stdarch subtree update #151711

folkertdev · 2026-01-26T21:22:37Z

Subtree update of stdarch to rust-lang/stdarch@9ba0a3f.

Created using https://github.com/rust-lang/josh-sync.

This allows the `types` macro to easily implement `From<Simd<T, N>>` and `Into<Simd<T, N>>`

…oid-unsafe-tests Use a generic type to implement SIMD types in `core_arch::simd` and avoid `unsafe fn` in more tests

For other targets we don't have such instances. We specifically decided agains implementing arithmetic operators on these low-level vector types

The inline(always) attribute is now applied to the call site when a function has target features enabled so that it can determine that the call is safe to inline.

thread, grid, and block dim/idx can only return non-negative values

remove `impl Neg` on s390x/powerpc vector types

Mark the neon intrinsics as inline(always).

use `simd_extract_dyn` for `extract`

Because `qemu` does not support these (yet), I haven't added any runtime tests

s390x: add `nnp-assist` intrinsics

test `vmulh_lane_f16` in intrinsic test

Stabilize `__jcvt`

This updates the rust-version file to 873d468.

Pull recent changes from https://github.com/rust-lang/rust via Josh. Upstream ref: rust-lang/rust@873d468 Filtered ref: rust-lang/stdarch@003d326 Upstream diff: rust-lang/rust@4862272...873d468 This merge was created using https://github.com/rust-lang/josh-sync.

Rustc pull update

use `simd_splat`

rustbot · 2026-01-26T21:22:40Z

stdarch is developed in its own repository. If possible, consider making this change to rust-lang/stdarch instead.

cc @Amanieu, @sayantn

this is required for compiling `stdarch`

folkertdev · 2026-01-26T22:21:56Z

I've enabled #![feature(target_feature_inline_always)] now, would that need T-lang/T-libs approval? It is just a hint, so I think it's fine, but just checking.

folkertdev · 2026-01-26T22:26:58Z

cc @davidtwco @RalfJung is #![feature(target_feature_inline_always)] ready for use in std?

RalfJung · 2026-01-27T07:12:34Z

The tracking issue contains a fresh example of what looks like a soundness problem with that feature...

folkertdev · 2026-01-27T13:47:49Z

based on #145574 (comment) I'm going to go ahead here.

@bors r+

rust-bors · 2026-01-27T13:47:53Z

📌 Commit fdad66a has been approved by folkertdev

It is now in the queue for this repository.

@ghost

…, r=folkertdev stdarch subtree update Subtree update of `stdarch` to rust-lang/stdarch@9ba0a3f. Created using https://github.com/rust-lang/josh-sync. r? @ghost

…uwer Rollup of 4 pull requests Successful merges: - #150491 (resolve: Mark items under exported ambiguous imports as exported) - #151161 (std: move time implementations to `sys`) - #151694 (more `proc_macro` bridge cleanups) - #151711 (stdarch subtree update)

…uwer Rollup of 4 pull requests Successful merges: - #151161 (std: move time implementations to `sys`) - #151694 (more `proc_macro` bridge cleanups) - #151711 (stdarch subtree update) - #150557 (Don't try to evaluate const blocks during constant promotion)

@ghost

Rollup merge of #151711 - folkertdev:stdarch-sync-2026-01-26, r=folkertdev stdarch subtree update Subtree update of `stdarch` to rust-lang/stdarch@9ba0a3f. Created using https://github.com/rust-lang/josh-sync. r? @ghost

…uwer Rollup of 4 pull requests Successful merges: - rust-lang/rust#151161 (std: move time implementations to `sys`) - rust-lang/rust#151694 (more `proc_macro` bridge cleanups) - rust-lang/rust#151711 (stdarch subtree update) - rust-lang/rust#150557 (Don't try to evaluate const blocks during constant promotion)

JonathanBrouwer · 2026-01-28T21:15:13Z

Congrats! This PR caused a significant performance boost in rollup: #151749 (comment)
Though if this will happen more often you might want to rollup=iffy or rollup=never these subtree updates

folkertdev · 2026-01-28T21:45:34Z

Huh, weird. I guess ripgrep might benefit from simd_splat (this PR includes the change where the platform intrinsics actually start using simd_splat instead of a weird shuffle pattern that LLVM didn't know how to optimize very well)? Otherwise most projects just don't use the stdarch intrinsics really I think.

Anyway, good point, I'll keep that in mind.

…uwer Rollup of 4 pull requests Successful merges: - rust-lang/rust#151161 (std: move time implementations to `sys`) - rust-lang/rust#151694 (more `proc_macro` bridge cleanups) - rust-lang/rust#151711 (stdarch subtree update) - rust-lang/rust#150557 (Don't try to evaluate const blocks during constant promotion)

RalfJung · 2026-01-29T07:01:13Z

That benchmark is measuring compile times -- so somehow rustc itself or a crate it uses is benefiting from this.

folkertdev · 2026-01-29T13:42:13Z

it's time spent compiling the crate right? So the compiler emitting better LLVM IR for some function could be the cause?

The compiler itself does not use the stdarch intrinsics that were touched as far as I know. We have some string/slice-related SIMD implementations but they don't appear to use splat.

ZuseZ4 and others added 24 commits December 9, 2025 10:52

thread, grid, and block dim/idx can only return non-negative values

79b30d6

Use a generic type to implement SIMD types in core_arch::simd

69c7a7e

This allows the `types` macro to easily implement `From<Simd<T, N>>` and `Into<Simd<T, N>>`

x86: avoid using transmute in some tests

dab5532

arm_shared: avoid using transmute and unsafe fn in tests

68701ff

Merge pull request rust-lang#1987 from eduardosm/generic-simd-type-av…

f9e6d4b

…oid-unsafe-tests Use a generic type to implement SIMD types in `core_arch::simd` and avoid `unsafe fn` in more tests

remove impl Neg on s390x/powerpc vector types

d0204a3

For other targets we don't have such instances. We specifically decided agains implementing arithmetic operators on these low-level vector types

Mark the neon intrinsics as inline(always).

78952f6

The inline(always) attribute is now applied to the call site when a function has target features enabled so that it can determine that the call is safe to inline.

stabilize __jscvt

5bf9ca7

Merge pull request rust-lang#1973 from ZuseZ4/u32-threadidx

aaf31e7

thread, grid, and block dim/idx can only return non-negative values

Merge pull request rust-lang#1992 from folkertdev/remove-impl-neg

0d802f5

remove `impl Neg` on s390x/powerpc vector types

Merge pull request rust-lang#1974 from JamieCunliffe/inline-always-neon

d251a38

Mark the neon intrinsics as inline(always).

test vmulh_lane_f16 in intrinsic test

bf7c1c2

use simd_extract_dyn for extract

b1ef655

Merge pull request rust-lang#1997 from folkertdev/use-extract-dyn

5829f21

use `simd_extract_dyn` for `extract`

s390x: add nnp-assist intrinsics

cf08a11

Because `qemu` does not support these (yet), I haven't added any runtime tests

Merge pull request rust-lang#1996 from folkertdev/s390x-nnp-assist

84fc953

s390x: add `nnp-assist` intrinsics

Merge pull request rust-lang#1998 from folkertdev/test-vmulh_lane_f16

5ec32b5

test `vmulh_lane_f16` in intrinsic test

Merge pull request rust-lang#1993 from folkertdev/stabilize-jscvt

464b309

Stabilize `__jcvt`

Prepare for merging from rust-lang/rust

a4f92e5

This updates the rust-version file to 873d468.

use simd_splat

246399c

Merge pull request rust-lang#2002 from rust-lang/rustc-pull

a45071b

Rustc pull update

use simd_splat for the set1 functions

0ef4c3a

Merge pull request rust-lang#2003 from folkertdev/use-simd-splat

f648ba2

use `simd_splat`

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Jan 26, 2026

This comment has been minimized.

Sign in to view

enable target_feature_inline_always in core and std

fdad66a

this is required for compiling `stdarch`

rust-bors bot added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jan 27, 2026

JonathanBrouwer mentioned this pull request Jan 27, 2026

Rollup of 4 pull requests #151743

Closed

JonathanBrouwer mentioned this pull request Jan 27, 2026

Rollup of 4 pull requests #151749

Merged

rust-bors bot merged commit 7d11720 into rust-lang:main Jan 27, 2026
11 checks passed

rustbot added this to the 1.95.0 milestone Jan 27, 2026

This was referenced Jan 27, 2026

Inlining + target_feature broken in powerpc64 #60637

Closed

Bad codegen for widen followed by ARM vdupq_n_* #137407

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

stdarch subtree update #151711

stdarch subtree update #151711

folkertdev commented Jan 26, 2026

Uh oh!

rustbot commented Jan 26, 2026

Uh oh!

This comment has been minimized.

folkertdev commented Jan 26, 2026

Uh oh!

folkertdev commented Jan 26, 2026

Uh oh!

RalfJung commented Jan 27, 2026 via email

Uh oh!

folkertdev commented Jan 27, 2026

Uh oh!

rust-bors bot commented Jan 27, 2026

Uh oh!

Uh oh!

JonathanBrouwer commented Jan 28, 2026 •

edited

Loading

Uh oh!

folkertdev commented Jan 28, 2026

Uh oh!

RalfJung commented Jan 29, 2026

Uh oh!

folkertdev commented Jan 29, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

10 participants

Uh oh!

stdarch subtree update #151711

stdarch subtree update #151711

Conversation

folkertdev commented Jan 26, 2026

Uh oh!

rustbot commented Jan 26, 2026

Uh oh!

This comment has been minimized.

folkertdev commented Jan 26, 2026

Uh oh!

folkertdev commented Jan 26, 2026

Uh oh!

RalfJung commented Jan 27, 2026 via email

Uh oh!

folkertdev commented Jan 27, 2026

Uh oh!

rust-bors bot commented Jan 27, 2026

Uh oh!

Uh oh!

JonathanBrouwer commented Jan 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

folkertdev commented Jan 28, 2026

Uh oh!

RalfJung commented Jan 29, 2026

Uh oh!

folkertdev commented Jan 29, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

10 participants

JonathanBrouwer commented Jan 28, 2026 •

edited

Loading