The DPP source and e64 destination operands remain unchanged for now.
Details
Details
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop2_dpp16.txt | ||
---|---|---|
60 ↗ | (On Diff #543468) | Why has only the dst changed to a 16 bit register? |
Comment Actions
After this patch, it seems the disassembly output for 16-bit instructions will be changed. We we be in a state where the ones using true16 print .l/.h suffixes and the other ones don't. Is that a problem for users?
llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop2_dpp16.txt | ||
---|---|---|
60 ↗ | (On Diff #543468) | The DPP source operand fields in VOPProfile_True16 still refer to the Fake16 operands. |
llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop2_dpp16.txt | ||
---|---|---|
60 ↗ | (On Diff #543468) | The disassembler changes in this patch include:
|