diff options
| author | H.J. Lu <hjl.tools@gmail.com> | 2017-08-23 06:16:12 -0700 |
|---|---|---|
| committer | H.J. Lu <hjl.tools@gmail.com> | 2017-08-23 06:26:44 -0700 |
| commit | b9eaca8fa0a9628a992e0f1478aaadde576804e1 (patch) | |
| tree | 10d8dcd9696ac057b485e94e7a003d3856de71dc /sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S | |
| parent | 5a706f649de3952271930a8340db4ca8aa50f485 (diff) | |
| download | glibc-b9eaca8fa0a9628a992e0f1478aaadde576804e1.tar.gz glibc-b9eaca8fa0a9628a992e0f1478aaadde576804e1.tar.xz glibc-b9eaca8fa0a9628a992e0f1478aaadde576804e1.zip | |
x86_64: Replace AVX512F .byte sequences with instructions
Since binutils 2.25 or later is required to build glibc, we can replace AVX512F .byte sequences with AVX512F instructions. Tested on x86-64 and x32. There are no code differences in libmvec.so and libmvec.a. * sysdeps/x86_64/fpu/svml_d_sincos8_core.S: Replace AVX512F .byte sequences with AVX512F instructions. * sysdeps/x86_64/fpu/svml_d_wrapper_impl.h: Likewise. * sysdeps/x86_64/fpu/svml_s_sincosf16_core.S: Likewise. * sysdeps/x86_64/fpu/svml_s_wrapper_impl.h: Likewise. * sysdeps/x86_64/fpu/multiarch/svml_d_sincos8_core_avx512.S: Likewise. * sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S: Likewise.
Diffstat (limited to 'sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S')
| -rw-r--r-- | sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S | 63 |
1 files changed, 6 insertions, 57 deletions
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S index f73ab7de7c..8fa4255d6d 100644 --- a/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S @@ -510,40 +510,11 @@ libmvec_hidden_def(_ZGVeN16vl4l4_sincosf_skx) cfi_def_cfa_register (%rbp) andq $-64, %rsp subq $384, %rsp - /* Encoding for vmovups %zmm1, 128(%rsp). */ - .byte 0x62 - .byte 0xf1 - .byte 0x7c - .byte 0x48 - .byte 0x11 - .byte 0x4c - .byte 0x24 - .byte 0x02 + vmovups %zmm1, 128(%rsp) lea (%rsp), %rdi - /* Encoding for vmovups %zmm2, 192(%rdi). */ - .byte 0x62 - .byte 0xf1 - .byte 0x7c - .byte 0x48 - .byte 0x11 - .byte 0x57 - .byte 0x03 - /* Encoding for vmovups %zmm3, 256(%rdi). */ - .byte 0x62 - .byte 0xf1 - .byte 0x7c - .byte 0x48 - .byte 0x11 - .byte 0x5f - .byte 0x04 - /* Encoding for vmovups %zmm4, 320(%rdi). */ - .byte 0x62 - .byte 0xf1 - .byte 0x7c - .byte 0x48 - .byte 0x11 - .byte 0x67 - .byte 0x05 + vmovups %zmm2, 192(%rdi) + vmovups %zmm3, 256(%rdi) + vmovups %zmm4, 320(%rdi) lea 64(%rsp), %rsi call HIDDEN_JUMPTARGET(\callee) movq 128(%rsp), %rdx @@ -661,30 +632,8 @@ libmvec_hidden_def(_ZGVeN16vl4l4_sincosf_skx) leal -112(%rbp), %esi leal -176(%rbp), %edi subl $296, %esp - /* Encoding for vmovdqa64 %zmm1, -240(%ebp). */ - .byte 0x67 - .byte 0x62 - .byte 0xf1 - .byte 0xfd - .byte 0x48 - .byte 0x7f - .byte 0x8d - .byte 0x10 - .byte 0xff - .byte 0xff - .byte 0xff - /* Encoding for vmovdqa64 %zmm2, -304(%ebp). */ - .byte 0x67 - .byte 0x62 - .byte 0xf1 - .byte 0xfd - .byte 0x48 - .byte 0x7f - .byte 0x95 - .byte 0xd0 - .byte 0xfe - .byte 0xff - .byte 0xff + vmovdqa64 %zmm1, -240(%ebp) + vmovdqa64 %zmm2, -304(%ebp) call HIDDEN_JUMPTARGET(\callee) movl -240(%ebp), %eax vmovss -176(%ebp), %xmm0 |