From: Rhys Perry <pendingchaos02@gmail.com>
Date: Fri, 22 Nov 2019 17:50:29 +0000 (+0000)
Subject: aco: improve clamp optimization
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=fa8357eb7008115413a9f3219e98a0f718687223;p=mesa.git

aco: improve clamp optimization

Not sure why it checked the use count, it doesn't apply the constants.

pipeline-db (Navi):
Totals from affected shaders:
SGPRS: 269409 -> 269745 (0.12 %)
VGPRS: 238120 -> 238132 (0.01 %)
Spilled SGPRs: 305 -> 305 (0.00 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Code Size: 22908584 -> 22904672 (-0.02 %) bytes
Max Waves: 20217 -> 20217 (0.00 %)
Instructions: 4275312 -> 4263869 (-0.27 %)

pipeline-db (Vega):
Totals from affected shaders:
SGPRS: 155409 -> 155233 (-0.11 %)
VGPRS: 153072 -> 153072 (0.00 %)
Spilled SGPRs: 269 -> 269 (0.00 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Code Size: 14650824 -> 14650396 (-0.00 %) bytes
Max Waves: 9609 -> 9609 (0.00 %)
Instructions: 2762802 -> 2755517 (-0.26 %)

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/2883>
---

diff --git a/src/amd/compiler/aco_optimizer.cpp b/src/amd/compiler/aco_optimizer.cpp
index 44ce5289ff5..96fa17e62d8 100644
--- a/src/amd/compiler/aco_optimizer.cpp
+++ b/src/amd/compiler/aco_optimizer.cpp
@@ -1794,6 +1794,9 @@ bool get_minmax_info(aco_opcode op, aco_opcode *min, aco_opcode *max, aco_opcode
 bool combine_clamp(opt_ctx& ctx, aco_ptr<Instruction>& instr,
                    aco_opcode min, aco_opcode max, aco_opcode med)
 {
+   /* TODO: GLSL's clamp(x, minVal, maxVal) and SPIR-V's
+    * FClamp(x, minVal, maxVal)/NClamp(x, minVal, maxVal) are undefined if
+    * minVal > maxVal, which means we can always select it to a v_med3_f32 */
    aco_opcode other_op;
    if (instr->opcode == min)
       other_op = max;
@@ -1818,8 +1821,7 @@ bool combine_clamp(opt_ctx& ctx, aco_ptr<Instruction>& instr,
             uint32_t val;
             if (operands[i].isConstant()) {
                val = operands[i].constantValue();
-            } else if (operands[i].isTemp() && ctx.uses[operands[i].tempId()] == 1 &&
-                       ctx.info[operands[i].tempId()].is_constant_or_literal()) {
+            } else if (operands[i].isTemp() && ctx.info[operands[i].tempId()].is_constant_or_literal()) {
                val = ctx.info[operands[i].tempId()].val;
             } else {
                continue;