Add AVX broadcast and conversion intrinsics #32140
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This defines `_mm256_broadcast_ps` and `_mm256_broadcast_pd`. The `_ss` and `_sd` variants are not supported by LLVM. In Clang these intrinsics are implemented as inline functions in C++. Intel reference: https://software.intel.com/en-us/node/514144. Note: the argument type should really be "0hPc" (a pointer to a vector of half the width), but internally the LLVM intrinsic takes a pointer to a signed integer, and for any other type LLVM will complain. This means that a transmute is required to call these intrinsics. The AVX2 broadcast intrinsics `_mm256_broadcastss_ps` and `_mm256_broadcastsd_pd` are not available as LLVM intrinsics. In Clang they are implemented using the shufflevector builtin.
This defines the following intrinsics: * `_mm256_cvtepi32_pd` * `_mm256_cvtepi32_ps` * `_mm256_cvtpd_epi32` * `_mm256_cvtpd_ps` * `_mm256_cvtps_epi32` * `_mm256_cvtps_pd` * `_mm256_cvttpd_epi32` * `_mm256_cvttps_epi32` Intel reference: https://software.intel.com/en-us/node/514130.
The exact command used was:
$ cd src/etc/platform-intrinsics/x86
$ python2 ../generator.py --format compiler-defs -i info.json \
sse.json sse2.json sse3.json ssse3.json sse41.json sse42.json \
avx.json avx2.json fma.json \
> ../../../librustc_platform_intrinsics/x86.rs
|
I also have a test for these, but I am not sure where to add that. There is #![feature(repr_simd, platform_intrinsics)]
use std::mem::transmute;
#[repr(simd)]
#[derive(Debug, PartialEq)]
struct F64x2(f64, f64);
#[repr(simd)]
#[derive(Debug, PartialEq)]
struct F64x4(f64, f64, f64, f64);
#[repr(simd)]
#[derive(Debug, PartialEq)]
struct F32x4(f32, f32, f32, f32);
#[repr(simd)]
#[derive(Debug, PartialEq)]
struct F32x8(f32, f32, f32, f32, f32, f32, f32, f32);
#[repr(simd)]
#[derive(Debug, PartialEq)]
struct I32x4(i32, i32, i32, i32);
#[repr(simd)]
#[derive(Debug, PartialEq)]
struct I32x8(i32, i32, i32, i32, i32, i32, i32, i32);
extern "platform-intrinsic" {
fn x86_mm256_broadcast_pd(ptr: *const i8) -> F64x4;
fn x86_mm256_broadcast_ps(ptr: *const i8) -> F32x8;
fn x86_mm256_cvtepi32_pd(x: I32x4) -> F64x4;
fn x86_mm256_cvtepi32_ps(x: I32x8) -> F32x8;
fn x86_mm256_cvtpd_epi32(x: F64x4) -> I32x4;
fn x86_mm256_cvtpd_ps(x: F64x4) -> F32x4;
fn x86_mm256_cvtps_epi32(x: F32x8) -> I32x8;
fn x86_mm256_cvtps_pd(x: F32x4) -> F64x4;
fn x86_mm256_cvttpd_epi32(x: F64x4) -> I32x4;
fn x86_mm256_cvttps_epi32(x: F32x8) -> I32x8;
}
fn main() {
let a = F64x2(1.0, 2.0);
let b = unsafe { x86_mm256_broadcast_pd(transmute(&a)) };
let c = F64x4(1.0, 2.0, 1.0, 2.0);
assert_eq!(b, c);
let a = F32x4(1.0, 2.0, 3.0, 4.0);
let b = unsafe { x86_mm256_broadcast_ps(transmute(&a)) };
let c = F32x8(1.0, 2.0, 3.0, 4.0, 1.0, 2.0, 3.0, 4.0);
assert_eq!(b, c);
let a = I32x4(1, -2, 3, -4);
let b = unsafe { x86_mm256_cvtepi32_pd(a) };
let c = F64x4(1.0, -2.0, 3.0, -4.0);
assert_eq!(b, c);
let a = I32x8(1, -2, 3, -4, 5, -6, 7, -8);
let b = unsafe { x86_mm256_cvtepi32_ps(a) };
let c = F32x8(1.0, -2.0, 3.0, -4.0, 5.0, -6.0, 7.0, -8.0);
assert_eq!(b, c);
let a = F64x4(1.0, -2.0, 3.0, -4.0);
let b = unsafe { x86_mm256_cvtpd_epi32(a) };
let c = I32x4(1, -2, 3, -4);
assert_eq!(b, c);
let a = F64x4(1.0, -2.0, 3.0, -4.0);
let b = unsafe { x86_mm256_cvtpd_ps(a) };
let c = F32x4(1.0, -2.0, 3.0, -4.0);
assert_eq!(b, c);
let a = F32x8(1.0, -2.0, 3.0, -4.0, 5.0, -6.0, 7.0, -8.0);
let b = unsafe { x86_mm256_cvtps_epi32(a) };
let c = I32x8(1, -2, 3, -4, 5, -6, 7, -8);
assert_eq!(b, c);
let a = F32x4(1.0, -2.0, 3.0, -4.0);
let b = unsafe { x86_mm256_cvtps_pd(a) };
let c = F64x4(1.0, -2.0, 3.0, -4.0);
assert_eq!(b, c);
let a = F64x4(1.6, -2.0, 3.0, -4.0);
let b = unsafe { x86_mm256_cvttpd_epi32(a) };
let c = I32x4(1, -2, 3, -4);
assert_eq!(b, c);
let a = F32x8(1.6, -2.0, 3.0, -4.0, 5.0, -6.0, 7.0, -8.0);
let b = unsafe { x86_mm256_cvttps_epi32(a) };
let c = I32x8(1, -2, 3, -4, 5, -6, 7, -8);
assert_eq!(b, c);
} |
Manishearth
added a commit
to Manishearth/rust
that referenced
this pull request
Mar 12, 2016
…hton Add AVX broadcast and conversion intrinsics This adds the following intrinsics: * `_mm256_broadcast_pd` * `_mm256_broadcast_ps` * `_mm256_cvtepi32_pd` * `_mm256_cvtepi32_ps` * `_mm256_cvtpd_epi32` * `_mm256_cvtpd_ps` * `_mm256_cvtps_epi32` * `_mm256_cvtps_pd` * `_mm256_cvttpd_epi32` * `_mm256_cvttps_epi32` The "avx" codegen feature must be enabled to use these.
|
⌛ Testing commit c306853 with merge cfcbbf1... |
|
💔 Test failed - auto-win-gnu-32-nopt-t |
|
@bors: retry On Sat, Mar 12, 2016 at 5:23 AM, bors notifications@github.com wrote:
|
bors
added a commit
that referenced
this pull request
Mar 13, 2016
Add AVX broadcast and conversion intrinsics This adds the following intrinsics: * `_mm256_broadcast_pd` * `_mm256_broadcast_ps` * `_mm256_cvtepi32_pd` * `_mm256_cvtepi32_ps` * `_mm256_cvtpd_epi32` * `_mm256_cvtpd_ps` * `_mm256_cvtps_epi32` * `_mm256_cvtps_pd` * `_mm256_cvttpd_epi32` * `_mm256_cvttps_epi32` The "avx" codegen feature must be enabled to use these.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This adds the following intrinsics:
_mm256_broadcast_pd_mm256_broadcast_ps_mm256_cvtepi32_pd_mm256_cvtepi32_ps_mm256_cvtpd_epi32_mm256_cvtpd_ps_mm256_cvtps_epi32_mm256_cvtps_pd_mm256_cvttpd_epi32_mm256_cvttps_epi32The "avx" codegen feature must be enabled to use these.