Konstantin Zhuravlyov
f9bcd7b189
[AMDGPU] Promote uniform i16 ops to i32 ops for targets that have 16 bit instructions
...
Differential Revision: https://reviews.llvm.org/D24125
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@282624 91177308-0d34-0410-b5e6-96231b3b80d8
2016-09-28 20:05:39 +00:00
Matt Arsenault
f799c706db
AMDGPU: Use rcp for fdiv 1, x with fpmath metadata
...
Using rcp should be OK for safe math usually, so this
should not be replacing the original fdiv.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@276823 91177308-0d34-0410-b5e6-96231b3b80d8
2016-07-26 23:25:44 +00:00
Matt Arsenault
63be72069d
AMDGPU: Change fdiv lowering based on !fpmath metadata
...
If 2.5 ulp is acceptable, denormals are not required, and
isn't a reciprocal which will already be handled, replace
with a faster fdiv.
Simplify the lowering tests by using per function
subtarget features.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@276051 91177308-0d34-0410-b5e6-96231b3b80d8
2016-07-19 23:16:53 +00:00
Matt Arsenault
11c2d4bf28
AMDGPU: Add stub custom CodeGenPrepare pass
...
This will do various things including ones
CodeGenPrepare does, but with knowledge of uniform
values.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@273657 91177308-0d34-0410-b5e6-96231b3b80d8
2016-06-24 07:07:55 +00:00
