diff options
| author | Samuel Pitoiset <[email protected]> | 2020-05-20 11:32:50 +0200 |
|---|---|---|
| committer | Marge Bot <[email protected]> | 2020-05-21 15:06:48 +0000 |
| commit | 2d4493ee11aa653ed4c2cc51b37f9e60b534b8bc (patch) | |
| tree | 4e0a6494bb4e11a45247804fb93cc5aa5bdd3cd2 | |
| parent | c76595aec2c47463a3ae580c56bd19191f185acf (diff) | |
aco: sign-extend the input and identity for 8-bit subgroup operations
Signed-off-by: Samuel Pitoiset <[email protected]>
Reviewed-by: Rhys Perry <[email protected]>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4494>
| -rw-r--r-- | src/amd/compiler/aco_lower_to_hw_instr.cpp | 12 |
1 files changed, 12 insertions, 0 deletions
diff --git a/src/amd/compiler/aco_lower_to_hw_instr.cpp b/src/amd/compiler/aco_lower_to_hw_instr.cpp index 1d3061d5dd9..3bd9647f781 100644 --- a/src/amd/compiler/aco_lower_to_hw_instr.cpp +++ b/src/amd/compiler/aco_lower_to_hw_instr.cpp @@ -506,6 +506,18 @@ void emit_reduction(lower_context *ctx, aco_opcode op, ReduceOp reduce_op, unsig Operand(stmp, bld.lm)); } + if (src.regClass() == v1b) { + aco_ptr<SDWA_instruction> sdwa{create_instruction<SDWA_instruction>(aco_opcode::v_mov_b32, asSDWA(Format::VOP1), 1, 1)}; + sdwa->operands[0] = Operand(PhysReg{tmp}, v1); + sdwa->definitions[0] = Definition(PhysReg{tmp}, v1); + if (reduce_op == imin8 || reduce_op == imax8) + sdwa->sel[0] = sdwa_sbyte; + else + sdwa->sel[0] = sdwa_ubyte; + sdwa->dst_sel = sdwa_udword; + bld.insert(std::move(sdwa)); + } + bool reduction_needs_last_op = false; switch (op) { case aco_opcode::p_reduce: |