Commit ee4bc13d authored by Samuel Pitoiset's avatar Samuel Pitoiset

aco: use v_bfe_u32 for unsigned reductions sign-extension on GFX6-GFX7

Signed-off-by: Samuel Pitoiset's avatarSamuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann's avatarDaniel Schürmann <daniel@schuermann.dev>
Part-of: <!5327>
parent a874132c
......@@ -591,7 +591,14 @@ void emit_reduction(lower_context *ctx, aco_opcode op, ReduceOp reduce_op, unsig
sdwa->dst_sel = sdwa_udword;
bld.insert(std::move(sdwa));
} else if (ctx->program->chip_class == GFX6 || ctx->program->chip_class == GFX7) {
bld.vop3(aco_opcode::v_bfe_i32, Definition(PhysReg{tmp}, v1),
aco_opcode opcode;
if (reduce_op == imin16 || reduce_op == imax16 || reduce_op == iadd16)
opcode = aco_opcode::v_bfe_i32;
else
opcode = aco_opcode::v_bfe_u32;
bld.vop3(opcode, Definition(PhysReg{tmp}, v1),
Operand(PhysReg{tmp}, v1), Operand(0u), Operand(16u));
}
}
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment