5 KB |
| utf8proc.version | -rw-r--r-- | 76 bytes |
Computing file changes ...
