summaryrefslogtreecommitdiff
path: root/src/compiler/nir/nir_opt_shrink_load.c
diff options
context:
space:
mode:
authorSamuel Pitoiset <samuel.pitoiset@gmail.com>2018-01-29 17:19:18 +0100
committerSamuel Pitoiset <samuel.pitoiset@gmail.com>2018-02-06 23:08:44 +0100
commit3488a3f0332ba1d77e2c4609a26206e77922e81b (patch)
tree732400c3f6e597d9c078eb48414914a8501c04f1 /src/compiler/nir/nir_opt_shrink_load.c
parente68562b94b79f5ce17f5804f8d7308faf57bd3f0 (diff)
radv: run nir_opt_shrink_load
LLVM can't shrink loads. Polaris10: Totals from affected shaders: SGPRS: 62528 -> 59955 (-4.11 %) VGPRS: 44708 -> 44616 (-0.21 %) Spilled SGPRs: 16 -> 8 (-50.00 %) Code Size: 1355504 -> 1355172 (-0.02 %) bytes Max Waves: 11710 -> 11670 (-0.34 %) Vega10: Totals from affected shaders: SGPRS: 51448 -> 50371 (-2.09 %) VGPRS: 39140 -> 39048 (-0.24 %) Spilled SGPRs: 16 -> 16 (0.00 %) Code Size: 1307188 -> 1304296 (-0.22 %) bytes Max Waves: 11312 -> 11292 (-0.18 %) This reduces SGPRs spilling in MadMax, and it also reduces number of SGPRs in DOW3 and F12017. The number of waves slightly decreases in F1 but I don't see any performance changes after benchmarking it. Talos and Serious Sam are not affected because they don't use any push constants. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Diffstat (limited to 'src/compiler/nir/nir_opt_shrink_load.c')
-rw-r--r--src/compiler/nir/nir_opt_shrink_load.c67
1 files changed, 67 insertions, 0 deletions
diff --git a/src/compiler/nir/nir_opt_shrink_load.c b/src/compiler/nir/nir_opt_shrink_load.c
new file mode 100644
index 00000000000..f97b7f9b67f
--- /dev/null
+++ b/src/compiler/nir/nir_opt_shrink_load.c
@@ -0,0 +1,67 @@
+/*
+ * Copyright © 2018 Valve Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "nir.h"
+
+static bool
+opt_shrink_load(nir_intrinsic_instr *instr)
+{
+ bool progress = false;
+
+ if (instr->intrinsic == nir_intrinsic_load_push_constant) {
+ unsigned mask = nir_ssa_def_components_read(&instr->dest.ssa);
+
+ if (instr->num_components > util_last_bit(mask)) {
+ instr->num_components = util_last_bit(mask);
+ instr->dest.ssa.num_components = instr->num_components;
+ progress = true;
+ }
+ }
+
+ return progress;
+}
+
+bool
+nir_opt_shrink_load(nir_shader *shader)
+{
+ bool progress = false;
+
+ nir_foreach_function(function, shader) {
+ if (!function->impl)
+ continue;
+
+ nir_foreach_block(block, function->impl) {
+ nir_foreach_instr(instr, block) {
+ if (instr->type != nir_instr_type_intrinsic)
+ continue;
+
+ progress |= opt_shrink_load(nir_instr_as_intrinsic(instr));
+ }
+ }
+
+ nir_metadata_preserve(function->impl, nir_metadata_block_index |
+ nir_metadata_dominance);
+ }
+
+ return progress;
+}