Module: Mesa Branch: main Commit: 3040aa2e264bd1351a14eb385486e0677e5e0a20 URL: http://cgit.freedesktop.org/mesa/mesa/commit/?id=3040aa2e264bd1351a14eb385486e0677e5e0a20
Author: Marek Olšák <[email protected]> Date: Sat Jul 8 22:25:19 2023 -0400 ac/llvm: don't convert undef to 0 because nir_opt_undef does it now TOTALS FROM AFFECTED SHADERS (29663/58918) Code Size: 39163724 -> 37842360 (-3.37 %) bytes Max Waves: 394813 -> 396334 (0.39 %) Outputs: 84616 -> 84616 (0.00 %) Patch Outputs: 0 -> 0 (0.00 %) Reviewed-by: Alyssa Rosenzweig <[email protected]> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24059> --- src/amd/llvm/ac_nir_to_llvm.c | 37 +++++---------------------- src/amd/llvm/ac_shader_abi.h | 3 --- src/gallium/drivers/radeonsi/si_shader_llvm.c | 1 - 3 files changed, 6 insertions(+), 35 deletions(-) diff --git a/src/amd/llvm/ac_nir_to_llvm.c b/src/amd/llvm/ac_nir_to_llvm.c index d857d4a85d3..44028d37e71 100644 --- a/src/amd/llvm/ac_nir_to_llvm.c +++ b/src/amd/llvm/ac_nir_to_llvm.c @@ -4129,44 +4129,19 @@ static void phi_post_pass(struct ac_nir_context *ctx) } } -static bool is_def_used_in_an_export(const nir_def *def) -{ - nir_foreach_use (use_src, def) { - if (use_src->parent_instr->type == nir_instr_type_intrinsic) { - nir_intrinsic_instr *instr = nir_instr_as_intrinsic(use_src->parent_instr); - if (instr->intrinsic == nir_intrinsic_store_deref) - return true; - } else if (use_src->parent_instr->type == nir_instr_type_alu) { - nir_alu_instr *instr = nir_instr_as_alu(use_src->parent_instr); - if (instr->op == nir_op_vec4 && is_def_used_in_an_export(&instr->def)) { - return true; - } - } - } - return false; -} - static void visit_ssa_undef(struct ac_nir_context *ctx, const nir_undef_instr *instr) { unsigned num_components = instr->def.num_components; LLVMTypeRef type = LLVMIntTypeInContext(ctx->ac.context, instr->def.bit_size); - if (!ctx->abi->convert_undef_to_zero || is_def_used_in_an_export(&instr->def)) { - LLVMValueRef undef; + LLVMValueRef undef; - if (num_components == 1) - undef = LLVMGetUndef(type); - else { - undef = LLVMGetUndef(LLVMVectorType(type, num_components)); - } - ctx->ssa_defs[instr->def.index] = undef; - } else { - LLVMValueRef zero = LLVMConstInt(type, 0, false); - if (num_components > 1) { - zero = ac_build_gather_values_extended(&ctx->ac, &zero, num_components, 0, false); - } - ctx->ssa_defs[instr->def.index] = zero; + if (num_components == 1) + undef = LLVMGetUndef(type); + else { + undef = LLVMGetUndef(LLVMVectorType(type, num_components)); } + ctx->ssa_defs[instr->def.index] = undef; } static bool visit_jump(struct ac_llvm_context *ctx, const nir_jump_instr *instr) diff --git a/src/amd/llvm/ac_shader_abi.h b/src/amd/llvm/ac_shader_abi.h index d505482b77f..4dee4ad2813 100644 --- a/src/amd/llvm/ac_shader_abi.h +++ b/src/amd/llvm/ac_shader_abi.h @@ -79,9 +79,6 @@ struct ac_shader_abi { /* Check for Inf interpolation coeff */ bool kill_ps_if_inf_interp; - /* Whether undef values must be converted to zero */ - bool convert_undef_to_zero; - /* Clamp div by 0 (so it won't produce NaN) */ bool clamp_div_by_zero; diff --git a/src/gallium/drivers/radeonsi/si_shader_llvm.c b/src/gallium/drivers/radeonsi/si_shader_llvm.c index 7e496a81605..83a106cfb9f 100644 --- a/src/gallium/drivers/radeonsi/si_shader_llvm.c +++ b/src/gallium/drivers/radeonsi/si_shader_llvm.c @@ -741,7 +741,6 @@ static bool si_llvm_translate_nir(struct si_shader_context *ctx, struct si_shade ctx->abi.clamp_shadow_reference = true; ctx->abi.robust_buffer_access = true; - ctx->abi.convert_undef_to_zero = true; ctx->abi.load_grid_size_from_user_sgpr = true; ctx->abi.clamp_div_by_zero = ctx->screen->options.clamp_div_by_zero || info->options & SI_PROFILE_CLAMP_DIV_BY_ZERO;
