summaryrefslogtreecommitdiffstats
path: root/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp
diff options
context:
space:
mode:
authorameerj <52414509+ameerj@users.noreply.github.com>2021-08-31 02:39:21 +0200
committerameerj <52414509+ameerj@users.noreply.github.com>2021-08-31 19:40:12 +0200
commit95213270efbc378561f18cc6eeb4428c2905b154 (patch)
tree1c55902fdb3ca12e9264311307057d29ec4e403a /src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp
parentMerge pull request #6897 from FernandoS27/pineapple-does-not-belong-in-pizza (diff)
downloadyuzu-95213270efbc378561f18cc6eeb4428c2905b154.tar
yuzu-95213270efbc378561f18cc6eeb4428c2905b154.tar.gz
yuzu-95213270efbc378561f18cc6eeb4428c2905b154.tar.bz2
yuzu-95213270efbc378561f18cc6eeb4428c2905b154.tar.lz
yuzu-95213270efbc378561f18cc6eeb4428c2905b154.tar.xz
yuzu-95213270efbc378561f18cc6eeb4428c2905b154.tar.zst
yuzu-95213270efbc378561f18cc6eeb4428c2905b154.zip
Diffstat (limited to 'src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp')
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp21
1 files changed, 11 insertions, 10 deletions
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp
index 78b1e1ba7..e7e7b4da1 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp
@@ -7,8 +7,13 @@
namespace Shader::Backend::SPIRV {
namespace {
+Id GetThreadId(EmitContext& ctx) {
+ return ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id);
+}
+
Id WarpExtract(EmitContext& ctx, Id value) {
- const Id local_index{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+ const Id thread_id{GetThreadId(ctx)};
+ const Id local_index{ctx.OpShiftRightArithmetic(ctx.U32[1], thread_id, ctx.Const(5U))};
return ctx.OpVectorExtractDynamic(ctx.U32[1], value, local_index);
}
@@ -51,11 +56,7 @@ Id SelectValue(EmitContext& ctx, Id in_range, Id value, Id src_thread_id) {
} // Anonymous namespace
Id EmitLaneId(EmitContext& ctx) {
- const Id id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
- if (!ctx.profile.warp_size_potentially_larger_than_guest) {
- return id;
- }
- return ctx.OpBitwiseAnd(ctx.U32[1], id, ctx.Const(31U));
+ return GetThreadId(ctx);
}
Id EmitVoteAll(EmitContext& ctx, Id pred) {
@@ -123,7 +124,7 @@ Id EmitSubgroupGeMask(EmitContext& ctx) {
Id EmitShuffleIndex(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
Id segmentation_mask) {
const Id not_seg_mask{ctx.OpNot(ctx.U32[1], segmentation_mask)};
- const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+ const Id thread_id{GetThreadId(ctx)};
const Id min_thread_id{ComputeMinThreadId(ctx, thread_id, segmentation_mask)};
const Id max_thread_id{ComputeMaxThreadId(ctx, min_thread_id, clamp, not_seg_mask)};
@@ -137,7 +138,7 @@ Id EmitShuffleIndex(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id cla
Id EmitShuffleUp(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
Id segmentation_mask) {
- const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+ const Id thread_id{GetThreadId(ctx)};
const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
const Id src_thread_id{ctx.OpISub(ctx.U32[1], thread_id, index)};
const Id in_range{ctx.OpSGreaterThanEqual(ctx.U1, src_thread_id, max_thread_id)};
@@ -148,7 +149,7 @@ Id EmitShuffleUp(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
Id EmitShuffleDown(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
Id segmentation_mask) {
- const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+ const Id thread_id{GetThreadId(ctx)};
const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
const Id src_thread_id{ctx.OpIAdd(ctx.U32[1], thread_id, index)};
const Id in_range{ctx.OpSLessThanEqual(ctx.U1, src_thread_id, max_thread_id)};
@@ -159,7 +160,7 @@ Id EmitShuffleDown(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clam
Id EmitShuffleButterfly(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
Id segmentation_mask) {
- const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+ const Id thread_id{GetThreadId(ctx)};
const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
const Id src_thread_id{ctx.OpBitwiseXor(ctx.U32[1], thread_id, index)};
const Id in_range{ctx.OpSLessThanEqual(ctx.U1, src_thread_id, max_thread_id)};