|
Revision tags: llvmorg-4.0.0-rc1 |
|
| #
cdff21b1 |
| 22-Dec-2016 |
Matt Arsenault <Matthew.Arsenault@amd.com> |
AMDGPU: Allow rcp and rsq usage with f16
llvm-svn: 290302
|
| #
269ffdac |
| 06-Dec-2016 |
Matt Arsenault <Matthew.Arsenault@amd.com> |
AMDGPU: Fix crash on i16 constant expression
llvm-svn: 288861
|
|
Revision tags: llvmorg-3.9.1, llvmorg-3.9.1-rc3, llvmorg-3.9.1-rc2, llvmorg-3.9.1-rc1 |
|
| #
c09e2d7e |
| 07-Oct-2016 |
Konstantin Zhuravlyov <kzhuravl_dev@outlook.com> |
[AMDGPU] AMDGPUCodeGenPrepare: remove extra ';'
llvm-svn: 283558
|
| #
f74fc60a |
| 07-Oct-2016 |
Konstantin Zhuravlyov <kzhuravl_dev@outlook.com> |
[AMDGPU] Promote uniform (i1, i16] operations to i32
Differential Revision: https://reviews.llvm.org/D25302
llvm-svn: 283555
|
| #
b4eb5d50 |
| 06-Oct-2016 |
Konstantin Zhuravlyov <kzhuravl_dev@outlook.com> |
[AMDGPU] Promote uniform i16 bitreverse intrinsic to i32
Differential Revision: https://reviews.llvm.org/D25121
llvm-svn: 283415
|
| #
691e2e02 |
| 03-Oct-2016 |
Konstantin Zhuravlyov <kzhuravl_dev@outlook.com> |
[AMDGPU] Sign extend AShr when promoting (instead of zero extending)
llvm-svn: 283130
|
| #
117296c0 |
| 01-Oct-2016 |
Mehdi Amini <mehdi.amini@apple.com> |
Use StringRef in Pass/PassManager APIs (NFC)
llvm-svn: 283004
|
| #
e14df4b2 |
| 28-Sep-2016 |
Konstantin Zhuravlyov <kzhuravl_dev@outlook.com> |
[AMDGPU] Promote uniform i16 ops to i32 ops for targets that have 16 bit instructions
Differential Revision: https://reviews.llvm.org/D24125
llvm-svn: 282624
|
|
Revision tags: llvmorg-3.9.0, llvmorg-3.9.0-rc3, llvmorg-3.9.0-rc2, llvmorg-3.9.0-rc1 |
|
| #
e3862cdc |
| 26-Jul-2016 |
Matt Arsenault <Matthew.Arsenault@amd.com> |
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.
llvm-svn: 276823
|
| #
a1fe17c9 |
| 19-Jul-2016 |
Matt Arsenault <Matthew.Arsenault@amd.com> |
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.
Simpli
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.
llvm-svn: 276051
show more ...
|
| #
86de486d |
| 24-Jun-2016 |
Matt Arsenault <Matthew.Arsenault@amd.com> |
AMDGPU: Add stub custom CodeGenPrepare pass
This will do various things including ones
CodeGenPrepare does, but with knowledge of uniform
values.
llvm-svn: 273657
|