Skip to content

Set target features depending on the target architecture #4127

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 7 commits into from
Jun 5, 2025

Conversation

zhassan-aws
Copy link
Contributor

Updates the target_config method to set the target features depending on the target architecture.

Resolves #4119

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 and MIT licenses.

@zhassan-aws zhassan-aws requested a review from a team as a code owner June 3, 2025 23:38
@github-actions github-actions bot added the Z-EndToEndBenchCI Tag a PR to run benchmark CI label Jun 3, 2025
Copy link
Contributor

@carolynzech carolynzech left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Smart fix! Thanks!

github-merge-queue bot pushed a commit that referenced this pull request Jun 5, 2025
This PR is co-authored with @tautschnig.

Background: In #4127 we
enabled certain target features that are platform dependent (e.g. `sse`
and `sse2` for x86_64 and `neon` for aarch64) which resulted in using
the `simd_bitmask` intrinsic more frequently. Kani's current model of
that intrinsic
(https://github.com/model-checking/kani/blob/5f14b735b74f3ae3f9f1c64ce5656e1e735d42ea/library/kani/src/models/mod.rs#L72)
uses a loop that iterates `LANE` times, which requires harnesses that
touch this code (e.g. that use the `HashSet` data structure) to have a
large unwinding value, which results in a significant slowdown.

This PR refactors this function through rewriting it using a nested loop
and manually unwinding the inner loop (which operates over the bits in a
byte) so that a large unwinding value is not needed.

By submitting this pull request, I confirm that my contribution is made
under the terms of the Apache 2.0 and MIT licenses.
@zhassan-aws
Copy link
Contributor Author

There is a serious slowdown in one regression:

hashset/check_contains_str | 767.537 | 33.0624 | +2221.480%

I filed #4132 to track this issue.

@zhassan-aws zhassan-aws added this pull request to the merge queue Jun 5, 2025
Merged via the queue into model-checking:main with commit 931661b Jun 5, 2025
25 of 26 checks passed
@zhassan-aws zhassan-aws deleted the iss4119 branch June 5, 2025 04:22
zhassan-aws added a commit that referenced this pull request Jun 5, 2025
This PR is co-authored with @tautschnig.

Background: In #4127 we
enabled certain target features that are platform dependent (e.g. `sse`
and `sse2` for x86_64 and `neon` for aarch64) which resulted in using
the `simd_bitmask` intrinsic more frequently. Kani's current model of
that intrinsic
(https://github.com/model-checking/kani/blob/5f14b735b74f3ae3f9f1c64ce5656e1e735d42ea/library/kani/src/models/mod.rs#L72)
uses a loop that iterates `LANE` times, which requires harnesses that
touch this code (e.g. that use the `HashSet` data structure) to have a
large unwinding value, which results in a significant slowdown.

This PR refactors this function through rewriting it using a nested loop
and manually unwinding the inner loop (which operates over the bits in a
byte) so that a large unwinding value is not needed.

By submitting this pull request, I confirm that my contribution is made
under the terms of the Apache 2.0 and MIT licenses.
zhassan-aws added a commit that referenced this pull request Jun 5, 2025
Updates the `target_config` method to set the target features depending
on the target architecture.

Resolves #4119

By submitting this pull request, I confirm that my contribution is made
under the terms of the Apache 2.0 and MIT licenses.
github-merge-queue bot pushed a commit that referenced this pull request Jun 9, 2025
These are the automatically-generated release notes:
```
## What's Changed
* Toolchain upgrade to nightly-2025-05-04 by @thanhnguyen-aws in #4059
* Automatic toolchain upgrade to nightly-2025-05-05 by @github-actions in #4060
* Automatic toolchain upgrade to nightly-2025-05-06 by @github-actions in #4061
* Enable target features: x87 and sse2 by @thanhnguyen-aws in #4062
* Fix the bug: Loop contracts are not composable with function contracts  by @thanhnguyen-aws in #3979
* Automatic cargo update to 2025-05-12 by @github-actions in #4066
* Bump tests/perf/s2n-quic from `6aa9975` to `5f323b7` by @dependabot in #4068
* Fix stabilization instructions in RFC intro by @carolynzech in #4067
* Add support for quantifiers by @qinheping in #3993
* Toolchain upgrade to nightly-2025-05-07 by @thanhnguyen-aws in #4070
* Automatic toolchain upgrade to nightly-2025-05-08 by @github-actions in #4071
* Automatic toolchain upgrade to nightly-2025-05-09 by @github-actions in #4072
* Automatic toolchain upgrade to nightly-2025-05-10 by @github-actions in #4073
* Clippy/Stylistic Fixes by @carolynzech in #4074
* Upgrade toolchain to 2025-05-14 by @zhassan-aws in #4076
* Autoharness argument validation: only error on `--quiet` if `--list` was passed by @carolynzech in #4069
* Upgrade Rust toolchain to 2025-05-16 by @zhassan-aws in #4080
* Automatic toolchain upgrade to nightly-2025-05-17 by @github-actions in #4081
* Add setup scripts for Ubuntu 20.04 by @zhassan-aws in #4082
* Automatic toolchain upgrade to nightly-2025-05-18 by @github-actions in #4083
* Automatic cargo update to 2025-05-19 by @github-actions in #4086
* Automatic toolchain upgrade to nightly-2025-05-19 by @github-actions in #4085
* Automatic toolchain upgrade to nightly-2025-05-20 by @github-actions in #4091
* Bump tests/perf/s2n-quic from `5f323b7` to `22434aa` by @dependabot in #4089
* Fix the error that Kani panics when there is no external parameter in quantifier's closure. by @thanhnguyen-aws in #4088
* Update toolchain to 2025-05-22 by @carolynzech in #4098
* Use our toolchain when invoking `cargo metadata` by @carolynzech in #4090
* Automatic toolchain upgrade to nightly-2025-05-23 by @github-actions in #4099
* Automatic toolchain upgrade to nightly-2025-05-24 by @github-actions in #4101
* Automatic toolchain upgrade to nightly-2025-05-25 by @github-actions in #4102
* Fix a bug codegening `SwitchInt`s with only an otherwise branch by @bkirwi in #4095
* Automatic toolchain upgrade to nightly-2025-05-26 by @github-actions in #4104
* Automatic cargo update to 2025-05-26 by @github-actions in #4105
* Bump tests/perf/s2n-quic from `22434aa` to `550afb3` by @dependabot in #4106
* Automatic toolchain upgrade to nightly-2025-05-27 by @github-actions in #4107
* Update `kani::mem` pointer validity documentation by @carolynzech in #4092
* Add support for edition 2018 crates using assert! (Fixes #3717) by @sintemal in #4096
* Automatic toolchain upgrade to nightly-2025-05-28 by @github-actions in #4113
* Automatic toolchain upgrade to nightly-2025-05-29 by @github-actions in #4115
* Automatic toolchain upgrade to nightly-2025-05-30 by @github-actions in #4118
* Handle generic defaults in BoundedArbitrary derives by @zhassan-aws in #4117
* Automatic cargo update to 2025-06-02 by @github-actions in #4121
* Bump tests/perf/s2n-quic from `550afb3` to `8f54b57` by @dependabot in #4122
* Upgrade Rust toolchain to 2025-06-02 by @zhassan-aws in #4123
* Automatic toolchain upgrade to nightly-2025-06-03 by @github-actions in #4125
* Finish deprecating `--enable-unstable`, `--restrict-vtable`, and `--write-json-symtab` by @carolynzech in #4110
* `ty_mangled_name`: only use non-mangled name if `-Zcffi` is enabled. by @carolynzech in #4114
* Improve Help Menu by @carolynzech in #4109
* Start stabilizing `--jobs` and `list`; deprecate default memory checks by @carolynzech in #4108
* Refactor simd_bitmask to reduce the number of iterations by @zhassan-aws in #4129
* Set target features depending on the target architecture by @zhassan-aws in #4127
* Bump some versions suggested by cargo-outdated by @zhassan-aws in #4131
* Improve linking error output for `#[no_std]` crates by @AlexanderPortland in #4126
* Fix the git log command in the toolchain update script by @zhassan-aws in #4139
* Gate quantifiers behind an experimental feature by @thanhnguyen-aws in #4141
* Automatic cargo update to 2025-06-09 by @github-actions in #4145

## New Contributors
* @bkirwi made their first contribution in #4095
* @sintemal made their first contribution in #4096
* @AlexanderPortland made their first contribution in #4126

**Full Changelog**: kani-0.62.0...kani-0.63.0
```


By submitting this pull request, I confirm that my contribution is made
under the terms of the Apache 2.0 and MIT licenses.

---------

Co-authored-by: Carolyn Zech <carolynzech@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Z-EndToEndBenchCI Tag a PR to run benchmark CI
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Compilation fails with errors in ring
3 participants