Closed
Description
std::arch::x86_64::_mm_extract_ps
(and x86
) is documented as
Extract a single-precision (32-bit) floating-point element from a, selected with imm8
but has the signature
pub unsafe fn _mm_extract_ps(a: __m128, imm8: i32) -> i32
Shouldn't this be returning f32
? The implementation just uses mem::transmute
with inferred types, so its just transmuting the result to i32
even though the implementation is using simd_transmute
to make an f32
.
Is this a bug or am I missing something here?