AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

radv,aco: implement nir_op_ffma

Browse Source 
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9805>
This commit is contained in:
Rhys Perry
2021-03-24 17:17:38 +00:00
committed by Marge Bot
parent c5f02a1cd3
commit 165ca5088b
4 changed files with 32 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/amd/compiler/aco_instruction_selection.cpp
+29
View File
@@ -2092,6 +2092,35 @@ visit_alu_instr(isel_context* ctx, nir_alu_instr* instr)
}
break;
}
case nir_op_ffma: {
if (dst.regClass() == v2b) {
emit_vop3a_instruction(ctx, instr, aco_opcode::v_fma_f16, dst, false, 3);
} else if (dst.regClass() == v1 && instr->dest.dest.ssa.bit_size == 16) {
assert(instr->dest.dest.ssa.num_components == 2);
Temp src0 = as_vgpr(ctx, get_alu_src_vop3p(ctx, instr->src[0]));
Temp src1 = as_vgpr(ctx, get_alu_src_vop3p(ctx, instr->src[1]));
Temp src2 = as_vgpr(ctx, get_alu_src_vop3p(ctx, instr->src[2]));
/* swizzle to opsel: all swizzles are either 0 (x) or 1 (y) */
unsigned opsel_lo = 0, opsel_hi = 0;
for (unsigned i = 0; i < 3; i++) {
opsel_lo |= (instr->src[i].swizzle[0] & 1) << i;
opsel_hi |= (instr->src[i].swizzle[1] & 1) << i;
}
bld.vop3p(aco_opcode::v_pk_fma_f16, Definition(dst), src0, src1, src2, opsel_lo, opsel_hi);
emit_split_vector(ctx, dst, 2);
} else if (dst.regClass() == v1) {
emit_vop3a_instruction(ctx, instr, aco_opcode::v_fma_f32, dst,
ctx->block->fp_mode.must_flush_denorms32, 3);
} else if (dst.regClass() == v2) {
emit_vop3a_instruction(ctx, instr, aco_opcode::v_fma_f64, dst, false, 3);
} else {
isel_err(&instr->instr, "Unimplemented NIR instr bit size");
}
break;
}
case nir_op_fmax: {
if (dst.regClass() == v2b) {
// TODO: check fp_mode.must_flush_denorms16_64