Skip to content

x86: fix abs error #26

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

Open
alanzhai219 wants to merge 1 commit into
base: master
Choose a base branch
from
Open

x86: fix abs error #26

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
6 changes: 3 additions & 3 deletions src/ppl/kernel/x86/fp32/abs/abs_fp32_avx.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,8 +37,8 @@ ppl::common::RetCode abs_fp32_avx(
for (int64_t i = 0; i < unroll_body; i += unroll_n) {
__m256 src0 = _mm256_loadu_ps(x + i + 0 * V_REG_ELTS);
__m256 src1 = _mm256_loadu_ps(x + i + 1 * V_REG_ELTS);
__m256 dst0 = _mm256_andnot_ps(src0, vsignbit);
__m256 dst1 = _mm256_andnot_ps(src1, vsignbit);
__m256 dst0 = _mm256_andnot_ps(vsignbit, src0);
__m256 dst1 = _mm256_andnot_ps(vsignbit, src1);
_mm256_storeu_ps(y + i + 0 * V_REG_ELTS, dst0);
_mm256_storeu_ps(y + i + 1 * V_REG_ELTS, dst1);
}
Expand All @@ -49,4 +49,4 @@ ppl::common::RetCode abs_fp32_avx(
return ppl::common::RC_SUCCESS;
}

}}}; // namespace ppl::kernel::x86
}}}; // namespace ppl::kernel::x86
10 changes: 5 additions & 5 deletions src/ppl/kernel/x86/fp32/abs/abs_fp32_sse.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,10 +39,10 @@ ppl::common::RetCode abs_fp32_sse(
__m128 src1 = _mm_loadu_ps(x + i + 1 * V_REG_ELTS);
__m128 src2 = _mm_loadu_ps(x + i + 2 * V_REG_ELTS);
__m128 src3 = _mm_loadu_ps(x + i + 3 * V_REG_ELTS);
__m128 dst0 = _mm_andnot_ps(src0, vsignbit);
__m128 dst1 = _mm_andnot_ps(src1, vsignbit);
__m128 dst2 = _mm_andnot_ps(src2, vsignbit);
__m128 dst3 = _mm_andnot_ps(src3, vsignbit);
__m128 dst0 = _mm_andnot_ps(vsignbit, src0);
__m128 dst1 = _mm_andnot_ps(vsignbit, src1);
__m128 dst2 = _mm_andnot_ps(vsignbit, src2);
__m128 dst3 = _mm_andnot_ps(vsignbit, src3);
_mm_storeu_ps(y + i + 0 * V_REG_ELTS, dst0);
_mm_storeu_ps(y + i + 1 * V_REG_ELTS, dst1);
_mm_storeu_ps(y + i + 2 * V_REG_ELTS, dst2);
Expand All @@ -55,4 +55,4 @@ ppl::common::RetCode abs_fp32_sse(
return ppl::common::RC_SUCCESS;
}

}}}; // namespace ppl::kernel::x86
}}}; // namespace ppl::kernel::x86