encounter
/
dawn-cmake
mirror of https://github.com/encounter/dawn-cmake.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
dawn-cmake/test/vk-gl-cts/graphicsfuzz/stable-mergesort-clamped-co.../1.wgsl.expected.spvasm

705 lines
26 KiB
Plaintext
Raw Normal View History

Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
; SPIR-V
; Version: 1.3
; Generator: Google Tint Compiler; 0
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
; Bound: 460
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
; Schema: 0
OpCapability Shader
%174 = OpExtInstImport "GLSL.std.450"
OpMemoryModel Logical GLSL450
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
OpEntryPoint Fragment %main "main" %gl_FragCoord_param_1 %x_GLF_color_1_1
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpExecutionMode %main OriginUpperLeft
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
OpName %gl_FragCoord_param_1 "gl_FragCoord_param_1"
OpName %x_GLF_color_1_1 "x_GLF_color_1_1"
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpName %data "data"
OpName %temp "temp"
OpName %gl_FragCoord "gl_FragCoord"
OpName %buf0 "buf0"
OpMemberName %buf0 0 "injectionSwitch"
OpName %x_34 "x_34"
OpName %x_GLF_color "x_GLF_color"
OpName %merge_i1_i1_i1_ "merge_i1_i1_i1_"
OpName %from "from"
OpName %mid "mid"
OpName %to "to"
OpName %k "k"
OpName %i "i"
OpName %j "j"
OpName %i_1 "i_1"
OpName %func_i1_i1_ "func_i1_i1_"
OpName %m "m"
OpName %high "high"
OpName %x "x"
OpName %x_335 "x_335"
OpName %x_336 "x_336"
OpName %mergeSort_ "mergeSort_"
OpName %low "low"
OpName %high_1 "high_1"
OpName %m_1 "m_1"
OpName %i_2 "i_2"
OpName %from_1 "from_1"
OpName %mid_1 "mid_1"
OpName %to_1 "to_1"
OpName %param "param"
OpName %param_1 "param_1"
OpName %param_2 "param_2"
OpName %param_3 "param_3"
OpName %param_4 "param_4"
OpName %main_1 "main_1"
OpName %i_3 "i_3"
OpName %j_1 "j_1"
OpName %grey "grey"
OpName %main_out "main_out"
OpMemberName %main_out 0 "x_GLF_color_1"
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
OpName %main_inner "main_inner"
OpName %gl_FragCoord_param "gl_FragCoord_param"
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpName %main "main"
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
OpDecorate %gl_FragCoord_param_1 BuiltIn FragCoord
OpDecorate %x_GLF_color_1_1 Location 0
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpDecorate %_arr_int_uint_10 ArrayStride 4
OpDecorate %buf0 Block
OpMemberDecorate %buf0 0 Offset 0
OpDecorate %x_34 NonWritable
OpDecorate %x_34 DescriptorSet 0
OpDecorate %x_34 Binding 0
OpMemberDecorate %main_out 0 Offset 0
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
%float = OpTypeFloat 32
%v4float = OpTypeVector %float 4
%_ptr_Input_v4float = OpTypePointer Input %v4float
%gl_FragCoord_param_1 = OpVariable %_ptr_Input_v4float Input
%_ptr_Output_v4float = OpTypePointer Output %v4float
%7 = OpConstantNull %v4float
%x_GLF_color_1_1 = OpVariable %_ptr_Output_v4float Output %7
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
%int = OpTypeInt 32 1
%uint = OpTypeInt 32 0
%uint_10 = OpConstant %uint 10
%_arr_int_uint_10 = OpTypeArray %int %uint_10
%_ptr_Private__arr_int_uint_10 = OpTypePointer Private %_arr_int_uint_10
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
%14 = OpConstantNull %_arr_int_uint_10
%data = OpVariable %_ptr_Private__arr_int_uint_10 Private %14
%temp = OpVariable %_ptr_Private__arr_int_uint_10 Private %14
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
%_ptr_Private_v4float = OpTypePointer Private %v4float
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
%gl_FragCoord = OpVariable %_ptr_Private_v4float Private %7
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
%v2float = OpTypeVector %float 2
%buf0 = OpTypeStruct %v2float
%_ptr_Uniform_buf0 = OpTypePointer Uniform %buf0
%x_34 = OpVariable %_ptr_Uniform_buf0 Uniform
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
%x_GLF_color = OpVariable %_ptr_Private_v4float Private %7
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
%void = OpTypeVoid
%_ptr_Function_int = OpTypePointer Function %int
%23 = OpTypeFunction %void %_ptr_Function_int %_ptr_Function_int %_ptr_Function_int
%32 = OpConstantNull %int
%int_1 = OpConstant %int 1
%bool = OpTypeBool
%_ptr_Private_int = OpTypePointer Private %int
%int_10 = OpConstant %int 10
%132 = OpTypeFunction %int %_ptr_Function_int %_ptr_Function_int
%uint_0 = OpConstant %uint 0
%_ptr_Private_float = OpTypePointer Private %float
%float_0 = OpConstant %float 0
%false = OpConstantFalse %bool
%int_0 = OpConstant %int 0
%int_4 = OpConstant %int 4
%int_2 = OpConstant %int 2
%180 = OpTypeFunction %void
%int_9 = OpConstant %int 9
%_ptr_Function_float = OpTypePointer Function %float
%251 = OpConstantNull %float
%_ptr_Uniform_float = OpTypePointer Uniform %float
%int_n5 = OpConstant %int -5
%int_n4 = OpConstant %int -4
%int_n3 = OpConstant %int -3
%int_n2 = OpConstant %int -2
%int_n1 = OpConstant %int -1
%int_3 = OpConstant %int 3
%uint_1 = OpConstant %uint 1
%int_30 = OpConstant %int 30
%float_0_5 = OpConstant %float 0.5
%float_10 = OpConstant %float 10
%int_60 = OpConstant %int 60
%int_90 = OpConstant %int 90
%int_120 = OpConstant %int 120
%int_150 = OpConstant %int 150
%int_180 = OpConstant %int 180
%int_5 = OpConstant %int 5
%int_210 = OpConstant %int 210
%int_6 = OpConstant %int 6
%int_240 = OpConstant %int 240
%int_7 = OpConstant %int 7
%int_270 = OpConstant %int 270
%int_8 = OpConstant %int 8
%v3float = OpTypeVector %float 3
%float_1 = OpConstant %float 1
%main_out = OpTypeStruct %v4float
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
%447 = OpTypeFunction %main_out %v4float
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
%merge_i1_i1_i1_ = OpFunction %void None %23
%from = OpFunctionParameter %_ptr_Function_int
%mid = OpFunctionParameter %_ptr_Function_int
%to = OpFunctionParameter %_ptr_Function_int
%30 = OpLabel
%k = OpVariable %_ptr_Function_int Function %32
%i = OpVariable %_ptr_Function_int Function %32
%j = OpVariable %_ptr_Function_int Function %32
%i_1 = OpVariable %_ptr_Function_int Function %32
%37 = OpLoad %int %from
OpStore %k %37
%39 = OpLoad %int %from
OpStore %i %39
%41 = OpLoad %int %mid
%43 = OpIAdd %int %41 %int_1
OpStore %j %43
OpBranch %44
%44 = OpLabel
OpLoopMerge %45 %46 None
OpBranch %47
%47 = OpLabel
%48 = OpLoad %int %i
%50 = OpLoad %int %mid
%51 = OpLoad %int %j
%53 = OpLoad %int %to
%54 = OpSLessThanEqual %bool %48 %50
OpSelectionMerge %56 None
OpBranchConditional %54 %57 %56
%57 = OpLabel
%58 = OpSLessThanEqual %bool %51 %53
OpBranch %56
%56 = OpLabel
%59 = OpPhi %bool %54 %47 %58 %57
OpSelectionMerge %60 None
OpBranchConditional %59 %61 %62
%61 = OpLabel
OpBranch %60
%62 = OpLabel
OpBranch %45
%60 = OpLabel
%63 = OpLoad %int %i
%65 = OpAccessChain %_ptr_Private_int %data %63
%66 = OpLoad %int %65
%67 = OpLoad %int %j
%68 = OpAccessChain %_ptr_Private_int %data %67
%69 = OpLoad %int %68
%70 = OpSLessThan %bool %66 %69
OpSelectionMerge %71 None
OpBranchConditional %70 %72 %73
%72 = OpLabel
%74 = OpLoad %int %k
%75 = OpIAdd %int %74 %int_1
OpStore %k %75
%76 = OpLoad %int %i
%77 = OpIAdd %int %76 %int_1
OpStore %i %77
%78 = OpAccessChain %_ptr_Private_int %data %76
%79 = OpLoad %int %78
%80 = OpAccessChain %_ptr_Private_int %temp %74
OpStore %80 %79
OpBranch %71
%73 = OpLabel
%81 = OpLoad %int %k
%82 = OpIAdd %int %81 %int_1
OpStore %k %82
%83 = OpLoad %int %j
%84 = OpIAdd %int %83 %int_1
OpStore %j %84
%85 = OpAccessChain %_ptr_Private_int %data %83
%86 = OpLoad %int %85
%87 = OpAccessChain %_ptr_Private_int %temp %81
OpStore %87 %86
OpBranch %71
%71 = OpLabel
OpBranch %46
%46 = OpLabel
OpBranch %44
%45 = OpLabel
OpBranch %88
%88 = OpLabel
OpLoopMerge %89 %90 None
OpBranch %91
%91 = OpLabel
%92 = OpLoad %int %i
%93 = OpLoad %int %i
%95 = OpLoad %int %mid
%97 = OpSLessThan %bool %92 %int_10
OpSelectionMerge %98 None
OpBranchConditional %97 %99 %98
%99 = OpLabel
%100 = OpSLessThanEqual %bool %93 %95
OpBranch %98
%98 = OpLabel
%101 = OpPhi %bool %97 %91 %100 %99
OpSelectionMerge %102 None
OpBranchConditional %101 %103 %104
%103 = OpLabel
OpBranch %102
%104 = OpLabel
OpBranch %89
%102 = OpLabel
%105 = OpLoad %int %k
%106 = OpIAdd %int %105 %int_1
OpStore %k %106
%107 = OpLoad %int %i
%108 = OpIAdd %int %107 %int_1
OpStore %i %108
%109 = OpAccessChain %_ptr_Private_int %data %107
%110 = OpLoad %int %109
%111 = OpAccessChain %_ptr_Private_int %temp %105
OpStore %111 %110
OpBranch %90
%90 = OpLabel
OpBranch %88
%89 = OpLabel
%113 = OpLoad %int %from
OpStore %i_1 %113
OpBranch %114
%114 = OpLabel
OpLoopMerge %115 %116 None
OpBranch %117
%117 = OpLabel
%118 = OpLoad %int %i_1
%120 = OpLoad %int %to
%121 = OpSLessThanEqual %bool %118 %120
OpSelectionMerge %122 None
OpBranchConditional %121 %123 %124
%123 = OpLabel
OpBranch %122
%124 = OpLabel
OpBranch %115
%122 = OpLabel
%125 = OpLoad %int %i_1
%126 = OpLoad %int %i_1
%127 = OpAccessChain %_ptr_Private_int %temp %126
%128 = OpLoad %int %127
%129 = OpAccessChain %_ptr_Private_int %data %125
OpStore %129 %128
OpBranch %116
%116 = OpLabel
%130 = OpLoad %int %i_1
%131 = OpIAdd %int %130 %int_1
OpStore %i_1 %131
OpBranch %114
%115 = OpLabel
OpReturn
OpFunctionEnd
%func_i1_i1_ = OpFunction %int None %132
%m = OpFunctionParameter %_ptr_Function_int
%high = OpFunctionParameter %_ptr_Function_int
%136 = OpLabel
%x = OpVariable %_ptr_Function_int Function %32
%x_335 = OpVariable %_ptr_Function_int Function %32
%x_336 = OpVariable %_ptr_Function_int Function %32
%142 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_0
%143 = OpLoad %float %142
%145 = OpFOrdGreaterThanEqual %bool %143 %float_0
OpSelectionMerge %146 None
OpBranchConditional %145 %147 %148
%147 = OpLabel
OpSelectionMerge %150 None
OpBranchConditional %false %151 %152
%151 = OpLabel
%154 = OpLoad %int %high
%155 = OpBitcast %uint %int_0
%157 = OpShiftLeftLogical %int %154 %155
OpStore %x_336 %157
OpBranch %150
%152 = OpLabel
OpStore %x_336 %int_4
OpBranch %150
%150 = OpLabel
%159 = OpLoad %int %x_336
%160 = OpBitcast %uint %159
%161 = OpShiftLeftLogical %int %int_1 %160
OpStore %x_335 %161
OpBranch %146
%148 = OpLabel
OpStore %x_335 %int_1
OpBranch %146
%146 = OpLabel
%162 = OpLoad %int %x_335
OpStore %x %162
%163 = OpLoad %int %x
%164 = OpBitcast %uint %int_4
%165 = OpShiftRightArithmetic %int %163 %164
OpStore %x %165
%167 = OpLoad %int %m
%169 = OpLoad %int %m
%171 = OpLoad %int %m
%172 = OpLoad %int %x
%176 = OpIMul %int %int_2 %167
%177 = OpIMul %int %int_2 %169
%178 = OpIMul %int %int_2 %171
%179 = OpSDiv %int %178 %172
%173 = OpExtInst %int %174 SClamp %176 %177 %179
OpReturnValue %173
OpFunctionEnd
%mergeSort_ = OpFunction %void None %180
%182 = OpLabel
%low = OpVariable %_ptr_Function_int Function %32
%high_1 = OpVariable %_ptr_Function_int Function %32
%m_1 = OpVariable %_ptr_Function_int Function %32
%i_2 = OpVariable %_ptr_Function_int Function %32
%from_1 = OpVariable %_ptr_Function_int Function %32
%mid_1 = OpVariable %_ptr_Function_int Function %32
%to_1 = OpVariable %_ptr_Function_int Function %32
%param = OpVariable %_ptr_Function_int Function %32
%param_1 = OpVariable %_ptr_Function_int Function %32
%param_2 = OpVariable %_ptr_Function_int Function %32
%param_3 = OpVariable %_ptr_Function_int Function %32
%param_4 = OpVariable %_ptr_Function_int Function %32
OpStore %low %int_0
OpStore %high_1 %int_9
OpStore %m_1 %int_1
OpBranch %196
%196 = OpLabel
OpLoopMerge %197 %198 None
OpBranch %199
%199 = OpLabel
%200 = OpLoad %int %m_1
%201 = OpLoad %int %high_1
%202 = OpSLessThanEqual %bool %200 %201
OpSelectionMerge %203 None
OpBranchConditional %202 %204 %205
%204 = OpLabel
OpBranch %203
%205 = OpLabel
OpBranch %197
%203 = OpLabel
%206 = OpLoad %int %low
OpStore %i_2 %206
OpBranch %207
%207 = OpLabel
OpLoopMerge %208 %209 None
OpBranch %210
%210 = OpLabel
%211 = OpLoad %int %i_2
%212 = OpLoad %int %high_1
%213 = OpSLessThan %bool %211 %212
OpSelectionMerge %214 None
OpBranchConditional %213 %215 %216
%215 = OpLabel
OpBranch %214
%216 = OpLabel
OpBranch %208
%214 = OpLabel
%217 = OpLoad %int %i_2
OpStore %from_1 %217
%218 = OpLoad %int %i_2
%219 = OpLoad %int %m_1
%220 = OpIAdd %int %218 %219
%221 = OpISub %int %220 %int_1
OpStore %mid_1 %221
%222 = OpLoad %int %i_2
%223 = OpLoad %int %m_1
%224 = OpLoad %int %high_1
%226 = OpIMul %int %int_2 %223
%227 = OpIAdd %int %222 %226
%228 = OpISub %int %227 %int_1
%225 = OpExtInst %int %174 SMin %228 %224
OpStore %to_1 %225
%229 = OpLoad %int %from_1
OpStore %param %229
%230 = OpLoad %int %mid_1
OpStore %param_1 %230
%231 = OpLoad %int %to_1
OpStore %param_2 %231
%232 = OpFunctionCall %void %merge_i1_i1_i1_ %param %param_1 %param_2
OpBranch %209
%209 = OpLabel
%236 = OpLoad %int %m_1
OpStore %param_3 %236
%237 = OpLoad %int %high_1
OpStore %param_4 %237
%238 = OpFunctionCall %int %func_i1_i1_ %param_3 %param_4
%241 = OpLoad %int %i_2
%242 = OpIAdd %int %241 %238
OpStore %i_2 %242
OpBranch %207
%208 = OpLabel
OpBranch %198
%198 = OpLabel
%243 = OpLoad %int %m_1
%244 = OpIMul %int %int_2 %243
OpStore %m_1 %244
OpBranch %196
%197 = OpLabel
OpReturn
OpFunctionEnd
%main_1 = OpFunction %void None %180
%246 = OpLabel
%i_3 = OpVariable %_ptr_Function_int Function %32
%j_1 = OpVariable %_ptr_Function_int Function %32
%grey = OpVariable %_ptr_Function_float Function %251
%253 = OpAccessChain %_ptr_Uniform_float %x_34 %uint_0 %uint_0
%254 = OpLoad %float %253
%255 = OpConvertFToS %int %254
OpStore %i_3 %255
OpBranch %256
%256 = OpLabel
OpLoopMerge %257 %258 None
OpBranch %259
%259 = OpLabel
%260 = OpLoad %int %i_3
OpSelectionMerge %261 None
OpSwitch %260 %262 9 %263 8 %264 7 %265 6 %266 5 %267 4 %268 3 %269 2 %270 1 %271 0 %272
%263 = OpLabel
%273 = OpLoad %int %i_3
%274 = OpAccessChain %_ptr_Private_int %data %273
OpStore %274 %int_n5
OpBranch %261
%264 = OpLabel
%276 = OpLoad %int %i_3
%277 = OpAccessChain %_ptr_Private_int %data %276
OpStore %277 %int_n4
OpBranch %261
%265 = OpLabel
%279 = OpLoad %int %i_3
%280 = OpAccessChain %_ptr_Private_int %data %279
OpStore %280 %int_n3
OpBranch %261
%266 = OpLabel
%282 = OpLoad %int %i_3
%283 = OpAccessChain %_ptr_Private_int %data %282
OpStore %283 %int_n2
OpBranch %261
%267 = OpLabel
%285 = OpLoad %int %i_3
%286 = OpAccessChain %_ptr_Private_int %data %285
OpStore %286 %int_n1
OpBranch %261
%268 = OpLabel
%288 = OpLoad %int %i_3
%289 = OpAccessChain %_ptr_Private_int %data %288
OpStore %289 %int_0
OpBranch %261
%269 = OpLabel
%290 = OpLoad %int %i_3
%291 = OpAccessChain %_ptr_Private_int %data %290
OpStore %291 %int_1
OpBranch %261
%270 = OpLabel
%292 = OpLoad %int %i_3
%293 = OpAccessChain %_ptr_Private_int %data %292
OpStore %293 %int_2
OpBranch %261
%271 = OpLabel
%294 = OpLoad %int %i_3
%295 = OpAccessChain %_ptr_Private_int %data %294
OpStore %295 %int_3
OpBranch %261
%272 = OpLabel
%297 = OpLoad %int %i_3
%298 = OpAccessChain %_ptr_Private_int %data %297
OpStore %298 %int_4
OpBranch %261
%262 = OpLabel
OpBranch %261
%261 = OpLabel
%299 = OpLoad %int %i_3
%300 = OpIAdd %int %299 %int_1
OpStore %i_3 %300
OpBranch %258
%258 = OpLabel
%301 = OpLoad %int %i_3
%302 = OpSLessThan %bool %301 %int_10
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
OpBranchConditional %302 %256 %257
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
%257 = OpLabel
OpStore %j_1 %int_0
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
OpBranch %303
%303 = OpLabel
OpLoopMerge %304 %305 None
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpBranch %306
%306 = OpLabel
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
%307 = OpLoad %int %j_1
%308 = OpSLessThan %bool %307 %int_10
OpSelectionMerge %309 None
OpBranchConditional %308 %310 %311
%310 = OpLabel
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpBranch %309
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
%311 = OpLabel
OpBranch %304
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
%309 = OpLabel
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
%312 = OpLoad %int %j_1
%313 = OpLoad %int %j_1
%314 = OpAccessChain %_ptr_Private_int %data %313
%315 = OpLoad %int %314
%316 = OpAccessChain %_ptr_Private_int %temp %312
OpStore %316 %315
OpBranch %305
%305 = OpLabel
%317 = OpLoad %int %j_1
%318 = OpIAdd %int %317 %int_1
OpStore %j_1 %318
OpBranch %303
%304 = OpLabel
%319 = OpFunctionCall %void %mergeSort_
%321 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%322 = OpLoad %float %321
%323 = OpConvertFToS %int %322
%325 = OpSLessThan %bool %323 %int_30
OpSelectionMerge %326 None
OpBranchConditional %325 %327 %328
%327 = OpLabel
%329 = OpAccessChain %_ptr_Private_int %data %int_0
%330 = OpLoad %int %329
%332 = OpConvertSToF %float %330
%334 = OpFDiv %float %332 %float_10
%335 = OpFAdd %float %float_0_5 %334
OpStore %grey %335
OpBranch %326
%328 = OpLabel
%336 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%337 = OpLoad %float %336
%338 = OpConvertFToS %int %337
%340 = OpSLessThan %bool %338 %int_60
OpSelectionMerge %341 None
OpBranchConditional %340 %342 %343
%342 = OpLabel
%344 = OpAccessChain %_ptr_Private_int %data %int_1
%345 = OpLoad %int %344
%346 = OpConvertSToF %float %345
%347 = OpFDiv %float %346 %float_10
%348 = OpFAdd %float %float_0_5 %347
OpStore %grey %348
OpBranch %341
%343 = OpLabel
%349 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%350 = OpLoad %float %349
%351 = OpConvertFToS %int %350
%353 = OpSLessThan %bool %351 %int_90
OpSelectionMerge %354 None
OpBranchConditional %353 %355 %356
%355 = OpLabel
%357 = OpAccessChain %_ptr_Private_int %data %int_2
%358 = OpLoad %int %357
%359 = OpConvertSToF %float %358
%360 = OpFDiv %float %359 %float_10
%361 = OpFAdd %float %float_0_5 %360
OpStore %grey %361
OpBranch %354
%356 = OpLabel
%362 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%363 = OpLoad %float %362
%364 = OpConvertFToS %int %363
%366 = OpSLessThan %bool %364 %int_120
OpSelectionMerge %367 None
OpBranchConditional %366 %368 %369
%368 = OpLabel
%370 = OpAccessChain %_ptr_Private_int %data %int_3
%371 = OpLoad %int %370
%372 = OpConvertSToF %float %371
%373 = OpFDiv %float %372 %float_10
%374 = OpFAdd %float %float_0_5 %373
OpStore %grey %374
OpBranch %367
%369 = OpLabel
%375 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%376 = OpLoad %float %375
%377 = OpConvertFToS %int %376
%379 = OpSLessThan %bool %377 %int_150
OpSelectionMerge %380 None
OpBranchConditional %379 %381 %382
%381 = OpLabel
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpKill
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
%382 = OpLabel
%383 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%384 = OpLoad %float %383
%385 = OpConvertFToS %int %384
%387 = OpSLessThan %bool %385 %int_180
OpSelectionMerge %388 None
OpBranchConditional %387 %389 %390
%389 = OpLabel
%392 = OpAccessChain %_ptr_Private_int %data %int_5
%393 = OpLoad %int %392
%394 = OpConvertSToF %float %393
%395 = OpFDiv %float %394 %float_10
%396 = OpFAdd %float %float_0_5 %395
OpStore %grey %396
OpBranch %388
%390 = OpLabel
%397 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%398 = OpLoad %float %397
%399 = OpConvertFToS %int %398
%401 = OpSLessThan %bool %399 %int_210
OpSelectionMerge %402 None
OpBranchConditional %401 %403 %404
%403 = OpLabel
%406 = OpAccessChain %_ptr_Private_int %data %int_6
%407 = OpLoad %int %406
%408 = OpConvertSToF %float %407
%409 = OpFDiv %float %408 %float_10
%410 = OpFAdd %float %float_0_5 %409
OpStore %grey %410
OpBranch %402
%404 = OpLabel
%411 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%412 = OpLoad %float %411
%413 = OpConvertFToS %int %412
%415 = OpSLessThan %bool %413 %int_240
OpSelectionMerge %416 None
OpBranchConditional %415 %417 %418
%417 = OpLabel
%420 = OpAccessChain %_ptr_Private_int %data %int_7
%421 = OpLoad %int %420
%422 = OpConvertSToF %float %421
%423 = OpFDiv %float %422 %float_10
%424 = OpFAdd %float %float_0_5 %423
OpStore %grey %424
OpBranch %416
%418 = OpLabel
%425 = OpAccessChain %_ptr_Private_float %gl_FragCoord %uint_1
%426 = OpLoad %float %425
%427 = OpConvertFToS %int %426
%429 = OpSLessThan %bool %427 %int_270
OpSelectionMerge %430 None
OpBranchConditional %429 %431 %432
%431 = OpLabel
%434 = OpAccessChain %_ptr_Private_int %data %int_8
%435 = OpLoad %int %434
%436 = OpConvertSToF %float %435
%437 = OpFDiv %float %436 %float_10
%438 = OpFAdd %float %float_0_5 %437
OpStore %grey %438
OpBranch %430
%432 = OpLabel
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpKill
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
%430 = OpLabel
OpBranch %416
%416 = OpLabel
OpBranch %402
%402 = OpLabel
OpBranch %388
%388 = OpLabel
OpBranch %380
%380 = OpLabel
OpBranch %367
%367 = OpLabel
OpBranch %354
%354 = OpLabel
OpBranch %341
%341 = OpLabel
OpBranch %326
%326 = OpLabel
%439 = OpLoad %float %grey
%441 = OpCompositeConstruct %v3float %439 %439 %439
%442 = OpCompositeExtract %float %441 0
%443 = OpCompositeExtract %float %441 1
%444 = OpCompositeExtract %float %441 2
%446 = OpCompositeConstruct %v4float %442 %443 %444 %float_1
OpStore %x_GLF_color %446
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpReturn
OpFunctionEnd
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
%main_inner = OpFunction %main_out None %447
%gl_FragCoord_param = OpFunctionParameter %v4float
spir-writer: handle break continuing block The continuing block can exit the loop in very constrained ways: When a break statement is placed such that it would exit from a loop’s § 7.3.8 Continuing Statement, then: - The break statement must appear as either: - The only statement in the if clause of an if statement that has: - no else clause or an empty else clause - no elseif clauses - The only statement in the else clause of an if statement that has an empty if clause and no elseif clauses. - That if statement must appear last in the continuing clause. By design, this allows a lossless round-trip from SPIR-V to WGSL and back to SPIR-V. But that requires this special case construct in WGSL to be translated to an OpBranchConditional with one target being the loop's megre block (which is where 'break' branches to), and the other targets the loop header (which is the loop backedge). That OpBranchConditional takes the place of the normal case of an unconditional backedge. Avoids errors like this: continue construct with the continue target X is not post dominated by the back-edge block Y Fixed: 1034 Change-Id: If472a179380b8d77af746a3cd8e279c8a5e56b37 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/59800 Auto-Submit: David Neto <dneto@google.com> Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: David Neto <dneto@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-27 15:12:27 +00:00
%451 = OpLabel
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
OpStore %gl_FragCoord %gl_FragCoord_param
%452 = OpFunctionCall %void %main_1
%453 = OpLoad %v4float %x_GLF_color
%454 = OpCompositeConstruct %main_out %453
OpReturnValue %454
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpFunctionEnd
%main = OpFunction %void None %180
spirv: Use generic transform to process shader IO The refactored CanonicalizeEntryPointIO transform makes it much easier to handle SPIR-V style IO as well, and doing this removes a lot of duplicated code. Remove all of the SPIR-V transform code for shader IO and vertex point size. Bug: tint:920 Change-Id: Id1b97517619b4d2fd09b45d5aee848259f3dfa77 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60840 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: James Price <jrprice@google.com> Auto-Submit: James Price <jrprice@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-08-05 17:34:19 +00:00
%456 = OpLabel
%458 = OpLoad %v4float %gl_FragCoord_param_1
%457 = OpFunctionCall %main_out %main_inner %458
%459 = OpCompositeExtract %v4float %457 0
OpStore %x_GLF_color_1_1 %459
Add tests derived from VK-GL-CTS This adds SPIR-V assembly and WGSL tests derived from VK-GL-CTS commit 571256871c2e2f03995373e1e4a02958d8cd8cf5. The following procedure was followed: - Those .amber files in VK-GL-CTS wholly owned by Google were identified - All GLSL and SPIR-V shaders were extracted from the Amber files and converted into SPIR-V binaries - The compact-ids pass of spirv-opt was applied to each binary - Duplicate binaries were removed - spirv-opt -O was used to obtain an optimized version of each remaining binary, with duplicates discarded - Binaries that failed validation using spirv-val with target environment SPIR-V 1.3 were discarded - Those binaries that tint could not successfully convert into WGSL were put aside for further investigation - SPIR-V assembly versions of the remaining binaries are included in this CL - test-runner with -generate-expected and -generate-skip was used to generate expected .spvasm, .msl, .hlsl and .wgsl outputs for these SPIR-V assembly tests - Each successfully-generated .expected.wgsl is included in this CL again, as a WGLSL test - test-runner with -generate-expected and -generate-skip was used again, to generate expected outputs for these WGSL tests Change-Id: Ibe9baf2729cf97e0b633db9a426f53362a5de540 Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58842 Kokoro: Kokoro <noreply+kokoro@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> Reviewed-by: Ben Clayton <bclayton@google.com>
2021-07-23 13:10:12 +00:00
OpReturn
OpFunctionEnd