mathvec/svml_d_sincos4_core.s

   1 .macro WRAPPER_IMPL_SSE2 callee
   2         subq $40, %rsp
   3         movaps %xmm0, (%rsp)
   4         call \callee
   5         movsd %xmm0, 16(%rsp)
   6         movsd 8(%rsp), %xmm0
   7         call \callee
   8         movsd 16(%rsp), %xmm1
   9         movsd %xmm0, 24(%rsp)
  10         unpcklpd %xmm0, %xmm1
  11         movaps %xmm1, %xmm0
  12         addq $40, %rsp
  13         ret
  14 .endm
  15 .macro WRAPPER_IMPL_SSE2_ff callee
  16         subq $56, %rsp
  17         movaps %xmm0, (%rsp)
  18         movaps %xmm1, 16(%rsp)
  19         call \callee
  20         movsd %xmm0, 32(%rsp)
  21         movsd 8(%rsp), %xmm0
  22         movsd 24(%rsp), %xmm1
  23         call \callee
  24         movsd 32(%rsp), %xmm1
  25         movsd %xmm0, 40(%rsp)
  26         unpcklpd %xmm0, %xmm1
  27         movaps %xmm1, %xmm0
  28         addq $56, %rsp
  29         ret
  30 .endm
  31 .macro WRAPPER_IMPL_SSE2_fFF callee
  32         pushq %rbp
  33         pushq %rbx
  34         movq %rdi, %rbp
  35         movq %rsi, %rbx
  36         subq $40, %rsp
  37         leaq 16(%rsp), %rsi
  38         leaq 24(%rsp), %rdi
  39         movaps %xmm0, (%rsp)
  40         call \callee
  41         leaq 16(%rsp), %rsi
  42         leaq 24(%rsp), %rdi
  43         movsd 24(%rsp), %xmm0
  44         movapd (%rsp), %xmm1
  45         movsd %xmm0, 0(%rbp)
  46         unpckhpd %xmm1, %xmm1
  47         movsd 16(%rsp), %xmm0
  48         movsd %xmm0, (%rbx)
  49         movapd %xmm1, %xmm0
  50         call \callee
  51         movsd 24(%rsp), %xmm0
  52         movsd %xmm0, 8(%rbp)
  53         movsd 16(%rsp), %xmm0
  54         movsd %xmm0, 8(%rbx)
  55         addq $40, %rsp
  56         popq %rbx
  57         popq %rbp
  58         ret
  59 .endm
  60 .macro WRAPPER_IMPL_AVX callee
  61         pushq %rbp
  62         movq %rsp, %rbp
  63         andq $-32, %rsp
  64         subq $32, %rsp
  65         vextractf128 $1, %ymm0, (%rsp)
  66         vzeroupper
  67         call \callee
  68         vmovapd %xmm0, 16(%rsp)
  69         vmovaps (%rsp), %xmm0
  70         call \callee
  71         vmovapd %xmm0, %xmm1
  72         vmovapd 16(%rsp), %xmm0
  73         vinsertf128 $1, %xmm1, %ymm0, %ymm0
  74         movq %rbp, %rsp
  75         popq %rbp
  76         ret
  77 .endm
  78 .macro WRAPPER_IMPL_AVX_ff callee
  79         pushq %rbp
  80         movq %rsp, %rbp
  81         andq $-32, %rsp
  82         subq $64, %rsp
  83         vextractf128 $1, %ymm0, 16(%rsp)
  84         vextractf128 $1, %ymm1, (%rsp)
  85         vzeroupper
  86         call \callee
  87         vmovaps %xmm0, 32(%rsp)
  88         vmovaps 16(%rsp), %xmm0
  89         vmovaps (%rsp), %xmm1
  90         call \callee
  91         vmovaps %xmm0, %xmm1
  92         vmovaps 32(%rsp), %xmm0
  93         vinsertf128 $1, %xmm1, %ymm0, %ymm0
  94         movq %rbp, %rsp
  95         popq %rbp
  96         ret
  97 .endm
  98 .macro WRAPPER_IMPL_AVX_fFF callee
  99         pushq %rbp
 100         movq %rsp, %rbp
 101         andq $-32, %rsp
 102         pushq %r13
 103         pushq %r14
 104         subq $48, %rsp
 105         movq %rsi, %r14
 106         movq %rdi, %r13
 107         vextractf128 $1, %ymm0, 32(%rsp)
 108         vzeroupper
 109         call \callee
 110         vmovaps 32(%rsp), %xmm0
 111         lea (%rsp), %rdi
 112         lea 16(%rsp), %rsi
 113         call \callee
 114         vmovapd (%rsp), %xmm0
 115         vmovapd 16(%rsp), %xmm1
 116         vmovapd %xmm0, 16(%r13)
 117         vmovapd %xmm1, 16(%r14)
 118         addq $48, %rsp
 119         popq %r14
 120         popq %r13
 121         movq %rbp, %rsp
 122         popq %rbp
 123         ret
 124 .endm
 125 .macro WRAPPER_IMPL_AVX512 callee
 126         pushq %rbp
 127         movq %rsp, %rbp
 128         andq $-64, %rsp
 129         subq $128, %rsp
 130         vmovups %zmm0, (%rsp)
 131         vmovupd (%rsp), %ymm0
 132         call \callee
 133         vmovupd %ymm0, 64(%rsp)
 134         vmovupd 32(%rsp), %ymm0
 135         call \callee
 136         vmovupd %ymm0, 96(%rsp)
 137         vmovups 64(%rsp), %zmm0
 138         movq %rbp, %rsp
 139         popq %rbp
 140         ret
 141 .endm
 142 .macro WRAPPER_IMPL_AVX512_ff callee
 143         pushq %rbp
 144         movq %rsp, %rbp
 145         andq $-64, %rsp
 146         subq $192, %rsp
 147         vmovups %zmm0, (%rsp)
 148         vmovups %zmm1, 64(%rsp)
 149         vmovupd (%rsp), %ymm0
 150         vmovupd 64(%rsp), %ymm1
 151         call \callee
 152         vmovupd %ymm0, 128(%rsp)
 153         vmovupd 32(%rsp), %ymm0
 154         vmovupd 96(%rsp), %ymm1
 155         call \callee
 156         vmovupd %ymm0, 160(%rsp)
 157         vmovups 128(%rsp), %zmm0
 158         movq %rbp, %rsp
 159         popq %rbp
 160         ret
 161 .endm
 162 .macro WRAPPER_IMPL_AVX512_fFF callee
 163         pushq %rbp
 164         movq %rsp, %rbp
 165         andq $-64, %rsp
 166         pushq %r12
 167         pushq %r13
 168         subq $176, %rsp
 169         movq %rsi, %r13
 170         vmovups %zmm0, (%rsp)
 171         movq %rdi, %r12
 172         vmovupd (%rsp), %ymm0
 173         call \callee
 174         vmovupd 32(%rsp), %ymm0
 175         lea 64(%rsp), %rdi
 176         lea 96(%rsp), %rsi
 177         call \callee
 178         vmovupd 64(%rsp), %ymm0
 179         vmovupd 96(%rsp), %ymm1
 180         vmovupd %ymm0, 32(%r12)
 181         vmovupd %ymm1, 32(%r13)
 182         vzeroupper
 183         addq $176, %rsp
 184         popq %r13
 185         popq %r12
 186         movq %rbp, %rsp
 187         popq %rbp
 188         ret
 189 .endm
 190  .text
 191 .globl _ZGVdN4vl8l8_sincos
 192  .type _ZGVdN4vl8l8_sincos,@function
 193  .align 1<<4
 194  _ZGVdN4vl8l8_sincos:
 195
 196
 197 WRAPPER_IMPL_AVX_fFF _ZGVbN2vl8l8_sincos
 198  .size _ZGVdN4vl8l8_sincos,.-_ZGVdN4vl8l8_sincos
 199
 200
 201 .macro WRAPPER_IMPL_AVX2_fFF_vvv callee
 202         pushq %rbp
 203         movq %rsp, %rbp
 204         andq $-32, %rsp
 205         subq $160, %rsp
 206         vmovupd %ymm0, 128(%rsp)
 207         lea (%rsp), %rdi
 208         vmovdqu %ymm1, 64(%rdi)
 209         vmovdqu %ymm2, 96(%rdi)
 210         lea 32(%rsp), %rsi
 211         vzeroupper
 212         call \callee
 213         vmovupd 144(%rsp), %xmm0
 214         lea 16(%rsp), %rdi
 215         lea 48(%rsp), %rsi
 216         call \callee
 217         movq 64(%rsp), %rdx
 218         movq 96(%rsp), %rsi
 219         movq 72(%rsp), %r8
 220         movq 104(%rsp), %r10
 221         movq (%rsp), %rax
 222         movq 32(%rsp), %rcx
 223         movq 8(%rsp), %rdi
 224         movq 40(%rsp), %r9
 225         movq %rax, (%rdx)
 226         movq %rcx, (%rsi)
 227         movq 80(%rsp), %rax
 228         movq 112(%rsp), %rcx
 229         movq %rdi, (%r8)
 230         movq %r9, (%r10)
 231         movq 88(%rsp), %rdi
 232         movq 120(%rsp), %r9
 233         movq 16(%rsp), %r11
 234         movq 48(%rsp), %rdx
 235         movq 24(%rsp), %rsi
 236         movq 56(%rsp), %r8
 237         movq %r11, (%rax)
 238         movq %rdx, (%rcx)
 239         movq %rsi, (%rdi)
 240         movq %r8, (%r9)
 241         movq %rbp, %rsp
 242         popq %rbp
 243         ret
 244 .endm
 245 .globl _ZGVdN4vvv_sincos
 246  .type _ZGVdN4vvv_sincos,@function
 247  .align 1<<4
 248  _ZGVdN4vvv_sincos:
 249
 250
 251 WRAPPER_IMPL_AVX2_fFF_vvv _ZGVbN2vl8l8_sincos
 252  .size _ZGVdN4vvv_sincos,.-_ZGVdN4vvv_sincos
 253
 254