aco: store 16-bit temporary outputs as v2b

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4966>

diff --git a/src/amd/compiler/aco_instruction_selection.cpp b/src/amd/compiler/aco_instruction_selection.cpp
index 4a7cef2d7d7d6624824f1cbef145fa1308029b13..f83aae2b179a05ce1bab6fb001dd70e0a5a8809b 100644 (file)
--- a/src/amd/compiler/aco_instruction_selection.cpp
+++ b/src/amd/compiler/aco_instruction_selection.cpp
@@ -4213,10 +4213,12 @@ bool store_output_to_temps(isel_context *ctx, nir_intrinsic_instr *instr)
    if (instr->src[0].ssa->bit_size == 64)
       write_mask = widen_mask(write_mask, 2);
 
+   RegClass rc = instr->src[0].ssa->bit_size == 16 ? v2b : v1;
+
    for (unsigned i = 0; i < 8; ++i) {
       if (write_mask & (1 << i)) {
          ctx->outputs.mask[idx / 4u] |= 1 << (idx % 4u);
-         ctx->outputs.temps[idx] = emit_extract_vector(ctx, src, i, v1);
+         ctx->outputs.temps[idx] = emit_extract_vector(ctx, src, i, rc);
       }
       idx++;
    }