Evghenii fe98fe8cdc added fast approximate rcp(double) accurate to 15 digits

2014-02-04 15:23:34 +01:00

..

builtins.c

Remove sprintf warnings on Windows and fix sprintf-related fails on Mac

2013-04-24 22:36:48 +02:00

dispatch.ll

Adding check for OS AVX support to auto-dispatch code

2013-09-19 15:39:56 +04:00

svml.m4

included suggested changes, ./tests/launch-*.ispc still fails. something is mask64 related, not sure what. help...

2013-09-12 17:18:12 +02:00

target-avx1-i64x4.ll

renamed avx-i64x4 -> avx1-i64x4

2013-09-13 15:07:14 +02:00

target-avx1-i64x4base.ll

added support for fast approximate rsqrt(double). Provide 16 digit accurancy but is over 3x faster than 1/sqrt(double)

2014-02-04 14:44:54 +01:00

target-avx1-x2.ll

Rename gen_gather() macro to gen_gather_factored()

2012-07-13 12:24:12 -07:00

target-avx1.ll

Rename gen_gather() macro to gen_gather_factored()

2012-07-13 12:24:12 -07:00

target-avx2-i64x4.ll

removed LLVM_3_0 legacy part and changed copyright to 2013

2013-10-18 08:53:01 +02:00

target-avx2-x2.ll

Revert "Remove support for building with LLVM 3.1"

2013-08-06 17:00:35 -07:00

target-avx2.ll

Revert "Remove support for building with LLVM 3.1"

2013-08-06 17:00:35 -07:00

target-avx11-i64x4.ll

removed LLVM_3_0 legacy part and changed copyright to 2013

2013-10-18 08:53:01 +02:00

target-avx11-x2.ll

Revert "Remove support for building with LLVM 3.1"

2013-08-06 17:00:35 -07:00

target-avx11.ll

Revert "Remove support for building with LLVM 3.1"

2013-08-06 17:00:35 -07:00

target-avx-common.ll

Ordering functions in target-[avx|sse2].ll to be in the same order. No real changes, except adding a few alwaysinline in SSE4 target

2013-11-14 15:34:30 +04:00

target-avx-x2.ll

+1

2014-02-02 18:16:48 +01:00

target-avx.ll

+1

2014-02-02 18:16:48 +01:00

target-generic-1.ll

+1

2014-02-02 18:16:48 +01:00

target-generic-4.ll

Add "generic" 4, 8, and 16-wide targets.

2011-12-19 13:46:50 -08:00

target-generic-8.ll

Add "generic" 4, 8, and 16-wide targets.

2011-12-19 13:46:50 -08:00

target-generic-16.ll

Add "generic" 4, 8, and 16-wide targets.

2011-12-19 13:46:50 -08:00

target-generic-32.ll

Add generic-32 target.

2012-05-03 11:11:06 -07:00

target-generic-64.ll

Add support for generic-64 target.

2012-05-25 11:57:28 -07:00

target-generic-common.ll

added support for fast approximate rsqrt(double). Provide 16 digit accurancy but is over 3x faster than 1/sqrt(double)

2014-02-04 14:44:54 +01:00

target-neon-8.ll

+1

2014-02-02 18:16:48 +01:00

target-neon-16.ll

+1

2014-02-02 18:16:48 +01:00

target-neon-32.ll

+1

2014-02-02 18:16:48 +01:00

target-neon-common.ll

included suggested changes, ./tests/launch-*.ispc still fails. something is mask64 related, not sure what. help...

2013-09-12 17:18:12 +02:00

target-sse2-common.ll

Add avg_{up,down}_int{8,16} routines to stdlib

2013-08-06 08:41:12 -07:00

target-sse2-x2.ll

+1

2014-02-02 18:16:48 +01:00

target-sse2.ll

+1

2014-02-02 18:16:48 +01:00

target-sse4-8.ll

+1

2014-02-02 18:16:48 +01:00

target-sse4-16.ll

+1

2014-02-02 18:16:48 +01:00

target-sse4-common.ll

target-[sse4|avx]_common.ll are twin brothers, which diffes only cosmetically. This commit makes them diffable. No real changes, except adding alwaysinline to sse version iof __max_uniform_int32/__max_uniform_uint32

2013-11-14 15:34:30 +04:00

target-sse4-x2.ll

+1

2014-02-02 18:16:48 +01:00

target-sse4.ll

+1

2014-02-02 18:16:48 +01:00

util.m4

added fast approximate rcp(double) accurate to 15 digits

2014-02-04 15:23:34 +01:00