Mesa (staging/21.0): aco: Mark VCC clobbered for iadd8 and iadd16 reductions on GFX6-7.

[GitLab Mirror]

Module: Mesa
Branch: staging/21.0
Commit: d32e751681e8aa8849e7a00561f91e0bd7167243
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=d32e751681e8aa8849e7a00561f91e0bd7167243

Author: Timur Kristóf <timur.kristof at gmail.com>
Date:   Tue Apr 20 13:55:23 2021 +0200

aco: Mark VCC clobbered for iadd8 and iadd16 reductions on GFX6-7.

On GFX6-7, the 8 and 16-bit integer add reductions use the 32-bit v_add
instruction, which clobbers the VCC register.

Cc: mesa-stable
Signed-off-by: Timur Kristóf <timur.kristof at gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10346>
(cherry picked from commit 74c467d9886cbb588d9e8f0ac23775d4b9d780bd)

---

 .pick_status.json                              | 2 +-
 src/amd/compiler/aco_instruction_selection.cpp | 2 ++
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/.pick_status.json b/.pick_status.json
index 31a85afeef1..78bdf90cf2d 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -148,7 +148,7 @@
         "description": "aco: Mark VCC clobbered for iadd8 and iadd16 reductions on GFX6-7.",
         "nominated": true,
         "nomination_type": 0,
-        "resolution": 0,
+        "resolution": 1,
         "master_sha": null,
         "because_sha": null
     },
diff --git a/src/amd/compiler/aco_instruction_selection.cpp b/src/amd/compiler/aco_instruction_selection.cpp
index 04f5a41c5d6..acb65c8c1c2 100644
--- a/src/amd/compiler/aco_instruction_selection.cpp
+++ b/src/amd/compiler/aco_instruction_selection.cpp
@@ -7828,6 +7828,8 @@ Temp emit_reduction_instr(isel_context *ctx, aco_opcode aco_op, ReduceOp op,
    bool clobber_vcc = false;
    if ((op == iadd32 || op == imul64) && ctx->program->chip_class < GFX9)
       clobber_vcc = true;
+   if ((op == iadd8 || op == iadd16) && ctx->program->chip_class < GFX8)
+      clobber_vcc = true;
    if (op == iadd64 || op == umin64 || op == umax64 || op == imin64 || op == imax64)
       clobber_vcc = true;