AuroraMiddleware/skia2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
95c1a08b24
skia2/tests/sksl/shared/golden/SwizzleByIndex.asm.frag

128 lines
4.1 KiB
GLSL
Raw Normal View History

Detect cases in Metal where out params are swizzled. We will need to emit a helper function to work around this case, as GLSL supports swizzled out params, but Metal does not. In this CL, we do not yet synthesize the helper function, but we annotate the code with a comment indicating affected calls. (Of course, this will be replaced with a helper function in a followup CL) Even detecting a swizzle is actually an interesting problem, because index expressions are sometimes actually swizzles, depending on the type of the base expression. Also, the index or swizzle might be nested in several other valid assignable expressions. Change-Id: I8c74f9a7daec08eff1f32387f8b6b96851c1bd6e Bug: skia:10855 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341057 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-04 20:36:49 +00:00
OpCapability Shader
%1 = OpExtInstImport "GLSL.std.450"
OpMemoryModel Logical GLSL450
OpEntryPoint Fragment %main "main" %sk_FragColor %sk_Clockwise
OpExecutionMode %main OriginUpperLeft
OpName %sk_FragColor "sk_FragColor"
OpName %sk_Clockwise "sk_Clockwise"
OpName %main "main"
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
OpName %_0_i "_0_i"
OpName %_1_v "_1_v"
OpName %_2_x "_2_x"
OpName %_3_y "_3_y"
OpName %_4_z "_4_z"
OpName %_5_w "_5_w"
OpName %_6_v "_6_v"
OpName %_7_x "_7_x"
OpName %_8_y "_8_y"
OpName %_9_z "_9_z"
OpName %_10_w "_10_w"
Detect cases in Metal where out params are swizzled. We will need to emit a helper function to work around this case, as GLSL supports swizzled out params, but Metal does not. In this CL, we do not yet synthesize the helper function, but we annotate the code with a comment indicating affected calls. (Of course, this will be replaced with a helper function in a followup CL) Even detecting a swizzle is actually an interesting problem, because index expressions are sometimes actually swizzles, depending on the type of the base expression. Also, the index or swizzle might be nested in several other valid assignable expressions. Change-Id: I8c74f9a7daec08eff1f32387f8b6b96851c1bd6e Bug: skia:10855 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341057 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-04 20:36:49 +00:00
OpDecorate %sk_FragColor RelaxedPrecision
OpDecorate %sk_FragColor Location 0
OpDecorate %sk_FragColor Index 0
OpDecorate %sk_Clockwise RelaxedPrecision
OpDecorate %sk_Clockwise BuiltIn FrontFacing
Revert "Declare all inlined variables at the topmost scope possible." This reverts commit e8e4aca955bc87732ada3dd42ec60f6b1a37443e. Reason for revert: can break ES2 for-loop rules Original change's description: > Declare all inlined variables at the topmost scope possible. > > By itself, this is uninteresting and even perhaps slightly > counterproductive (as it separates vardecl from its initializer, > increasing LOC). However, this enables a followup CL > (http://review.skia.org/344665) which allows single-return functions to > be inlined without the creation of a temporary variable at all. This > applies to the majority of fragment processors in a typical Ganesh > hierarchy. This change will greatly reduce the number of inliner-created > temporary copies when compiling a typical tree of FPs. > > Change-Id: I03423a13cf35050637dabace4a32973a08a4ed0a > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344764 > Reviewed-by: Brian Osman <brianosman@google.com> > Commit-Queue: John Stiles <johnstiles@google.com> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com Change-Id: Ica01d6906bcb9cef1f49d22dda714fc9cbfa3885 No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/345121 Reviewed-by: John Stiles <johnstiles@google.com> Commit-Queue: John Stiles <johnstiles@google.com>
2020-12-16 18:25:14 +00:00
OpDecorate %27 RelaxedPrecision
OpDecorate %32 RelaxedPrecision
OpDecorate %37 RelaxedPrecision
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
OpDecorate %42 RelaxedPrecision
OpDecorate %46 RelaxedPrecision
OpDecorate %47 RelaxedPrecision
OpDecorate %48 RelaxedPrecision
OpDecorate %49 RelaxedPrecision
Revert "Declare all inlined variables at the topmost scope possible." This reverts commit e8e4aca955bc87732ada3dd42ec60f6b1a37443e. Reason for revert: can break ES2 for-loop rules Original change's description: > Declare all inlined variables at the topmost scope possible. > > By itself, this is uninteresting and even perhaps slightly > counterproductive (as it separates vardecl from its initializer, > increasing LOC). However, this enables a followup CL > (http://review.skia.org/344665) which allows single-return functions to > be inlined without the creation of a temporary variable at all. This > applies to the majority of fragment processors in a typical Ganesh > hierarchy. This change will greatly reduce the number of inliner-created > temporary copies when compiling a typical tree of FPs. > > Change-Id: I03423a13cf35050637dabace4a32973a08a4ed0a > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344764 > Reviewed-by: Brian Osman <brianosman@google.com> > Commit-Queue: John Stiles <johnstiles@google.com> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com Change-Id: Ica01d6906bcb9cef1f49d22dda714fc9cbfa3885 No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/345121 Reviewed-by: John Stiles <johnstiles@google.com> Commit-Queue: John Stiles <johnstiles@google.com>
2020-12-16 18:25:14 +00:00
OpDecorate %55 RelaxedPrecision
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
OpDecorate %58 RelaxedPrecision
Revert "Declare all inlined variables at the topmost scope possible." This reverts commit e8e4aca955bc87732ada3dd42ec60f6b1a37443e. Reason for revert: can break ES2 for-loop rules Original change's description: > Declare all inlined variables at the topmost scope possible. > > By itself, this is uninteresting and even perhaps slightly > counterproductive (as it separates vardecl from its initializer, > increasing LOC). However, this enables a followup CL > (http://review.skia.org/344665) which allows single-return functions to > be inlined without the creation of a temporary variable at all. This > applies to the majority of fragment processors in a typical Ganesh > hierarchy. This change will greatly reduce the number of inliner-created > temporary copies when compiling a typical tree of FPs. > > Change-Id: I03423a13cf35050637dabace4a32973a08a4ed0a > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344764 > Reviewed-by: Brian Osman <brianosman@google.com> > Commit-Queue: John Stiles <johnstiles@google.com> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com Change-Id: Ica01d6906bcb9cef1f49d22dda714fc9cbfa3885 No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/345121 Reviewed-by: John Stiles <johnstiles@google.com> Commit-Queue: John Stiles <johnstiles@google.com>
2020-12-16 18:25:14 +00:00
OpDecorate %61 RelaxedPrecision
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
OpDecorate %64 RelaxedPrecision
OpDecorate %66 RelaxedPrecision
OpDecorate %67 RelaxedPrecision
OpDecorate %68 RelaxedPrecision
OpDecorate %69 RelaxedPrecision
Detect cases in Metal where out params are swizzled. We will need to emit a helper function to work around this case, as GLSL supports swizzled out params, but Metal does not. In this CL, we do not yet synthesize the helper function, but we annotate the code with a comment indicating affected calls. (Of course, this will be replaced with a helper function in a followup CL) Even detecting a swizzle is actually an interesting problem, because index expressions are sometimes actually swizzles, depending on the type of the base expression. Also, the index or swizzle might be nested in several other valid assignable expressions. Change-Id: I8c74f9a7daec08eff1f32387f8b6b96851c1bd6e Bug: skia:10855 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341057 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-04 20:36:49 +00:00
%float = OpTypeFloat 32
%v4float = OpTypeVector %float 4
%_ptr_Output_v4float = OpTypePointer Output %v4float
%sk_FragColor = OpVariable %_ptr_Output_v4float Output
%bool = OpTypeBool
%_ptr_Input_bool = OpTypePointer Input %bool
%sk_Clockwise = OpVariable %_ptr_Input_bool Input
%void = OpTypeVoid
%11 = OpTypeFunction %void
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
%int = OpTypeInt 32 1
%v4int = OpTypeVector %int 4
%_ptr_Function_v4int = OpTypePointer Function %v4int
Revert "Declare all inlined variables at the topmost scope possible." This reverts commit e8e4aca955bc87732ada3dd42ec60f6b1a37443e. Reason for revert: can break ES2 for-loop rules Original change's description: > Declare all inlined variables at the topmost scope possible. > > By itself, this is uninteresting and even perhaps slightly > counterproductive (as it separates vardecl from its initializer, > increasing LOC). However, this enables a followup CL > (http://review.skia.org/344665) which allows single-return functions to > be inlined without the creation of a temporary variable at all. This > applies to the majority of fragment processors in a typical Ganesh > hierarchy. This change will greatly reduce the number of inliner-created > temporary copies when compiling a typical tree of FPs. > > Change-Id: I03423a13cf35050637dabace4a32973a08a4ed0a > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344764 > Reviewed-by: Brian Osman <brianosman@google.com> > Commit-Queue: John Stiles <johnstiles@google.com> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com Change-Id: Ica01d6906bcb9cef1f49d22dda714fc9cbfa3885 No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/345121 Reviewed-by: John Stiles <johnstiles@google.com> Commit-Queue: John Stiles <johnstiles@google.com>
2020-12-16 18:25:14 +00:00
%float_1 = OpConstant %float 1
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
%_ptr_Function_v4float = OpTypePointer Function %v4float
Detect cases in Metal where out params are swizzled. We will need to emit a helper function to work around this case, as GLSL supports swizzled out params, but Metal does not. In this CL, we do not yet synthesize the helper function, but we annotate the code with a comment indicating affected calls. (Of course, this will be replaced with a helper function in a followup CL) Even detecting a swizzle is actually an interesting problem, because index expressions are sometimes actually swizzles, depending on the type of the base expression. Also, the index or swizzle might be nested in several other valid assignable expressions. Change-Id: I8c74f9a7daec08eff1f32387f8b6b96851c1bd6e Bug: skia:10855 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341057 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-04 20:36:49 +00:00
%_ptr_Function_float = OpTypePointer Function %float
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
%float_2 = OpConstant %float 2
%71 = OpConstantComposite %v4float %float_2 %float_2 %float_2 %float_2
Detect cases in Metal where out params are swizzled. We will need to emit a helper function to work around this case, as GLSL supports swizzled out params, but Metal does not. In this CL, we do not yet synthesize the helper function, but we annotate the code with a comment indicating affected calls. (Of course, this will be replaced with a helper function in a followup CL) Even detecting a swizzle is actually an interesting problem, because index expressions are sometimes actually swizzles, depending on the type of the base expression. Also, the index or swizzle might be nested in several other valid assignable expressions. Change-Id: I8c74f9a7daec08eff1f32387f8b6b96851c1bd6e Bug: skia:10855 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341057 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-04 20:36:49 +00:00
%main = OpFunction %void None %11
%12 = OpLabel
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
%_0_i = OpVariable %_ptr_Function_v4int Function
%_1_v = OpVariable %_ptr_Function_v4float Function
%_2_x = OpVariable %_ptr_Function_float Function
%_3_y = OpVariable %_ptr_Function_float Function
%_4_z = OpVariable %_ptr_Function_float Function
%_5_w = OpVariable %_ptr_Function_float Function
%_6_v = OpVariable %_ptr_Function_v4float Function
%_7_x = OpVariable %_ptr_Function_float Function
%_8_y = OpVariable %_ptr_Function_float Function
%_9_z = OpVariable %_ptr_Function_float Function
%_10_w = OpVariable %_ptr_Function_float Function
Revert "Declare all inlined variables at the topmost scope possible." This reverts commit e8e4aca955bc87732ada3dd42ec60f6b1a37443e. Reason for revert: can break ES2 for-loop rules Original change's description: > Declare all inlined variables at the topmost scope possible. > > By itself, this is uninteresting and even perhaps slightly > counterproductive (as it separates vardecl from its initializer, > increasing LOC). However, this enables a followup CL > (http://review.skia.org/344665) which allows single-return functions to > be inlined without the creation of a temporary variable at all. This > applies to the majority of fragment processors in a typical Ganesh > hierarchy. This change will greatly reduce the number of inliner-created > temporary copies when compiling a typical tree of FPs. > > Change-Id: I03423a13cf35050637dabace4a32973a08a4ed0a > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344764 > Reviewed-by: Brian Osman <brianosman@google.com> > Commit-Queue: John Stiles <johnstiles@google.com> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com Change-Id: Ica01d6906bcb9cef1f49d22dda714fc9cbfa3885 No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/345121 Reviewed-by: John Stiles <johnstiles@google.com> Commit-Queue: John Stiles <johnstiles@google.com>
2020-12-16 18:25:14 +00:00
%18 = OpExtInst %float %1 Sqrt %float_1
%17 = OpConvertFToS %int %18
%20 = OpCompositeConstruct %v4int %17 %17 %17 %17
OpStore %_0_i %20
%23 = OpExtInst %float %1 Sqrt %float_1
%24 = OpCompositeConstruct %v4float %23 %23 %23 %23
OpStore %_1_v %24
%27 = OpLoad %v4float %_1_v
%28 = OpLoad %v4int %_0_i
%29 = OpCompositeExtract %int %28 0
%30 = OpVectorExtractDynamic %float %27 %29
OpStore %_2_x %30
%32 = OpLoad %v4float %_1_v
%33 = OpLoad %v4int %_0_i
%34 = OpCompositeExtract %int %33 1
%35 = OpVectorExtractDynamic %float %32 %34
OpStore %_3_y %35
%37 = OpLoad %v4float %_1_v
%38 = OpLoad %v4int %_0_i
%39 = OpCompositeExtract %int %38 2
%40 = OpVectorExtractDynamic %float %37 %39
OpStore %_4_z %40
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
%42 = OpLoad %v4float %_1_v
%43 = OpLoad %v4int %_0_i
%44 = OpCompositeExtract %int %43 3
%45 = OpVectorExtractDynamic %float %42 %44
OpStore %_5_w %45
%46 = OpLoad %float %_2_x
%47 = OpLoad %float %_3_y
%48 = OpLoad %float %_4_z
%49 = OpLoad %float %_5_w
%50 = OpCompositeConstruct %v4float %46 %47 %48 %49
OpStore %sk_FragColor %50
Revert "Declare all inlined variables at the topmost scope possible." This reverts commit e8e4aca955bc87732ada3dd42ec60f6b1a37443e. Reason for revert: can break ES2 for-loop rules Original change's description: > Declare all inlined variables at the topmost scope possible. > > By itself, this is uninteresting and even perhaps slightly > counterproductive (as it separates vardecl from its initializer, > increasing LOC). However, this enables a followup CL > (http://review.skia.org/344665) which allows single-return functions to > be inlined without the creation of a temporary variable at all. This > applies to the majority of fragment processors in a typical Ganesh > hierarchy. This change will greatly reduce the number of inliner-created > temporary copies when compiling a typical tree of FPs. > > Change-Id: I03423a13cf35050637dabace4a32973a08a4ed0a > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344764 > Reviewed-by: Brian Osman <brianosman@google.com> > Commit-Queue: John Stiles <johnstiles@google.com> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com Change-Id: Ica01d6906bcb9cef1f49d22dda714fc9cbfa3885 No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/345121 Reviewed-by: John Stiles <johnstiles@google.com> Commit-Queue: John Stiles <johnstiles@google.com>
2020-12-16 18:25:14 +00:00
%52 = OpExtInst %float %1 Sqrt %float_1
%53 = OpCompositeConstruct %v4float %52 %52 %52 %52
OpStore %_6_v %53
%55 = OpLoad %v4float %_6_v
%56 = OpCompositeExtract %float %55 0
OpStore %_7_x %56
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
%58 = OpLoad %v4float %_6_v
Revert "Declare all inlined variables at the topmost scope possible." This reverts commit e8e4aca955bc87732ada3dd42ec60f6b1a37443e. Reason for revert: can break ES2 for-loop rules Original change's description: > Declare all inlined variables at the topmost scope possible. > > By itself, this is uninteresting and even perhaps slightly > counterproductive (as it separates vardecl from its initializer, > increasing LOC). However, this enables a followup CL > (http://review.skia.org/344665) which allows single-return functions to > be inlined without the creation of a temporary variable at all. This > applies to the majority of fragment processors in a typical Ganesh > hierarchy. This change will greatly reduce the number of inliner-created > temporary copies when compiling a typical tree of FPs. > > Change-Id: I03423a13cf35050637dabace4a32973a08a4ed0a > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/344764 > Reviewed-by: Brian Osman <brianosman@google.com> > Commit-Queue: John Stiles <johnstiles@google.com> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com Change-Id: Ica01d6906bcb9cef1f49d22dda714fc9cbfa3885 No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/345121 Reviewed-by: John Stiles <johnstiles@google.com> Commit-Queue: John Stiles <johnstiles@google.com>
2020-12-16 18:25:14 +00:00
%59 = OpCompositeExtract %float %58 1
OpStore %_8_y %59
%61 = OpLoad %v4float %_6_v
%62 = OpCompositeExtract %float %61 2
OpStore %_9_z %62
Replace array indexing on vector types with swizzles. Our optimizer ignores index expressions, but has a few simplifications that it can perform on swizzles. (Added extra code to SwizzleByIndex which demonstrates this.) Change-Id: If3c85a0456d98749008d796e422944b602ee6933 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341460 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-07 20:25:31 +00:00
%64 = OpLoad %v4float %_6_v
%65 = OpCompositeExtract %float %64 3
OpStore %_10_w %65
%66 = OpLoad %float %_7_x
%67 = OpLoad %float %_8_y
%68 = OpLoad %float %_9_z
%69 = OpLoad %float %_10_w
%70 = OpCompositeConstruct %v4float %66 %67 %68 %69
OpStore %sk_FragColor %70
OpStore %sk_FragColor %71
Detect cases in Metal where out params are swizzled. We will need to emit a helper function to work around this case, as GLSL supports swizzled out params, but Metal does not. In this CL, we do not yet synthesize the helper function, but we annotate the code with a comment indicating affected calls. (Of course, this will be replaced with a helper function in a followup CL) Even detecting a swizzle is actually an interesting problem, because index expressions are sometimes actually swizzles, depending on the type of the base expression. Also, the index or swizzle might be nested in several other valid assignable expressions. Change-Id: I8c74f9a7daec08eff1f32387f8b6b96851c1bd6e Bug: skia:10855 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341057 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
2020-12-04 20:36:49 +00:00
OpReturn
OpFunctionEnd
Reference in New Issue Copy Permalink