Update the JIT to share the xplat vector tables across platforms by tannergooding · Pull Request #130049 · dotnet/runtime

tannergooding · 2026-06-30T18:40:01Z

This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set.

It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times.

I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

dotnet-policy-service · 2026-06-30T18:42:05Z

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

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

…ted synthetic instruction set

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 2

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 6

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 0 new

… marking

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

adamperlin

I'm going to approve this pending CI (and the latest 64 bit spmi runs). This was a fairly mechanical refactor overall and basic checks for consistency seem to look good. It seems like any asmdiffs that did come up are accounted for at this point, and we have minimal TP diff.

tannergooding · 2026-07-02T22:28:17Z

/ba-g unrelated riscv build failure, occurring in multiple prs

…0049) This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set. It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times. I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

tannergooding · 2026-06-30T18:40:01Z

This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set.

It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times.

I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

dotnet-policy-service · 2026-06-30T18:42:05Z

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

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

…ted synthetic instruction set

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 2

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 6

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 0 new

… marking

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

adamperlin

I'm going to approve this pending CI (and the latest 64 bit spmi runs). This was a fairly mechanical refactor overall and basic checks for consistency seem to look good. It seems like any asmdiffs that did come up are accounted for at this point, and we have minimal TP diff.

tannergooding · 2026-07-02T22:28:17Z

/ba-g unrelated riscv build failure, occurring in multiple prs

…0049) This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set. It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times. I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

tannergooding · 2026-06-30T18:40:01Z

This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set.

It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times.

I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

dotnet-policy-service · 2026-06-30T18:42:05Z

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

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

…ted synthetic instruction set

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 2

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 6

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 0 new

… marking

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

adamperlin

I'm going to approve this pending CI (and the latest 64 bit spmi runs). This was a fairly mechanical refactor overall and basic checks for consistency seem to look good. It seems like any asmdiffs that did come up are accounted for at this point, and we have minimal TP diff.

tannergooding · 2026-07-02T22:28:17Z

/ba-g unrelated riscv build failure, occurring in multiple prs

…0049) This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set. It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times. I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

tannergooding · 2026-06-30T18:40:01Z

This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set.

It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times.

I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

dotnet-policy-service · 2026-06-30T18:42:05Z

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

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

…ted synthetic instruction set

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 2

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 6

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 0 new

… marking

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

adamperlin

I'm going to approve this pending CI (and the latest 64 bit spmi runs). This was a fairly mechanical refactor overall and basic checks for consistency seem to look good. It seems like any asmdiffs that did come up are accounted for at this point, and we have minimal TP diff.

tannergooding · 2026-07-02T22:28:17Z

/ba-g unrelated riscv build failure, occurring in multiple prs

…0049) This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set. It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times. I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

tannergooding · 2026-06-30T18:40:01Z

This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set.

It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times.

I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

dotnet-policy-service · 2026-06-30T18:42:05Z

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

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

…ted synthetic instruction set

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 2

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 6

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 0 new

… marking

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

adamperlin

I'm going to approve this pending CI (and the latest 64 bit spmi runs). This was a fairly mechanical refactor overall and basic checks for consistency seem to look good. It seems like any asmdiffs that did come up are accounted for at this point, and we have minimal TP diff.

tannergooding · 2026-07-02T22:28:17Z

/ba-g unrelated riscv build failure, occurring in multiple prs

…0049) This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set. It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times. I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

tannergooding · 2026-06-30T18:40:01Z

This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set.

It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times.

I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

dotnet-policy-service · 2026-06-30T18:42:05Z

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

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

…ted synthetic instruction set

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 2

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 6

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 0 new

… marking

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

adamperlin

I'm going to approve this pending CI (and the latest 64 bit spmi runs). This was a fairly mechanical refactor overall and basic checks for consistency seem to look good. It seems like any asmdiffs that did come up are accounted for at this point, and we have minimal TP diff.

tannergooding · 2026-07-02T22:28:17Z

/ba-g unrelated riscv build failure, occurring in multiple prs

…0049) This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set. It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times. I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

tannergooding · 2026-06-30T18:40:01Z

This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set.

It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times.

I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

dotnet-policy-service · 2026-06-30T18:42:05Z

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

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

…ted synthetic instruction set

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 2

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 6

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 0 new

… marking

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

adamperlin

I'm going to approve this pending CI (and the latest 64 bit spmi runs). This was a fairly mechanical refactor overall and basic checks for consistency seem to look good. It seems like any asmdiffs that did come up are accounted for at this point, and we have minimal TP diff.

tannergooding · 2026-07-02T22:28:17Z

/ba-g unrelated riscv build failure, occurring in multiple prs

…0049) This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set. It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times. I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

tannergooding · 2026-06-30T18:40:01Z

This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set.

It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times.

I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

dotnet-policy-service · 2026-06-30T18:42:05Z

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

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

…ted synthetic instruction set

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 2

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 6

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 0 new

… marking

Copilot

Copilot's findings

Files reviewed: 34/39 changed files
Comments generated: 1

adamperlin

I'm going to approve this pending CI (and the latest 64 bit spmi runs). This was a fairly mechanical refactor overall and basic checks for consistency seem to look good. It seems like any asmdiffs that did come up are accounted for at this point, and we have minimal TP diff.

tannergooding · 2026-07-02T22:28:17Z

/ba-g unrelated riscv build failure, occurring in multiple prs

…0049) This massively simplifies and unifies the xplat vector intrinsic support by unifying it to a single table and instruction set. It removes around 3000 lines of code (more simplifications are possible in the future as well) and enables us to add WASM and Arm64 SVE support for the xplat intrinsics without duplicating the import logic two more times. I've generally split it out into two commits. The first is the general refactoring and the second is some of the base support for WASM lightup. There's some pending platform specific work and some more xplat work required before WASM can be fully enabled.

tannergooding added 2 commits June 30, 2026 11:02

Update the JIT to share the xplat vector tables across platforms
444176b

Prepatory work for some of the xplat SIMD support for WASM
3f84a72

CopilotAI review requested due to automatic review settings June 30, 2026 18:40

github-actionsBot added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jun 30, 2026

Copilotstarted reviewing on behalf of tannergooding June 30, 2026 18:41 View session

dotnet-policy-serviceBot assigned tannergooding Jun 30, 2026

CopilotAI reviewed Jun 30, 2026

tannergooding added 3 commits June 30, 2026 14:23

Apply formatting patch
65a0edb

Merge remote-tracking branch 'dotnet/main' into wasm-simd
8ed5bf5

Resolve merge conflicts and don't bump JIT/EE version for a non repor…
cace755
…ted synthetic instruction set

build-analysisBot mentioned this pull request Jun 30, 2026
Multiple Helix work items fail on maccatalyst/tvos CoreCLR Release #126460
Closed

Fix EMSB to use the right intrinsic
8b2b938

CopilotAI review requested due to automatic review settings June 30, 2026 22:40

Copilotstarted reviewing on behalf of tannergooding June 30, 2026 22:41 View session

CopilotAI reviewed Jun 30, 2026
View reviewed changes

Comment threadsrc/coreclr/jit/rationalize.cpp

tannergooding added 3 commits June 30, 2026 16:36

Fix a case that was falling through incorrectly
e840822

Fix NarrowWithSaturation on Arm64
fa9aca2

Don't handle V128 Get/With Lower/Upper on xarch
0a63d85

CopilotAI review requested due to automatic review settings July 1, 2026 01:24

Copilotstarted reviewing on behalf of tannergooding July 1, 2026 01:25 View session

CopilotAI reviewed Jul 1, 2026
View reviewed changes

Comment threadsrc/coreclr/jit/valuenum.cpp
Comment threadsrc/coreclr/jit/rationalize.cpp

tannergooding added 2 commits June 30, 2026 19:43

Fix an assert on xarch and ConvertToDouble on Arm64
1963f49

Fix WithElement and don't introduce new folding to avoid diffs
ec57ba7

CopilotAI review requested due to automatic review settings July 1, 2026 10:58

Copilotstarted reviewing on behalf of tannergooding July 1, 2026 10:59 View session

CopilotAI reviewed Jul 1, 2026
View reviewed changes

Comment threadsrc/coreclr/jit/compiler.h
Comment threadsrc/coreclr/jit/compiler.h
Comment threadsrc/coreclr/jit/lclmorph.cpp
Comment threadsrc/coreclr/jit/lclmorph.cpp
Comment threadsrc/coreclr/jit/hwintrinsiclist.h Outdated
Comment threadsrc/coreclr/jit/hwintrinsicwasm.cpp

CopilotAI review requested due to automatic review settings July 1, 2026 15:57

tannergooding force-pushed the wasm-simd branch from 75b8a84 to 7d172cbCompare July 1, 2026 15:57

Copilotstarted reviewing on behalf of tannergooding July 1, 2026 15:57 View session

adamperlin reviewed Jul 2, 2026
View reviewed changes

Comment threadsrc/coreclr/jit/simd.h
Comment threadsrc/coreclr/jit/hwintrinsiclist.h

Don't import WithElement on Arm64 for out of range constant
f15a0e5

CopilotAI review requested due to automatic review settings July 2, 2026 02:45

Copilotstarted reviewing on behalf of tannergooding July 2, 2026 02:46 View session

tannergooding commented Jul 2, 2026
View reviewed changes

Comment threadsrc/coreclr/jit/hwintrinsic.cpp Outdated

CopilotAI reviewed Jul 2, 2026
View reviewed changes

Comment threadsrc/coreclr/jit/hwintrinsiccodegenxarch.cpp

tannergooding added 2 commits July 1, 2026 20:00

Use the return type for the selected size
c9ed274

Ensure the arm64 check is before arguments are popped
7ef0e19

CopilotAI review requested due to automatic review settings July 2, 2026 13:04

Copilotstarted reviewing on behalf of tannergooding July 2, 2026 13:05 View session

CopilotAI reviewed Jul 2, 2026
View reviewed changes

tannergooding added 2 commits July 2, 2026 09:27

Ensure that NI_Vector_AsVector128Unsafe is still handled in delayFree…
065068f
… marking

Merge remote-tracking branch 'dotnet/main' into wasm-simd
f8ecf3f

CopilotAI review requested due to automatic review settings July 2, 2026 18:04

Copilotstarted reviewing on behalf of tannergooding July 2, 2026 18:04 View session

CopilotAI reviewed Jul 2, 2026
View reviewed changes

Comment threadsrc/coreclr/tools/Common/JitInterface/ThunkGenerator/InstructionSetGenerator.cs

adamperlin approved these changes Jul 2, 2026
View reviewed changes

tannergooding enabled auto-merge (squash) July 2, 2026 19:09

tannergooding merged commit 57fc9dc into dotnet:mainJul 2, 2026
143 of 145 checks passed

tannergooding deleted the wasm-simd branch July 2, 2026 22:28

tannergooding mentioned this pull request Jul 2, 2026
ILC failing with exit code 139 for risc-v build #130145
Closed

dotnet-maestroBot mentioned this pull request Jul 3, 2026
[main] Source code updates from dotnet/runtime dotnet/dotnet#7538
Merged

dotnet-milestone-botBot added this to the 11.0-preview7 milestone Jul 3, 2026

EgorBo mentioned this pull request Jul 6, 2026
Assertion failed 'comp->compIsaSupportedDebugOnly(InstructionSet_AVX2)' #130230
Closed

github-actionsBot locked and limited conversation to collaborators Aug 3, 2026