llvm-project

Commit Graph

Author	SHA1	Message	Date
Davide Italiano	9c26161b2e	[SimplifyLibCalls] Remove useless bits of this tests. llvm-svn: 254318	2015-11-30 19:38:35 +00:00
Davide Italiano	1aeed6a955	[SimplifyLibCalls] Transform log(exp2(y)) to y*log(2) under fast-math. llvm-svn: 254317	2015-11-30 19:36:35 +00:00
Davide Italiano	b8b7133c94	[SimplifyLibCalls] Tranform log(pow(x, y)) -> ylog(x). This one is enabled only under -ffast-math. There are cases where the difference between the value computed and the correct value is huge even for ffast-math, e.g. as Steven pointed out: x = -1, y = -4 log(pow(-1), 4) = 0 4log(-1) = NaN I checked what GCC does and apparently they do the same optimization (which result in the dramatic difference). Future work might try to make this (slightly) less worse. Differential Revision: http://reviews.llvm.org/D14400 llvm-svn: 254263	2015-11-29 20:58:04 +00:00

Author

SHA1

Message

Date

Davide Italiano

9c26161b2e

[SimplifyLibCalls] Remove useless bits of this tests.

llvm-svn: 254318

2015-11-30 19:38:35 +00:00

Davide Italiano

1aeed6a955

[SimplifyLibCalls] Transform log(exp2(y)) to y*log(2) under fast-math.

llvm-svn: 254317

2015-11-30 19:36:35 +00:00

Davide Italiano

b8b7133c94

[SimplifyLibCalls] Tranform log(pow(x, y)) -> y*log(x).

This one is enabled only under -ffast-math. There are cases where the
difference between the value computed and the correct value is huge
even for ffast-math, e.g. as Steven pointed out:

x = -1, y = -4
log(pow(-1), 4) = 0
4*log(-1) = NaN

I checked what GCC does and apparently they do the same optimization
(which result in the dramatic difference). Future work might try to
make this (slightly) less worse.

Differential Revision:	http://reviews.llvm.org/D14400

llvm-svn: 254263

2015-11-29 20:58:04 +00:00

3 Commits