fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-04-22 16:10:36 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 3

aco: fix integer tg4 workaround with unnormalized coordinates

Browse source 
Same as LLVM from 2abf62d348.

fossil-db (GFX8):
Totals from 15 (0.01% of 147787) affected shaders:
VGPRs: 744 -> 748 (+0.54%)
CodeSize: 100472 -> 100732 (+0.26%)
Instrs: 19995 -> 20059 (+0.32%)
Latency: 1001530 -> 1001859 (+0.03%)
InvThroughput: 378508 -> 378747 (+0.06%)
SClause: 676 -> 675 (-0.15%)
Copies: 1655 -> 1654 (-0.06%)
PreSGPRs: 735 -> 742 (+0.95%)

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/10053>
This commit is contained in:
Rhys Perry 2021-03-31 17:46:52 +01:00 • committed by Marge Bot
parent 5a70c4d4a0
commit 835c5b7ebf
1 changed files with 18 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
src/amd/compiler/aco_instruction_selection.cpp
View file

@ -9036,6 +9036,24 @@ void visit_tex(isel_context *ctx, nir_tex_instr *instr)
half_texel[i] = bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), Operand(0xbf000000/*-0.5*/), half_texel[i]);
}
if (instr->sampler_dim == GLSL_SAMPLER_DIM_2D && !instr->is_array) {
/* In vulkan, whether the sampler uses unnormalized
* coordinates or not is a dynamic property of the
* sampler. Hence, to figure out whether or not we
* need to divide by the texture size, we need to test
* the sampler at runtime. This tests the bit set by
* radv_init_sampler().
*/
unsigned bit_idx = ffs(S_008F30_FORCE_UNNORMALIZED(1)) - 1;
Temp not_needed = bld.sopc(aco_opcode::s_bitcmp0_b32, bld.def(s1, scc), sampler, Operand(bit_idx));
not_needed = bool_to_vector_condition(ctx, not_needed);
half_texel[0] = bld.vop2(aco_opcode::v_cndmask_b32, bld.def(v1),
Operand(0xbf000000/*-0.5*/), half_texel[0], not_needed);
half_texel[1] = bld.vop2(aco_opcode::v_cndmask_b32, bld.def(v1),
Operand(0xbf000000/*-0.5*/), half_texel[1], not_needed);
}
Temp new_coords[2] = {
bld.vop2(aco_opcode::v_add_f32, bld.def(v1), coords[0], half_texel[0]),
bld.vop2(aco_opcode::v_add_f32, bld.def(v1), coords[1], half_texel[1])