Skip to content

Use intrinsics for non-temporal loads. #2187

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

Draft
ruriww wants to merge 1 commit into
base: main
Choose a base branch
from
+3 −24
Draft

Use intrinsics for non-temporal loads. #2187

Show file tree
Hide file tree
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
9 changes: 1 addition & 8 deletions crates/core_arch/src/x86/avx2.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3374,14 +3374,7 @@ pub const fn _mm256_srlv_epi64(a: __m256i, count: __m256i) -> __m256i {
#[cfg_attr(test, assert_instr(vmovntdqa))]
#[stable(feature = "simd_x86_updates", since = "1.82.0")]
pub unsafe fn _mm256_stream_load_si256(mem_addr: *const __m256i) -> __m256i {
let dst: __m256i;
crate::arch::asm!(
vpl!("vmovntdqa {a}"),
a = out(ymm_reg) dst,
p = in(reg) mem_addr,
options(pure, readonly, nostack, preserves_flags),
);
dst
intrinsics::nontemporal_load(mem_addr)
}

/// Subtract packed 16-bit integers in `b` from packed 16-bit integers in `a`
Expand Down
9 changes: 1 addition & 8 deletions crates/core_arch/src/x86/avx512f.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31071,14 +31071,7 @@ pub unsafe fn _mm512_stream_si512(mem_addr: *mut __m512i, a: __m512i) {
#[target_feature(enable = "avx512f")]
#[stable(feature = "stdarch_x86_avx512", since = "1.89")]
pub unsafe fn _mm512_stream_load_si512(mem_addr: *const __m512i) -> __m512i {
let dst: __m512i;
crate::arch::asm!(
vpl!("vmovntdqa {a}"),
a = out(zmm_reg) dst,
p = in(reg) mem_addr,
options(pure, readonly, nostack, preserves_flags),
);
dst
intrinsics::nontemporal_load(mem_addr)
}

/// Sets packed 32-bit integers in `dst` with the supplied values.
Expand Down
9 changes: 1 addition & 8 deletions crates/core_arch/src/x86/sse41.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1170,14 +1170,7 @@ pub fn _mm_test_mix_ones_zeros(a: __m128i, mask: __m128i) -> i32 {
#[cfg_attr(test, assert_instr(movntdqa))]
#[stable(feature = "simd_x86_updates", since = "1.82.0")]
pub unsafe fn _mm_stream_load_si128(mem_addr: *const __m128i) -> __m128i {
let dst: __m128i;
crate::arch::asm!(
vpl!("movntdqa {a}"),
a = out(xmm_reg) dst,
p = in(reg) mem_addr,
options(pure, readonly, nostack, preserves_flags),
);
dst
intrinsics::nontemporal_load(mem_addr)
}

#[allow(improper_ctypes)]
Expand Down
Loading