KonstantinSeurer
/
mesa
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

radv, aco: Round texture array layer in NIR. 

Foz-DB Navi21:
Totals from 9100 (6.75% of 134913) affected shaders:
VGPRs: 609912 -> 610104 (+0.03%); split: -0.01%, +0.05%
SpillSGPRs: 1459 -> 1489 (+2.06%)
CodeSize: 66705920 -> 66620288 (-0.13%); split: -0.13%, +0.00%
MaxWaves: 148546 -> 148518 (-0.02%); split: +0.00%, -0.02%
Instrs: 12278485 -> 12255821 (-0.18%); split: -0.19%, +0.00%
Latency: 277414916 -> 277261192 (-0.06%); split: -0.08%, +0.02%
InvThroughput: 48431180 -> 48394637 (-0.08%); split: -0.11%, +0.03%
VClause: 250866 -> 251062 (+0.08%); split: -0.04%, +0.11%
SClause: 498377 -> 498173 (-0.04%); split: -0.08%, +0.04%
Copies: 652835 -> 655371 (+0.39%); split: -0.09%, +0.48%
Branches: 284367 -> 284371 (+0.00%); split: -0.00%, +0.00%
PreSGPRs: 498580 -> 498477 (-0.02%)
PreVGPRs: 558436 -> 558709 (+0.05%); split: -0.01%, +0.06%

Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16871>
This commit is contained in:
Georg Lehmann 2022-06-04 21:26:45 +02:00 committed by Marge Bot
parent 1b68d3e43a
commit 5ebb014bdf
2 changed files with 4 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
src/amd/compiler/aco_instruction_selection.cpp
View File

@ -9169,13 +9169,9 @@ prepare_cube_coords(isel_context* ctx, std::vector<Temp>& coords, Temp* ddx, Tem
aco_opcode madmk =
ctx->program->gfx_level >= GFX10_3 ? aco_opcode::v_fmamk_f32 : aco_opcode::v_madmk_f32;
if (is_array) {
coords[3] = bld.vop1(aco_opcode::v_rndne_f32, bld.def(v1), coords[3]);
/* see comment in ac_prepare_cube_coords() */
if (ctx->options->gfx_level <= GFX8)
coords[3] = bld.vop2(aco_opcode::v_max_f32, bld.def(v1), Operand::zero(), coords[3]);
}
/* see comment in ac_prepare_cube_coords() */
if (is_array && ctx->options->gfx_level <= GFX8)
coords[3] = bld.vop2(aco_opcode::v_max_f32, bld.def(v1), Operand::zero(), coords[3]);
ma = bld.vop3(aco_opcode::v_cubema_f32, bld.def(v1), coords[0], coords[1], coords[2]);
@ -9425,18 +9421,6 @@ visit_tex(isel_context* ctx, nir_tex_instr* instr)
has_derivs = true;
}
if (instr->coord_components > 1 && instr->sampler_dim == GLSL_SAMPLER_DIM_1D &&
instr->is_array && instr->op != nir_texop_txf)
coords[1] = bld.vop1(aco_opcode::v_rndne_f32, bld.def(v1), coords[1]);
if (instr->coord_components > 2 &&
(instr->sampler_dim == GLSL_SAMPLER_DIM_2D || instr->sampler_dim == GLSL_SAMPLER_DIM_MS ||
instr->sampler_dim == GLSL_SAMPLER_DIM_SUBPASS ||
instr->sampler_dim == GLSL_SAMPLER_DIM_SUBPASS_MS) &&
instr->is_array && instr->op != nir_texop_txf && instr->op != nir_texop_fragment_fetch_amd &&
instr->op != nir_texop_fragment_mask_fetch_amd)
coords[2] = bld.vop1(aco_opcode::v_rndne_f32, bld.def(v1), coords[2]);
if (ctx->options->gfx_level == GFX9 && instr->sampler_dim == GLSL_SAMPLER_DIM_1D &&
instr->op != nir_texop_lod && instr->coord_components) {
assert(coords.size() > 0 && coords.size() < 3);

1
src/amd/vulkan/radv_shader.c
View File

@ -827,6 +827,7 @@ radv_shader_spirv_to_nir(struct radv_device *device, const struct radv_pipeline_
.lower_to_fragment_fetch_amd = true,
.lower_lod_zero_width = true,
.lower_invalid_implicit_lod = true,
.lower_array_layer_round_even = true,
};
NIR_PASS(_, nir, nir_lower_tex, &tex_options);