diff options
Diffstat (limited to 'sysdeps/x86_64/fpu/svml_s_sincosf8_core_avx.S')
| -rw-r--r-- | sysdeps/x86_64/fpu/svml_s_sincosf8_core_avx.S | 179 |
1 files changed, 176 insertions, 3 deletions
diff --git a/sysdeps/x86_64/fpu/svml_s_sincosf8_core_avx.S b/sysdeps/x86_64/fpu/svml_s_sincosf8_core_avx.S index 55b8b2d768..cd88195ee7 100644 --- a/sysdeps/x86_64/fpu/svml_s_sincosf8_core_avx.S +++ b/sysdeps/x86_64/fpu/svml_s_sincosf8_core_avx.S @@ -20,6 +20,179 @@ #include "svml_s_wrapper_impl.h" .text -ENTRY(_ZGVcN8vvv_sincosf) -WRAPPER_IMPL_AVX_fFF _ZGVbN4vvv_sincosf -END(_ZGVcN8vvv_sincosf) +ENTRY (_ZGVcN8vl4l4_sincosf) +WRAPPER_IMPL_AVX_fFF _ZGVbN4vl4l4_sincosf +END (_ZGVcN8vl4l4_sincosf) + +/* AVX ISA version as wrapper to SSE ISA version (for vector + function declared with #pragma omp declare simd notinbranch). */ +.macro WRAPPER_IMPL_AVX_fFF_vvv callee +#ifndef __ILP32__ + pushq %rbp + movq %rsp, %rbp + andq $-32, %rsp + subq $224, %rsp + vmovups %ymm0, 64(%rsp) + lea (%rsp), %rdi + vmovdqu %xmm1, 96(%rdi) + vmovdqu %xmm2, 112(%rdi) + vmovdqu %xmm3, 128(%rdi) + vmovdqu %xmm4, 144(%rdi) + vmovdqu %xmm5, 160(%rdi) + lea 32(%rsp), %rsi + vmovdqu %xmm6, 144(%rsi) + vmovdqu %xmm7, 160(%rsi) + vzeroupper + call HIDDEN_JUMPTARGET(\callee) + vmovdqu 80(%rsp), %xmm0 + lea 16(%rsp), %rdi + lea 48(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + movq 96(%rsp), %rdx + movq 104(%rsp), %rsi + movq 112(%rsp), %r8 + movq 120(%rsp), %r10 + movl (%rsp), %eax + movl 4(%rsp), %ecx + movl 8(%rsp), %edi + movl 12(%rsp), %r9d + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 128(%rsp), %rax + movq 136(%rsp), %rcx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 144(%rsp), %rdi + movq 152(%rsp), %r9 + movl 16(%rsp), %r11d + movl 20(%rsp), %edx + movl 24(%rsp), %esi + movl 28(%rsp), %r8d + movl %r11d, (%rax) + movl %edx, (%rcx) + movq 160(%rsp), %r11 + movq 168(%rsp), %rdx + movl %esi, (%rdi) + movl %r8d, (%r9) + movq 176(%rsp), %rsi + movq 184(%rsp), %r8 + movl 32(%rsp), %r10d + movl 36(%rsp), %eax + movl 40(%rsp), %ecx + movl 44(%rsp), %edi + movl %r10d, (%r11) + movl %eax, (%rdx) + movq 192(%rsp), %r10 + movq 200(%rsp), %rax + movl %ecx, (%rsi) + movl %edi, (%r8) + movq 16(%rbp), %rcx + movq 24(%rbp), %rdi + movl 48(%rsp), %r9d + movl 52(%rsp), %r11d + movl 56(%rsp), %edx + movl 60(%rsp), %esi + movl %r9d, (%r10) + movl %r11d, (%rax) + movl %edx, (%rcx) + movl %esi, (%rdi) + movq %rbp, %rsp + popq %rbp + ret +#else + leal 8(%rsp), %r10d + .cfi_def_cfa 10, 0 + andl $-32, %esp + pushq -8(%r10d) + pushq %rbp + .cfi_escape 0x10,0x6,0x2,0x76,0 + movl %esp, %ebp + pushq %r12 + leal -80(%rbp), %esi + pushq %r10 + .cfi_escape 0xf,0x3,0x76,0x70,0x6 + .cfi_escape 0x10,0xc,0x2,0x76,0x78 + leal -112(%rbp), %edi + movq %rsi, %r12 + pushq %rbx + .cfi_escape 0x10,0x3,0x2,0x76,0x68 + movq %rdi, %rbx + subl $184, %esp + vmovaps %xmm1, -128(%ebp) + vmovaps %xmm2, -144(%ebp) + vmovaps %xmm3, -160(%ebp) + vmovaps %xmm4, -176(%ebp) + vmovaps %ymm0, -208(%ebp) + vzeroupper + call HIDDEN_JUMPTARGET(\callee) + leal 16(%r12), %esi + vmovups -192(%ebp), %xmm0 + leal 16(%rbx), %edi + call HIDDEN_JUMPTARGET(\callee) + movq -128(%ebp), %rax + vmovss -112(%ebp), %xmm0 + vmovdqa -128(%ebp), %xmm7 + vmovdqa -144(%ebp), %xmm3 + vmovss %xmm0, (%eax) + vmovss -108(%ebp), %xmm0 + vpextrd $1, %xmm7, %eax + vmovss %xmm0, (%eax) + movq -120(%ebp), %rax + vmovss -104(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -100(%ebp), %xmm0 + vpextrd $3, %xmm7, %eax + vmovdqa -160(%ebp), %xmm7 + vmovss %xmm0, (%eax) + movq -144(%ebp), %rax + vmovss -96(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -92(%ebp), %xmm0 + vpextrd $1, %xmm3, %eax + vmovss %xmm0, (%eax) + movq -136(%ebp), %rax + vmovss -88(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -84(%ebp), %xmm0 + vpextrd $3, %xmm3, %eax + vmovss %xmm0, (%eax) + movq -160(%ebp), %rax + vmovss -80(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -76(%ebp), %xmm0 + vpextrd $1, %xmm7, %eax + vmovss %xmm0, (%eax) + movq -152(%ebp), %rax + vmovss -72(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -68(%ebp), %xmm0 + vpextrd $3, %xmm7, %eax + vmovss %xmm0, (%eax) + movq -176(%ebp), %rax + vmovss -64(%ebp), %xmm0 + vmovdqa -176(%ebp), %xmm3 + vmovss %xmm0, (%eax) + vmovss -60(%ebp), %xmm0 + vpextrd $1, %xmm3, %eax + vmovss %xmm0, (%eax) + movq -168(%ebp), %rax + vmovss -56(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -52(%ebp), %xmm0 + vpextrd $3, %xmm3, %eax + vmovss %xmm0, (%eax) + addl $184, %esp + popq %rbx + popq %r10 + .cfi_def_cfa 10, 0 + popq %r12 + popq %rbp + leal -8(%r10), %esp + .cfi_def_cfa 7, 8 + ret +#endif +.endm + +ENTRY (_ZGVcN8vvv_sincosf) +WRAPPER_IMPL_AVX_fFF_vvv _ZGVbN4vl4l4_sincosf +END (_ZGVcN8vvv_sincosf) |