[Wasm RyuJIT] Codegen for Const Vector Create and Basic Packed SIMD operations by adamperlin · Pull Request #129703 · dotnet/runtime

adamperlin · 2026-06-22T18:02:00Z

This PR implements support for Vector128.Create on Wasm with known constant args, as well as table-driven expansion and codegen for a large set of PackedSimd opcodes that operate on v128, (v128, v128) and (v128, v128, v128) type operands.

Notably absent from this PR: Vector128.Create with non-constants (requires additional lowering work), and SIMD compares (need special codegen for ulong since i64_u compare is not natively supported on Wasm).

Enable Wasm SIMD ISA for use in the JIT

…-basic-simd

…ector operands)

…simd, simd) param type operations (no integer operands or immediates supported yet)

dotnet-policy-service · 2026-06-22T18:03:16Z

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

Copilot

Pull request overview

This PR extends CoreCLR’s Wasm (RyuJIT) pipeline to recognize Wasm SIMD instruction-set support and begin emitting v128 SIMD code, including constant Vector128.Create lowering to v128.const and table-driven codegen for many PackedSimd operations.

Changes:

Enable Wasm ISA flags (WasmBase, PackedSimd, Vector128) and wire up instruction-set support for Wasm32 (base, simd128).
Add GT_CNS_VEC emission via v128.const, plus v128 load/store selection for TYP_SIMD16.
Introduce initial importer/list/codegen plumbing for PackedSimd and Vector128.Create constants.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
src/libraries/System.Private.CoreLib/src/System.Private.CoreLib.Shared.projitems	Enables `SupportsWasmIntrinsics` for CoreCLR Wasm builds.
src/coreclr/tools/Common/InstructionSetHelpers.cs	Adds Wasm32 baseline instruction sets (`base`, `simd128`).
src/coreclr/jit/lowerwasm.cpp	Starts allowing SIMD-category `GT_HWINTRINSIC` through lowering.
src/coreclr/jit/lower.cpp	Relaxes `TYP_SIMD12` assert for Wasm during lowering checks.
src/coreclr/jit/instr.cpp	Adds Wasm `v128_load`/`v128_store` mapping for `TYP_SIMD16`.
src/coreclr/jit/hwintrinsicwasm.cpp	Implements constant `Vector128.Create` import to `GT_CNS_VEC`.
src/coreclr/jit/hwintrinsiclistwasm.h	Adds `PackedSimd` intrinsic definitions and updates some `Vector128` entries.
src/coreclr/jit/hwintrinsiccodegenwasm.cpp	Adds initial table-driven Wasm HW intrinsic emission.
src/coreclr/jit/hwintrinsic.h	Adds a Wasm `HWIntrinsic` helper wrapper (but currently has a preprocessor issue).
src/coreclr/jit/hwintrinsic.cpp	Adds Wasm `PackedSimd` ISA range and simdSize validation for Wasm.
src/coreclr/jit/compiler.cpp	Forces Wasm baseline ISAs into JIT supported set, adds `Vector128`.
src/coreclr/jit/codegenwasm.cpp	Adds `GT_HWINTRINSIC` dispatch and `GT_CNS_VEC` emission (`v128.const`).
src/coreclr/jit/codegen.h	Declares Wasm vector-constant codegen helper.

Copilot

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 7 comments.

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 4 comments.

…case in SPMI (can happen if NYI_WASM_SIMD fires during import)

…tor elements

Copilot

Pull request overview

Copilot reviewed 14 out of 14 changed files in this pull request and generated 4 comments.

Copilot

Pull request overview

Copilot reviewed 14 out of 14 changed files in this pull request and generated 1 comment.

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 14 out of 14 changed files in this pull request and generated 3 comments.

 HARDWARE_INTRINSIC(Vector128,      Dot,                                                              16,      2,    {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_InvalidNodeId)
 HARDWARE_INTRINSIC(Vector128,      ExtractMostSignificantBits,                                       16,      1,    {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_InvalidNodeId)
-HARDWARE_INTRINSIC(Vector128,      GetElement,                                                       16,      1,    {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_IMM,                   HW_Flag_InvalidNodeId)
+HARDWARE_INTRINSIC(Vector128,      GetElement,                                                       16,      2,    {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_InvalidNodeId)


-HARDWARE_INTRINSIC(Vector128,      ToScalar,                                                         16,     -1,    {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Scalar,                HW_Flag_InvalidNodeId)
-HARDWARE_INTRINSIC(Vector128,      WithElement,                                                      16,      2,    {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_InvalidNodeId)
+HARDWARE_INTRINSIC(Vector128,      ToScalar,                                                         16,      1,    {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_SIMD,                  HW_Flag_InvalidNodeId|HW_Flag_BaseTypeFromFirstArg|HW_Flag_ReturnsScalarT)
+HARDWARE_INTRINSIC(Vector128,      WithElement,                                                      16,      3,    {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_InvalidNodeId)


+                                # A Wasm JIT may exit successfully without writing any disassembly to DOTNET_JitStdOutFile. For example, the wasm JIT
+                                # with JitWasmSimdNyiToR2RUnsupported=1 exits via
+                                # implReadyToRunUnsupported() (CORJIT_R2R_UNSUPPORTED) for NYI_WASM_SIMD during import, so no code is produced. This is an expected behavior.
+                                # TODO-WASM: This check can potentially be removed once we no longer have any NYI's in the import stage.
+                                if not os.path.exists(item_path) and self.coreclr_args.target_arch == "wasm":
+                                    return ""


…-basic-simd

…ck to managed implementation, and implement HARDWARE_INTRINSIC() stubs for all packed simd opcodes to ensure NYI

…ntime into adamperlin/wasm-basic-simd

adamperlin · 2026-06-25T01:20:41Z

@AndyAyersMS Copilot feedback has been implemented for the most part, so this one should be ready for another look. Now all of the PackedSimd intrinsics are at least stubbed out in the table and should cause NYI, so we won't end up trying to call the infinitely recursive managed versions.

adamperlin · 2026-06-25T01:21:44Z

@tannergooding Could you please take a look at this when you have a chance? I don't think a fallback to managed path is possible for the PackedSimd intrinsics as they're just defined as self-calling.

tannergooding · 2026-06-25T13:36:58Z

Could you please take a look at this when you have a chance?

Looking.

I don't think a fallback to managed path is possible for the PackedSimd intrinsics as they're just defined as self-calling.

This is expected. It's only the xplat APIs (i.e. Vector128<T> for WASM) where they fallback to managed if they cannot be accelerated by the JIT.

For the platform specified APIs (i.e. PackedSimd for WASM) they are "must expand" and so the JIT must handle them. This is generally setup in a way that if Isa.IsSupported is configured to return true then the JIT will create a relevant IR node and emit the codegen; otherwise it sees mustExpand = true and forces it to generate a throw new PlatformNotSupportedException instead. See https://github.com/dotnet/runtime/blob/main/src/coreclr/jit/importercalls.cpp#L11472-L11531 and https://github.com/dotnet/runtime/blob/main/src/coreclr/jit/importercalls.cpp#L3170-L3285, for where we handle the platform mismatch or the must expand calls for ISAs that aren't supported

For cases where the ISA is supported, but is not fully lit up, there is then a fallback to throw new NotImplementedException here: https://github.com/dotnet/runtime/blob/main/src/coreclr/jit/importercalls.cpp#L3331-L3340; but this should only be hit when you're doing bringup for a new ISA and never in production.

tannergooding · 2026-06-25T13:55:55Z

+            else if (targetArchitecture == TargetArchitecture.Wasm32)
+            {
+                instructionSetSupportBuilder.AddSupportedInstructionSet("base");
+                instructionSetSupportBuilder.AddSupportedInstructionSet("simd128");


Why simd128 and not packedsimd or similar?

tannergooding · 2026-06-25T14:02:39Z

+            NYI_WASM_SIMD("PackedSimd.ReplaceScalar");
+            break;
+
+        case NI_PackedSimd_LoadVector128:


This one should just be gtNewSimdLoadNode, it only produces a GT_IND so the regular addressing mode and other support can kick in, similar with Store.

It's only the scalar and special semantic instructions which can't be represented as a "regular indirection" that we want to persist as GT_HWINTRINSIC nodes here.

tannergooding · 2026-06-25T14:04:34Z

+            NYI_WASM_SIMD("PackedSimd.Splat");
+            break;
+
+        case NI_Vector128_Create:


We have nearly this same logic duplicated across xarch, arm64, and wasm now. It'd probably be good to have some impSimdCreate(intrinsicId) or similar so that the whole "pop stack and check if everything is constant" logic can be shared across them.

Co-authored-by: Tanner Gooding <tagoo@microsoft.com>

adamperlin added 8 commits June 17, 2026 15:05

Implement Wasm codegen for Vector128.Create with const operands

a1cd1f8

Enable Wasm SIMD ISA for use in the JIT

Merge branch 'main' of github.com:dotnet/runtime into adamperlin/wasm…

2921181

…-basic-simd

Fix categories and flags for some Vector128 Wasm hardware intrinsics

f4fd2fe

Checkpoint: PackedSimd.Add now compiles as an intrinsic (with const v…

48e0731

…ector operands)

Add a few more PackedSimd intrinsics

7bf550b

Add table-driven codegen entries for (simd), (simd, simd) and (simd, …

50f2359

…simd, simd) param type operations (no integer operands or immediates supported yet)

Add ifdefs and remove not fully implemented opcode

250aa59

Cleanup (for now) unused code

3766e1b

adamperlin requested a review from MichalStrehovsky as a code owner June 22, 2026 18:02

Copilot AI review requested due to automatic review settings June 22, 2026 18:02

github-actions Bot added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jun 22, 2026

dotnet-policy-service Bot assigned adamperlin Jun 22, 2026

adamperlin requested review from AndyAyersMS and tannergooding June 22, 2026 18:04

Add #if FEATURE_SIMD

46ae244

Copilot started reviewing on behalf of adamperlin June 22, 2026 18:06 View session

Copilot AI reviewed Jun 22, 2026

View reviewed changes

adamperlin added 2 commits June 22, 2026 11:50

jit-format

cb5a7e9

Fix ifdef condition syntax

c0f347d

Copilot AI review requested due to automatic review settings June 22, 2026 18:54

Copilot started reviewing on behalf of adamperlin June 22, 2026 18:54 View session

Copilot AI reviewed Jun 22, 2026

View reviewed changes

Apply suggestions from code review

3d4abae

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings June 22, 2026 19:32

Copilot started reviewing on behalf of adamperlin June 22, 2026 19:32 View session

Copilot AI reviewed Jun 22, 2026

View reviewed changes

Comment thread src/coreclr/jit/hwintrinsicwasm.cpp

Comment thread src/coreclr/jit/hwintrinsicwasm.cpp

Comment thread src/coreclr/jit/hwintrinsicwasm.cpp

Comment thread src/coreclr/jit/hwintrinsic.cpp Outdated

build-analysis Bot mentioned this pull request Jun 22, 2026

Intermittent build failure in AfterSourceBuild: "Could not write state file" #76488

Closed

Fix unconditional NYI_WASM_SIMD lower bailout, handle no disasm file …

628c47a

…case in SPMI (can happen if NYI_WASM_SIMD fires during import)

adamperlin commented Jun 23, 2026

View reviewed changes

Comment thread src/coreclr/jit/instr.cpp

Comment thread src/coreclr/jit/hwintrinsicwasm.cpp

Copilot started reviewing on behalf of adamperlin June 23, 2026 18:33 View session

Copilot feedback: remove unconditional import NYI on non-constant vec…

6ecc662

…tor elements

Copilot AI reviewed Jun 23, 2026

View reviewed changes

Comment thread src/coreclr/jit/hwintrinsiclistwasm.h Outdated

Comment thread src/coreclr/jit/hwintrinsiclistwasm.h Outdated

Comment thread src/coreclr/scripts/superpmi.py Outdated

Comment thread src/coreclr/jit/hwintrinsicwasm.cpp

adamperlin added 2 commits June 23, 2026 12:09

Format hwintrinsiclistwasm.h

292bcaf

More copilot feedback

e7cef58

Copilot AI review requested due to automatic review settings June 23, 2026 20:57

Copilot started reviewing on behalf of adamperlin June 23, 2026 20:58 View session

Copilot AI reviewed Jun 23, 2026

View reviewed changes

Comment thread src/coreclr/jit/codegenwasm.cpp

Potential fix for pull request finding

5a93f14

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings June 23, 2026 22:05

Copilot started reviewing on behalf of adamperlin June 23, 2026 22:05 View session

Copilot AI reviewed Jun 23, 2026

View reviewed changes

Merge branch 'main' of github.com:dotnet/runtime into adamperlin/wasm…

3b52445

…-basic-simd

Copilot AI mentioned this pull request Jun 24, 2026

Fix flaky SendAsync_Success_ConnectionSetupActivityGraphRecorded test (Http1.1 + Http2) #129822

Closed

adamperlin added 2 commits June 24, 2026 18:14

Address more review feedback: Ensure Vector128.GetElement can fall ba…

01e61c6

…ck to managed implementation, and implement HARDWARE_INTRINSIC() stubs for all packed simd opcodes to ensure NYI

Merge branch 'adamperlin/wasm-basic-simd' of github.com:adamperlin/ru…

ef9b870

…ntime into adamperlin/wasm-basic-simd

This was referenced Jun 25, 2026

TestNativeDigits fails for ur-IN on Apple platforms (xunit v3 exposed previously non-running test) #125933

Open

System.IO.Tests.Directory_Move_Tests.Directory_Move_Multiple_From_Watched_To_Unwatched_Mac #127651

Open

tannergooding reviewed Jun 25, 2026

View reviewed changes

Comment thread src/coreclr/jit/hwintrinsic.cpp Outdated

tannergooding reviewed Jun 25, 2026

View reviewed changes

Update src/coreclr/jit/hwintrinsic.cpp

aff1238

Co-authored-by: Tanner Gooding <tagoo@microsoft.com>

Copilot AI review requested due to automatic review settings June 25, 2026 18:10

Copilot started reviewing on behalf of adamperlin June 25, 2026 18:10 View session

Uh oh!

Conversation

adamperlin commented Jun 22, 2026

Uh oh!

dotnet-policy-service Bot commented Jun 22, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

adamperlin commented Jun 25, 2026

Uh oh!

adamperlin commented Jun 25, 2026

Uh oh!

tannergooding commented Jun 25, 2026

Uh oh!

tannergooding Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

tannergooding Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

tannergooding Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants