diff options
| author | Marek Olšák <marek.olsak@amd.com> | 2016-10-02 21:23:26 +0200 |
|---|---|---|
| committer | Marek Olšák <marek.olsak@amd.com> | 2016-10-04 16:12:11 +0200 |
| commit | 046c199c3af3d329aefd87498c77c014a6bb83a6 (patch) | |
| tree | fe332543705a14057d18068ef679b39cf5c5b04c /src/gallium/drivers/radeonsi/si_shader.c | |
| parent | bcc55e1f3288f200e996c381736e86201fb3531a (diff) | |
| download | external_mesa3d-046c199c3af3d329aefd87498c77c014a6bb83a6.zip external_mesa3d-046c199c3af3d329aefd87498c77c014a6bb83a6.tar.gz external_mesa3d-046c199c3af3d329aefd87498c77c014a6bb83a6.tar.bz2 | |
radeonsi: don't call build_gep0 in si_llvm_emit_ddxy on VI
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Diffstat (limited to 'src/gallium/drivers/radeonsi/si_shader.c')
| -rw-r--r-- | src/gallium/drivers/radeonsi/si_shader.c | 14 |
1 files changed, 9 insertions, 5 deletions
diff --git a/src/gallium/drivers/radeonsi/si_shader.c b/src/gallium/drivers/radeonsi/si_shader.c index 67ab16b..414810e 100644 --- a/src/gallium/drivers/radeonsi/si_shader.c +++ b/src/gallium/drivers/radeonsi/si_shader.c @@ -5014,9 +5014,10 @@ static void si_llvm_emit_ddxy( unsigned c; int idx; unsigned mask; + bool has_ds_bpermute = HAVE_LLVM >= 0x0309 && + ctx->screen->b.chip_class >= VI; thread_id = get_thread_id(ctx);; - store_ptr = build_gep0(ctx, ctx->lds, thread_id); if (opcode == TGSI_OPCODE_DDX_FINE) mask = TID_MASK_LEFT; @@ -5027,13 +5028,17 @@ static void si_llvm_emit_ddxy( tl_tid = LLVMBuildAnd(gallivm->builder, thread_id, lp_build_const_int32(gallivm, mask), ""); - load_ptr0 = build_gep0(ctx, ctx->lds, tl_tid); /* for DDX we want to next X pixel, DDY next Y pixel. */ idx = (opcode == TGSI_OPCODE_DDX || opcode == TGSI_OPCODE_DDX_FINE) ? 1 : 2; trbl_tid = LLVMBuildAdd(gallivm->builder, tl_tid, lp_build_const_int32(gallivm, idx), ""); - load_ptr1 = build_gep0(ctx, ctx->lds, trbl_tid); + + if (!has_ds_bpermute) { + store_ptr = build_gep0(ctx, ctx->lds, thread_id); + load_ptr0 = build_gep0(ctx, ctx->lds, tl_tid); + load_ptr1 = build_gep0(ctx, ctx->lds, trbl_tid); + } for (c = 0; c < 4; ++c) { unsigned i; @@ -5054,8 +5059,7 @@ static void si_llvm_emit_ddxy( lp_build_emit_fetch(bld_base, inst, 0, c), ctx->i32, ""); - if ((HAVE_LLVM >= 0x0309) && ctx->screen->b.family >= CHIP_TONGA) { - + if (has_ds_bpermute) { args[0] = LLVMBuildMul(gallivm->builder, tl_tid, lp_build_const_int32(gallivm, 4), ""); args[1] = val; |