fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-25 00:00:11 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

nir/algebraic: ad-hoc constant-fold ALU instructions

Browse source 
Slight differences due to different optimization order.
Totals from 135 (0.17% of 79839) affected shaders: (Navi48)
Instrs: 287852 -> 287527 (-0.11%); split: -0.15%, +0.03%
CodeSize: 1522972 -> 1521764 (-0.08%); split: -0.12%, +0.04%
Latency: 1806803 -> 1825754 (+1.05%); split: -0.08%, +1.12%
InvThroughput: 242693 -> 244703 (+0.83%); split: -0.02%, +0.84%
VClause: 4092 -> 4084 (-0.20%)
SClause: 7462 -> 7478 (+0.21%)
Copies: 20509 -> 20401 (-0.53%); split: -0.74%, +0.21%
Branches: 6395 -> 6386 (-0.14%)
PreSGPRs: 7334 -> 7337 (+0.04%); split: -0.03%, +0.07%
PreVGPRs: 6375 -> 6382 (+0.11%)
VALU: 151787 -> 151595 (-0.13%); split: -0.15%, +0.02%
SALU: 52967 -> 52910 (-0.11%); split: -0.23%, +0.12%
VMEM: 6704 -> 6696 (-0.12%)
SMEM: 12099 -> 12129 (+0.25%)

Tested on a small collection of 2518 shaders from Dredge with callgrind using RADV:
baseline:
  nir_opt_algebraic was called 12917 times from radv_optimize_nir()
  nir_opt_cse was called 15204 times from radv_optimize_nir()
  relative time spent in radv_optimize_nir(): 31.48%
  total instruction fetch cost: 28,642,638,021

with nir/algebraic: ad-hoc constant-fold ALU instructions
  nir_opt_algebraic was called 12797 times from radv_optimize_nir()
  nir_opt_cse was called 12963 times from radv_optimize_nir()
  relative time spent in radv_optimize_nir(): 30.63%
  total instruction fetch cost: 28,284,386,123

=> ~1.27% improvement in total compile times

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37195>
This commit is contained in:
Daniel Schürmann 2025-09-05 10:57:45 +02:00 • committed by Marge Bot
parent 10be538851
commit b3615e5d6f
2 changed files with 20 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

19
src/compiler/nir/nir_search.c
View file

@ -494,15 +494,24 @@ construct_value(nir_builder *build,
state, instr);
}
nir_builder_instr_insert(build, &alu->instr);
assert(alu->def.index ==
/* Immediately try to constant-fold the expression, in order to allow
* for more expressions to be matched within a single pass.
*/
nir_def *def = &alu->def;
nir_def *const_expr = nir_try_constant_fold_alu(build, alu);
if (const_expr) {
nir_instr_free(&alu->instr);
def = const_expr;
} else {
nir_builder_instr_insert(build, &alu->instr);
}
assert(def->index ==
util_dynarray_num_elements(state->states, uint16_t));
util_dynarray_append_typed(state->states, uint16_t, 0);
nir_algebraic_automaton(&alu->instr, state->states, state->pass_op_table);
nir_algebraic_automaton(def->parent_instr, state->states, state->pass_op_table);
nir_alu_src val;
val.src = nir_src_for_ssa(&alu->def);
val.src = nir_src_for_ssa(def);
if (expr->swizzle < 0)
memcpy(val.swizzle, identity_swizzle, sizeof(val.swizzle));
else

12
src/intel/ci/traces-iris.yml
View file

@ -36,20 +36,20 @@ traces:
checksum: f53ac20e17da91c0359c31f2fa3f401e
0ad/0ad-v2.trace:
gl-intel-apl:
checksum: b25b9dfce05def046f6277d05b3cba98
checksum: 6c9f66db38f03847574c543de48fa203
gl-intel-glk:
checksum: b25b9dfce05def046f6277d05b3cba98
checksum: 6c9f66db38f03847574c543de48fa203
gl-intel-amly:
checksum: b25b9dfce05def046f6277d05b3cba98
checksum: 6c9f66db38f03847574c543de48fa203
gl-intel-kbl:
checksum: b25b9dfce05def046f6277d05b3cba98
checksum: 6c9f66db38f03847574c543de48fa203
gl-intel-whl:
checksum: e67b7a93bac02e41de0326419ee17a3e
gl-intel-cml:
checksum: b25b9dfce05def046f6277d05b3cba98
checksum: 6c9f66db38f03847574c543de48fa203
gl-intel-adl:
label: [no-perf]
checksum: b25b9dfce05def046f6277d05b3cba98
checksum: 6c9f66db38f03847574c543de48fa203
pathfinder/demo-v2.trace:
gl-intel-apl:
checksum: d9b33f0a2efe17c21b7933242afd9ec7