[PATCH] D26767: [AMDGPU] Promote f16/i16 conversions to f32/i32 + custom lower f16 = fp_round f64

[Konstantin Zhuravlyov via llvm-commits]

kzhuravl created this revision.
kzhuravl added reviewers: tstellarAMD, arsenm.
kzhuravl added a subscriber: llvm-commits.
Herald added subscribers: tony-tye, yaxunl, nhaehnle, wdng.

I will split this change in 2 when submitting (it is easier to test everything in bulk):

- Promote f16/i16 conversions to f32/i32
- Custom lower f16 = fp_round f64

Testing done:

- Conformance: half (passed), conversions (in progress)
- make check-all (passed)


https://reviews.llvm.org/D26767

Files:
  lib/Target/AMDGPU/SIISelLowering.cpp
  lib/Target/AMDGPU/SIISelLowering.h
  test/CodeGen/AMDGPU/fp_to_sint.ll
  test/CodeGen/AMDGPU/fp_to_uint.ll
  test/CodeGen/AMDGPU/fptosi.f16.ll
  test/CodeGen/AMDGPU/fptoui.f16.ll
  test/CodeGen/AMDGPU/sitofp.f16.ll
  test/CodeGen/AMDGPU/uitofp.f16.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D26767.78256.patch
Type: text/x-patch
Size: 16120 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20161116/652c73cf/attachment.bin>