user0/rust
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Add SSE2 trivial aliases and conversions. (#165)

Browse source 
`_mm_cvtsd_f64`, `_mm_cvtsd_si64x` and `_mm_cvttsd_si64x`.
See #40.
This commit is contained in:
Malo Jaffré 2017-11-02 19:10:50 +01:00 • committed by Alex Crichton
parent 542aac988a
commit 74870635e5
1 changed files with 42 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

46
library/stdarch/src/x86/sse2.rs
View file

@ -1787,6 +1787,13 @@ pub unsafe fn _mm_cvtsd_si64(a: f64x2) -> i64 {
cvtsd2si64(a)
}
/// Alias for [`_mm_cvtsd_si64`](fn._mm_cvtsd_si64_ss.html).
#[cfg(target_arch = "x86_64")]
#[inline(always)]
#[target_feature = "+sse2"]
#[cfg_attr(test, assert_instr(cvtsd2si))]
pub unsafe fn _mm_cvtsd_si64x(a: f64x2) -> i64 { _mm_cvtsd_si64(a) }
/// Convert the lower double-precision (64-bit) floating-point element in `b`
/// to a single-precision (32-bit) floating-point element, store the result in
/// the lower element of the return value, and copy the upper element from `a`
@ -1798,6 +1805,14 @@ pub unsafe fn _mm_cvtsd_ss(a: f32x4, b: f64x2) -> f32x4 {
cvtsd2ss(a, b)
}
/// Return the lower double-precision (64-bit) floating-point element of "a".
#[inline(always)]
#[target_feature = "+sse2"]
// no particular instruction to test
pub unsafe fn _mm_cvtsd_f64(a: f64x2) -> f64 {
a.extract(0)
}
/// Convert the lower single-precision (32-bit) floating-point element in `b`
/// to a double-precision (64-bit) floating-point element, store the result in
/// the lower element of the return value, and copy the upper element from `a`
@ -1837,6 +1852,13 @@ pub unsafe fn _mm_cvttsd_si64(a: f64x2) -> i64 {
cvttsd2si64(a)
}
/// Alias for [`_mm_cvttsd_si64`](fn._mm_cvttsd_si64_ss.html).
#[cfg(target_arch = "x86_64")]
#[inline(always)]
#[target_feature = "+sse2"]
#[cfg_attr(test, assert_instr(cvttsd2si))]
pub unsafe fn _mm_cvttsd_si64x(a: f64x2) -> i64 { _mm_cvttsd_si64(a) }
/// Convert packed single-precision (32-bit) floating-point elements in `a` to
/// packed 32-bit integers with truncation.
#[inline(always)]
@ -3979,8 +4001,14 @@ mod tests {
let r = sse2::_mm_cvtsd_si64(f64x2::new(f64::MAX, f64::MIN));
assert_eq!(r, i64::MIN);
}
let r = sse2::_mm_cvtsd_si64(f64x2::new(f64::NAN, f64::NAN));
#[cfg(target_arch = "x86_64")]
#[simd_test = "sse2"]
unsafe fn _mm_cvtsd_si64x() {
use std::{f64, i64};
let r = sse2::_mm_cvtsd_si64x(f64x2::new(f64::NAN, f64::NAN));
assert_eq!(r, i64::MIN);
}
@ -4012,6 +4040,12 @@ mod tests {
);
}
#[simd_test = "sse2"]
unsafe fn _mm_cvtsd_f64() {
let r = sse2::_mm_cvtsd_f64(f64x2::new(-1.1, 2.2));
assert_eq!(r, -1.1);
}
#[simd_test = "sse2"]
unsafe fn _mm_cvtss_sd() {
use std::{f32, f64};
@ -4058,14 +4092,18 @@ mod tests {
#[cfg(target_arch = "x86_64")]
#[simd_test = "sse2"]
unsafe fn _mm_cvttsd_si64() {
use std::{f64, i64};
let a = f64x2::new(-1.1, 2.2);
let r = sse2::_mm_cvttsd_si64(a);
assert_eq!(r, -1_i64);
}
#[cfg(target_arch = "x86_64")]
#[simd_test = "sse2"]
unsafe fn _mm_cvttsd_si64x() {
use std::{f64, i64};
let a = f64x2::new(f64::NEG_INFINITY, f64::NAN);
let r = sse2::_mm_cvttsd_si64(a);
let r = sse2::_mm_cvttsd_si64x(a);
assert_eq!(r, i64::MIN);
}