144 lines
4.5 KiB
ArmAsm
144 lines
4.5 KiB
ArmAsm
/* Function sincos vectorized in AVX ISA as wrapper to SSE4 ISA version.
|
|
Copyright (C) 2014-2019 Free Software Foundation, Inc.
|
|
This file is part of the GNU C Library.
|
|
|
|
The GNU C Library is free software; you can redistribute it and/or
|
|
modify it under the terms of the GNU Lesser General Public
|
|
License as published by the Free Software Foundation; either
|
|
version 2.1 of the License, or (at your option) any later version.
|
|
|
|
The GNU C Library is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
Lesser General Public License for more details.
|
|
|
|
You should have received a copy of the GNU Lesser General Public
|
|
License along with the GNU C Library; if not, see
|
|
<http://www.gnu.org/licenses/>. */
|
|
|
|
#include <sysdep.h>
|
|
#include "svml_d_wrapper_impl.h"
|
|
|
|
.text
|
|
ENTRY (_ZGVcN4vl8l8_sincos)
|
|
WRAPPER_IMPL_AVX_fFF _ZGVbN2vl8l8_sincos
|
|
END (_ZGVcN4vl8l8_sincos)
|
|
|
|
/* 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 $160, %rsp
|
|
vmovupd %ymm0, 64(%rsp)
|
|
lea (%rsp), %rdi
|
|
vmovdqu %xmm1, 96(%rdi)
|
|
vmovdqu %xmm2, 112(%rdi)
|
|
vmovdqu %xmm3, 128(%rdi)
|
|
vmovdqu %xmm4, 144(%rdi)
|
|
lea 32(%rsp), %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
|
|
movq (%rsp), %rax
|
|
movq 8(%rsp), %rcx
|
|
movq 16(%rsp), %rdi
|
|
movq 24(%rsp), %r9
|
|
movq %rax, (%rdx)
|
|
movq %rcx, (%rsi)
|
|
movq 128(%rsp), %rax
|
|
movq 136(%rsp), %rcx
|
|
movq %rdi, (%r8)
|
|
movq %r9, (%r10)
|
|
movq 144(%rsp), %rdi
|
|
movq 152(%rsp), %r9
|
|
movq 32(%rsp), %r11
|
|
movq 40(%rsp), %rdx
|
|
movq 48(%rsp), %rsi
|
|
movq 56(%rsp), %r8
|
|
movq %r11, (%rax)
|
|
movq %rdx, (%rcx)
|
|
movq %rsi, (%rdi)
|
|
movq %r8, (%r9)
|
|
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 $152, %esp
|
|
vmovaps %xmm1, -128(%ebp)
|
|
vmovaps %xmm2, -144(%ebp)
|
|
vmovapd %ymm0, -176(%ebp)
|
|
vzeroupper
|
|
call HIDDEN_JUMPTARGET(\callee)
|
|
leal 16(%r12), %esi
|
|
vmovupd -160(%ebp), %xmm0
|
|
leal 16(%rbx), %edi
|
|
call HIDDEN_JUMPTARGET(\callee)
|
|
movq -128(%ebp), %rax
|
|
vmovsd -112(%ebp), %xmm0
|
|
vmovdqa -128(%ebp), %xmm5
|
|
vmovdqa -144(%ebp), %xmm1
|
|
vmovsd %xmm0, (%eax)
|
|
vmovsd -104(%ebp), %xmm0
|
|
vpextrd $1, %xmm5, %eax
|
|
vmovsd %xmm0, (%eax)
|
|
movq -120(%ebp), %rax
|
|
vmovsd -96(%ebp), %xmm0
|
|
vmovsd %xmm0, (%eax)
|
|
vmovsd -88(%ebp), %xmm0
|
|
vpextrd $3, %xmm5, %eax
|
|
vmovsd %xmm0, (%eax)
|
|
movq -144(%ebp), %rax
|
|
vmovsd -80(%ebp), %xmm0
|
|
vmovsd %xmm0, (%eax)
|
|
vmovsd -72(%ebp), %xmm0
|
|
vpextrd $1, %xmm1, %eax
|
|
vmovsd %xmm0, (%eax)
|
|
movq -136(%ebp), %rax
|
|
vmovsd -64(%ebp), %xmm0
|
|
vmovsd %xmm0, (%eax)
|
|
vmovsd -56(%ebp), %xmm0
|
|
vpextrd $3, %xmm1, %eax
|
|
vmovsd %xmm0, (%eax)
|
|
addl $152, %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 (_ZGVcN4vvv_sincos)
|
|
WRAPPER_IMPL_AVX_fFF_vvv _ZGVbN2vl8l8_sincos
|
|
END (_ZGVcN4vvv_sincos)
|