Skip to content

s390x: add f32 min/max tests #2059

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.

Sign up for GitHub

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

Closed
RalfJung wants to merge 1 commit into from
+13 −0
Closed

s390x: add f32 min/max tests #2059

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 13 additions & 0 deletions crates/core_arch/src/s390x/vector.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7477,6 +7477,19 @@ mod tests {
[0, !0, !0, !0]
}

// f32 is the tricky case for max/min as that needs a fallback on z13
test_vec_2! { test_vec_max, vec_max, f32x4, f32x4 -> f32x4,
[1.0, f32::NAN, f32::INFINITY, 2.0],
[-10.0, -10.0, 5.0, f32::NAN],
[1.0, -10.0, f32::INFINITY, 2.0]
Copy link
Copy Markdown
Member Author

@RalfJung RalfJung Mar 10, 2026

Choose a reason for hiding this comment

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

Actually I expect this to fail as the fallback does not handle NaNs correctly.

Copy link
Copy Markdown
Member Author

@RalfJung RalfJung Mar 10, 2026

Choose a reason for hiding this comment

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

Hm, strange. Tests pass with and without the enhancement target feature. Is the fallback actually more accurate than the docs say?

Copy link
Copy Markdown
Member Author

@RalfJung RalfJung Mar 11, 2026 •
edited
Loading

Choose a reason for hiding this comment

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

Ah no, the test probably gets optimized away. (EDIT: Or maybe not, LLVM might just do a proper fallback that matches what it documents rather than the IBM docs.)

Still useful to have the test as it would have shown the issue with #2058.

}

test_vec_2! { test_vec_min, vec_min, f32x4, f32x4 -> f32x4,
[1.0, f32::NAN, f32::INFINITY, 2.0],
[-10.0, -10.0, 5.0, f32::NAN],
[-10.0, -10.0, 5.0, 2.0]
}

#[simd_test(enable = "vector")]
fn test_vec_meadd() {
let a = vector_unsigned_short([1, 0, 2, 0, 3, 0, 4, 0]);
Expand Down
Loading