[PATCH] D25298: [AMDGPU] Refactor waitcnt encoding

[Konstantin Zhuravlyov via llvm-commits]

kzhuravl added inline comments.


================
Comment at: lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp:55-71
+/// \returns Vmcnt bit shift for given isa \p Version.
+unsigned getVmcntBitShift(IsaVersion Version) { return 0; }
+
+/// \returns Vmcnt bit width for given isa \p Version.
+unsigned getVmcntBitWidth(IsaVersion Version) { return 4; }
+
+/// \returns Expcnt bit shift for given isa \p Version.
----------------
kzhuravl wrote:
> arsenm wrote:
> > These don't use the version?
> Not at this time, but will in a separate patch
After giving it a second thought, we do not need version in get*BitShift and get*BitWidth functions. I have removed version from those functions.


https://reviews.llvm.org/D25298