Matt Arsenault 01807baf69 TTI: Improve default costs for addrspacecast ...

For some reason multiple places need to do this, and the variant the
loop unroller and inliner use was not handling it.

Also, introduce a new wrapper to be slightly more precise, since on
AMDGPU some addrspacecasts are free, but not no-ops.

llvm-svn: 362436

2019-06-03 18:41:34 +00:00

..

add-sub.ll

[AMDGPU] Prepare for introduction of v3 and v5 MVTs

2019-03-17 21:04:16 +00:00

addrspacecast.ll

TTI: Improve default costs for addrspacecast

2019-06-03 18:41:34 +00:00

bit-ops.ll

…

br.ll

…

extractelement.ll

[AMDGPU] Prepare for introduction of v3 and v5 MVTs

2019-03-17 21:04:16 +00:00

fabs.ll

[AMDGPU] Prepare for introduction of v3 and v5 MVTs

2019-03-17 21:04:16 +00:00

fadd.ll

[AMDGPU] Prepare for introduction of v3 and v5 MVTs

2019-03-17 21:04:16 +00:00

fdiv.ll

AMDGPU: Partially fix default device for HSA

2019-03-17 21:31:35 +00:00

fmul.ll

[AMDGPU] Prepare for introduction of v3 and v5 MVTs

2019-03-17 21:04:16 +00:00

fsub.ll

[AMDGPU] Prepare for introduction of v3 and v5 MVTs

2019-03-17 21:04:16 +00:00

insertelement.ll

…

lit.local.cfg

…

mul.ll

[AMDGPU] Prepare for introduction of v3 and v5 MVTs

2019-03-17 21:04:16 +00:00

shifts.ll

…

shufflevector.ll

…