diff options
| author | Andrew Senkevich <andrew.senkevich@intel.com> | 2016-10-17 17:55:28 +0300 |
|---|---|---|
| committer | Mike Frysinger <vapier@gentoo.org> | 2016-11-12 00:44:39 -0500 |
| commit | 92deb83907f771dc38614628c60309bb7ea84511 (patch) | |
| tree | beb16530b662fb42ba471f12b53610603bc9f7df | |
| parent | 402a6602288ef72d0f9061bef42c7a9235c9dd07 (diff) | |
| download | glibc-92deb83907f771dc38614628c60309bb7ea84511.tar.gz glibc-92deb83907f771dc38614628c60309bb7ea84511.tar.xz glibc-92deb83907f771dc38614628c60309bb7ea84511.zip | |
Fixed x86_64 vector sincos/sincosf ABI.
Fixed wrong vector sincos/sincosf ABI to have it compatible with current vector function declaration "#pragma omp declare simd notinbranch", according to which vector sincos should have vector of pointers for second and third parameters. It is fixed with implementation as wrapper to version having second and third parameters as pointers. (cherry-picked from commit ee2196bb6766ca7e63a1ba22ebb7619a3266776a) (cherry picked from commit 0350824fabf12da92c3ed5e1f3938bc1453a5938)
32 files changed, 2545 insertions, 33 deletions
diff --git a/sysdeps/x86/fpu/test-math-vector-sincos.h b/sysdeps/x86/fpu/test-math-vector-sincos.h new file mode 100644 index 0000000000..0263fc5935 --- /dev/null +++ b/sysdeps/x86/fpu/test-math-vector-sincos.h @@ -0,0 +1,98 @@ +/* Wrappers definitions for tests of ABI of vector sincos/sincosf having + vector declaration "#pragma omp declare simd notinbranch". + Copyright (C) 2016 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/>. */ + +#define INIT_VEC_PTRS_LOOP(vec, val, len) \ + do \ + { \ + for (i = 0; i < len; i++) \ + { \ + vec[i] = &val[i]; \ + } \ + } \ + while (0) + +/* Wrapper for vector sincos/sincosf compatible with x86_64 and x32 variants + of _ZGVbN2vvv_sincos, _ZGVdN4vvv_sincos, _ZGVeN8vvv_sincos; + x32 variants of _ZGVbN4vvv_sincosf, _ZGVcN4vvv_sincos, _ZGVdN8vvv_sincosf, + _ZGVeN16vvv_sincosf. */ +#define VECTOR_WRAPPER_fFF_2(scalar_func, vector_func) \ +extern void vector_func (VEC_TYPE, VEC_INT_TYPE, VEC_INT_TYPE); \ +void scalar_func (FLOAT x, FLOAT * r, FLOAT * r1) \ +{ \ + int i; \ + FLOAT r_loc[VEC_LEN], r1_loc[VEC_LEN]; \ + VEC_TYPE mx; \ + VEC_INT_TYPE mr, mr1; \ + INIT_VEC_LOOP (mx, x, VEC_LEN); \ + INIT_VEC_PTRS_LOOP (((FLOAT **) &mr), r_loc, VEC_LEN); \ + INIT_VEC_PTRS_LOOP (((FLOAT **) &mr1), r1_loc, VEC_LEN); \ + vector_func (mx, mr, mr1); \ + TEST_VEC_LOOP (r_loc, VEC_LEN); \ + TEST_VEC_LOOP (r1_loc, VEC_LEN); \ + *r = r_loc[0]; \ + *r1 = r1_loc[0]; \ + return; \ +} + +/* Wrapper for vector sincos/sincosf compatible with x86_64 variants of + _ZGVcN4vvv_sincos, _ZGVeN16vvv_sincosf, _ZGVbN4vvv_sincosf, + _ZGVdN8vvv_sincosf, _ZGVcN8vvv_sincosf. */ +#define VECTOR_WRAPPER_fFF_3(scalar_func, vector_func) \ +extern void vector_func (VEC_TYPE, VEC_INT_TYPE, VEC_INT_TYPE, \ + VEC_INT_TYPE, VEC_INT_TYPE); \ +void scalar_func (FLOAT x, FLOAT * r, FLOAT * r1) \ +{ \ + int i; \ + FLOAT r_loc[VEC_LEN/2], r1_loc[VEC_LEN/2]; \ + VEC_TYPE mx; \ + VEC_INT_TYPE mr, mr1; \ + INIT_VEC_LOOP (mx, x, VEC_LEN); \ + INIT_VEC_PTRS_LOOP (((FLOAT **) &mr), r_loc, VEC_LEN/2); \ + INIT_VEC_PTRS_LOOP (((FLOAT **) &mr1), r1_loc, VEC_LEN/2); \ + vector_func (mx, mr, mr, mr1, mr1); \ + TEST_VEC_LOOP (r_loc, VEC_LEN/2); \ + TEST_VEC_LOOP (r1_loc, VEC_LEN/2); \ + *r = r_loc[0]; \ + *r1 = r1_loc[0]; \ + return; \ +} + +/* Wrapper for vector sincosf compatible with x86_64 variant of + _ZGVcN8vvv_sincosf. */ +#define VECTOR_WRAPPER_fFF_4(scalar_func, vector_func) \ +extern void vector_func (VEC_TYPE, VEC_INT_TYPE, VEC_INT_TYPE, \ + VEC_INT_TYPE, VEC_INT_TYPE, \ + VEC_INT_TYPE, VEC_INT_TYPE, \ + VEC_INT_TYPE, VEC_INT_TYPE); \ +void scalar_func (FLOAT x, FLOAT * r, FLOAT * r1) \ +{ \ + int i; \ + FLOAT r_loc[VEC_LEN/4], r1_loc[VEC_LEN/4]; \ + VEC_TYPE mx; \ + VEC_INT_TYPE mr, mr1; \ + INIT_VEC_LOOP (mx, x, VEC_LEN); \ + INIT_VEC_PTRS_LOOP (((FLOAT **) &mr), r_loc, VEC_LEN/4); \ + INIT_VEC_PTRS_LOOP (((FLOAT **) &mr1), r1_loc, VEC_LEN/4); \ + vector_func (mx, mr, mr, mr, mr, mr1, mr1, mr1, mr1); \ + TEST_VEC_LOOP (r_loc, VEC_LEN/4); \ + TEST_VEC_LOOP (r1_loc, VEC_LEN/4); \ + *r = r_loc[0]; \ + *r1 = r1_loc[0]; \ + return; \ +} diff --git a/sysdeps/x86_64/fpu/Makefile b/sysdeps/x86_64/fpu/Makefile index 88742faff1..83a219fc60 100644 --- a/sysdeps/x86_64/fpu/Makefile +++ b/sysdeps/x86_64/fpu/Makefile @@ -30,9 +30,37 @@ ifeq ($(subdir),math) ifeq ($(build-mathvec),yes) libmvec-tests += double-vlen2 double-vlen4 double-vlen4-avx2 \ float-vlen4 float-vlen8 float-vlen8-avx2 +tests += test-double-libmvec-sincos test-double-libmvec-sincos-avx \ + test-double-libmvec-sincos-avx2 test-float-libmvec-sincosf \ + test-float-libmvec-sincosf-avx test-float-libmvec-sincosf-avx2 + +$(objpfx)test-double-libmvec-sincos: \ + $(objpfx)test-double-libmvec-sincos.o $(libmvec) + +$(objpfx)test-double-libmvec-sincos-avx: \ + $(objpfx)test-double-libmvec-sincos-avx.o $(libmvec) + +$(objpfx)test-double-libmvec-sincos-avx2: \ + $(objpfx)test-double-libmvec-sincos-avx2.o $(libmvec) + +$(objpfx)test-float-libmvec-sincosf: \ + $(objpfx)test-float-libmvec-sincosf.o $(libmvec) + +$(objpfx)test-float-libmvec-sincosf-avx: \ + $(objpfx)test-float-libmvec-sincosf-avx.o $(libmvec) + +$(objpfx)test-float-libmvec-sincosf-avx2: \ + $(objpfx)test-float-libmvec-sincosf-avx2.o $(libmvec) ifeq (yes,$(config-cflags-avx512)) libmvec-tests += double-vlen8 float-vlen16 +tests += test-double-libmvec-sincos-avx512 test-float-libmvec-sincosf-avx512 + +$(objpfx)test-double-libmvec-sincos-avx512: \ + $(objpfx)test-double-libmvec-sincos-avx512.o $(libmvec) + +$(objpfx)test-float-libmvec-sincosf-avx512: \ + $(objpfx)test-float-libmvec-sincosf-avx512.o $(libmvec) endif double-vlen4-arch-ext-cflags = -mavx @@ -43,11 +71,22 @@ float-vlen8-arch-ext-cflags = -mavx float-vlen8-arch-ext2-cflags = -mavx2 float-vlen16-arch-ext-cflags = -mavx512f +libmvec-sincos-cflags = $(libm-test-fast-math-cflags) -fno-inline -fopenmp -Wno-unknown-pragmas + CFLAGS-test-double-vlen4-avx2.c = $(libm-test-vec-cflags) CFLAGS-test-double-vlen4-avx2-wrappers.c = $(double-vlen4-arch-ext2-cflags) CFLAGS-test-float-vlen8-avx2.c = $(libm-test-vec-cflags) CFLAGS-test-float-vlen8-avx2-wrappers.c = $(float-vlen8-arch-ext2-cflags) +CFLAGS-test-double-libmvec-sincos.c = $(libmvec-sincos-cflags) +CFLAGS-test-double-libmvec-sincos-avx.c = $(libmvec-sincos-cflags) $(double-vlen4-arch-ext-cflags) -DREQUIRE_AVX +CFLAGS-test-double-libmvec-sincos-avx2.c = $(libmvec-sincos-cflags) $(double-vlen4-arch-ext2-cflags) -DREQUIRE_AVX2 +CFLAGS-test-double-libmvec-sincos-avx512.c = $(libmvec-sincos-cflags) $(double-vlen8-arch-ext-cflags) -DREQUIRE_AVX512F + +CFLAGS-test-float-libmvec-sincosf.c = $(libmvec-sincos-cflags) +CFLAGS-test-float-libmvec-sincosf-avx.c = $(libmvec-sincos-cflags) $(float-vlen8-arch-ext-cflags) -DREQUIRE_AVX +CFLAGS-test-float-libmvec-sincosf-avx2.c = $(libmvec-sincos-cflags) $(float-vlen8-arch-ext2-cflags) -DREQUIRE_AVX2 +CFLAGS-test-float-libmvec-sincosf-avx512.c = $(libmvec-sincos-cflags) $(float-vlen16-arch-ext-cflags) -DREQUIRE_AVX512F endif endif diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_sincos2_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_d_sincos2_core_sse4.S index 65ad540122..ab701a3a5f 100644 --- a/sysdeps/x86_64/fpu/multiarch/svml_d_sincos2_core_sse4.S +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_sincos2_core_sse4.S @@ -20,7 +20,7 @@ #include "svml_d_trig_data.h" .text -ENTRY (_ZGVbN2vvv_sincos_sse4) +ENTRY (_ZGVbN2vl8l8_sincos_sse4) /* ALGORITHM DESCRIPTION: @@ -311,4 +311,58 @@ ENTRY (_ZGVbN2vvv_sincos_sse4) movsd %xmm0, 256(%rsp,%r15) jmp .LBL_1_7 +END (_ZGVbN2vl8l8_sincos_sse4) +libmvec_hidden_def(_ZGVbN2vl8l8_sincos_sse4) + +/* vvv version implemented with wrapper to vl8l8 variant. */ +ENTRY (_ZGVbN2vvv_sincos_sse4) +#ifndef __ILP32__ + subq $72, %rsp + .cfi_def_cfa_offset 80 + movdqu %xmm1, 32(%rsp) + lea (%rsp), %rdi + movdqu %xmm2, 48(%rdi) + lea 16(%rsp), %rsi + call HIDDEN_JUMPTARGET(_ZGVbN2vl8l8_sincos_sse4) + movq 32(%rsp), %rdx + movq 48(%rsp), %rsi + movq 40(%rsp), %r8 + movq 56(%rsp), %r10 + movq (%rsp), %rax + movq 16(%rsp), %rcx + movq 8(%rsp), %rdi + movq 24(%rsp), %r9 + movq %rax, (%rdx) + movq %rcx, (%rsi) + movq %rdi, (%r8) + movq %r9, (%r10) + addq $72, %rsp + .cfi_def_cfa_offset 8 + ret +#else + subl $72, %esp + .cfi_def_cfa_offset 80 + leal 48(%rsp), %esi + movaps %xmm1, 16(%esp) + leal 32(%rsp), %edi + movaps %xmm2, (%esp) + call HIDDEN_JUMPTARGET(_ZGVbN2vl8l8_sincos_sse4) + movdqa 16(%esp), %xmm1 + movsd 32(%esp), %xmm0 + movq %xmm1, %rax + movdqa (%esp), %xmm2 + movsd %xmm0, (%eax) + movsd 40(%esp), %xmm0 + pextrd $1, %xmm1, %eax + movsd %xmm0, (%eax) + movsd 48(%esp), %xmm0 + movq %xmm2, %rax + movsd %xmm0, (%eax) + movsd 56(%esp), %xmm0 + pextrd $1, %xmm2, %eax + movsd %xmm0, (%eax) + addl $72, %esp + .cfi_def_cfa_offset 8 + ret +#endif END (_ZGVbN2vvv_sincos_sse4) diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_sincos4_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_sincos4_core_avx2.S index 60d03e9f8b..f534afbb67 100644 --- a/sysdeps/x86_64/fpu/multiarch/svml_d_sincos4_core_avx2.S +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_sincos4_core_avx2.S @@ -20,7 +20,7 @@ #include "svml_d_trig_data.h" .text -ENTRY (_ZGVdN4vvv_sincos_avx2) +ENTRY (_ZGVdN4vl8l8_sincos_avx2) /* ALGORITHM DESCRIPTION: @@ -274,4 +274,100 @@ ENTRY (_ZGVdN4vvv_sincos_avx2) vmovsd %xmm0, 384(%rsp,%r15) jmp .LBL_1_7 +END (_ZGVdN4vl8l8_sincos_avx2) +libmvec_hidden_def(_ZGVdN4vl8l8_sincos_avx2) + +/* vvv version implemented with wrapper to vl8l8 variant. */ +ENTRY (_ZGVdN4vvv_sincos_avx2) +#ifndef __ILP32__ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-32, %rsp + subq $128, %rsp + vmovdqu %ymm1, 64(%rsp) + lea (%rsp), %rdi + vmovdqu %ymm2, 96(%rdi) + lea 32(%rsp), %rsi + call HIDDEN_JUMPTARGET(_ZGVdN4vl8l8_sincos_avx2) + movq 64(%rsp), %rdx + movq 96(%rsp), %rsi + movq 72(%rsp), %r8 + movq 104(%rsp), %r10 + movq (%rsp), %rax + movq 32(%rsp), %rcx + movq 8(%rsp), %rdi + movq 40(%rsp), %r9 + movq %rax, (%rdx) + movq %rcx, (%rsi) + movq 80(%rsp), %rax + movq 112(%rsp), %rcx + movq %rdi, (%r8) + movq %r9, (%r10) + movq 88(%rsp), %rdi + movq 120(%rsp), %r9 + movq 16(%rsp), %r11 + movq 48(%rsp), %rdx + movq 24(%rsp), %rsi + movq 56(%rsp), %r8 + movq %r11, (%rax) + movq %rdx, (%rcx) + movq %rsi, (%rdi) + movq %r8, (%r9) + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%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 %r10 + .cfi_escape 0xf,0x3,0x76,0x78,0x6 + leal -48(%rbp), %esi + leal -80(%rbp), %edi + subl $104, %esp + vmovaps %xmm1, -96(%ebp) + vmovaps %xmm2, -112(%ebp) + call HIDDEN_JUMPTARGET(_ZGVdN4vl8l8_sincos_avx2) + movl -96(%ebp), %eax + vmovsd -80(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -92(%ebp), %eax + vmovsd -72(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -88(%ebp), %eax + vmovsd -64(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -84(%ebp), %eax + vmovsd -56(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -112(%ebp), %eax + vmovsd -48(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -108(%ebp), %eax + vmovsd -40(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -104(%ebp), %eax + vmovsd -32(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -100(%ebp), %eax + vmovsd -24(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + addl $104, %esp + popq %r10 + .cfi_def_cfa 10, 0 + popq %rbp + leal -8(%r10), %esp + .cfi_def_cfa 7, 8 + ret +#endif END (_ZGVdN4vvv_sincos_avx2) diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_sincos8_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_d_sincos8_core_avx512.S index 44700f90b8..2ff217be8d 100644 --- a/sysdeps/x86_64/fpu/multiarch/svml_d_sincos8_core_avx512.S +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_sincos8_core_avx512.S @@ -36,9 +36,9 @@ sin(R), sin(R') are approximated by corresponding polynomial. */ .text -ENTRY (_ZGVeN8vvv_sincos_knl) +ENTRY (_ZGVeN8vl8l8_sincos_knl) #ifndef HAVE_AVX512_ASM_SUPPORT -WRAPPER_IMPL_AVX512_fFF _ZGVdN4vvv_sincos +WRAPPER_IMPL_AVX512_fFF _ZGVdN4vl8l8_sincos #else pushq %rbp cfi_adjust_cfa_offset (8) @@ -304,11 +304,12 @@ WRAPPER_IMPL_AVX512_fFF _ZGVdN4vvv_sincos jmp .LBL_1_7 #endif -END (_ZGVeN8vvv_sincos_knl) +END (_ZGVeN8vl8l8_sincos_knl) +libmvec_hidden_def(_ZGVeN8vl8l8_sincos_knl) -ENTRY (_ZGVeN8vvv_sincos_skx) +ENTRY (_ZGVeN8vl8l8_sincos_skx) #ifndef HAVE_AVX512_ASM_SUPPORT -WRAPPER_IMPL_AVX512_fFF _ZGVdN4vvv_sincos +WRAPPER_IMPL_AVX512_fFF _ZGVdN4vl8l8_sincos #else pushq %rbp cfi_adjust_cfa_offset (8) @@ -585,6 +586,175 @@ WRAPPER_IMPL_AVX512_fFF _ZGVdN4vvv_sincos jmp .LBL_2_7 #endif +END (_ZGVeN8vl8l8_sincos_skx) +libmvec_hidden_def(_ZGVeN8vl8l8_sincos_skx) + +/* Wrapper between vvv and vl8l8 vector variants. */ +.macro WRAPPER_AVX512_vvv_vl8l8 callee +#ifndef __ILP32__ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-64, %rsp + subq $256, %rsp + /* Encoding for vmovups %zmm1, 128(%rsp). */ + .byte 0x62 + .byte 0xf1 + .byte 0x7c + .byte 0x48 + .byte 0x11 + .byte 0x4c + .byte 0x24 + .byte 0x02 + lea (%rsp), %rdi + /* Encoding for vmovups %zmm2, 192(%rdi). */ + .byte 0x62 + .byte 0xf1 + .byte 0x7c + .byte 0x48 + .byte 0x11 + .byte 0x57 + .byte 0x03 + lea 64(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + movq 128(%rsp), %rdx + movq 136(%rsp), %rsi + movq 144(%rsp), %r8 + movq 152(%rsp), %r10 + movq (%rsp), %rax + movq 8(%rsp), %rcx + movq 16(%rsp), %rdi + movq 24(%rsp), %r9 + movq %rax, (%rdx) + movq %rcx, (%rsi) + movq 160(%rsp), %rax + movq 168(%rsp), %rcx + movq %rdi, (%r8) + movq %r9, (%r10) + movq 176(%rsp), %rdi + movq 184(%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 192(%rsp), %r11 + movq 200(%rsp), %rdx + movq %rsi, (%rdi) + movq %r8, (%r9) + movq 208(%rsp), %rsi + movq 216(%rsp), %r8 + movq 64(%rsp), %r10 + movq 72(%rsp), %rax + movq 80(%rsp), %rcx + movq 88(%rsp), %rdi + movq %r10, (%r11) + movq %rax, (%rdx) + movq 224(%rsp), %r10 + movq 232(%rsp), %rax + movq %rcx, (%rsi) + movq %rdi, (%r8) + movq 240(%rsp), %rcx + movq 248(%rsp), %rdi + movq 96(%rsp), %r9 + movq 104(%rsp), %r11 + movq 112(%rsp), %rdx + movq 120(%rsp), %rsi + movq %r9, (%r10) + movq %r11, (%rax) + movq %rdx, (%rcx) + movq %rsi, (%rdi) + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%rbp) + ret +#else + leal 8(%rsp), %r10d + .cfi_def_cfa 10, 0 + andl $-64, %esp + pushq -8(%r10d) + pushq %rbp + .cfi_escape 0x10,0x6,0x2,0x76,0 + movl %esp, %ebp + pushq %r10 + .cfi_escape 0xf,0x3,0x76,0x78,0x6 + leal -112(%rbp), %esi + leal -176(%rbp), %edi + subl $232, %esp + vmovdqa %ymm1, -208(%ebp) + vmovdqa %ymm2, -240(%ebp) + call HIDDEN_JUMPTARGET(\callee) + vmovdqa -208(%ebp), %xmm0 + vmovq %xmm0, %rax + vmovsd -176(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + shrq $32, %rax + vmovsd -168(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movq -200(%ebp), %rax + vmovsd -160(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + shrq $32, %rax + vmovsd -152(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movq -192(%ebp), %rax + vmovsd -144(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + shrq $32, %rax + vmovsd -136(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movq -184(%ebp), %rax + vmovsd -128(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + shrq $32, %rax + vmovsd -120(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + vmovdqa -240(%ebp), %xmm0 + vmovq %xmm0, %rax + vmovsd -112(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + shrq $32, %rax + vmovsd -104(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movq -232(%ebp), %rax + vmovsd -96(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + shrq $32, %rax + vmovsd -88(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movq -224(%ebp), %rax + vmovsd -80(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + shrq $32, %rax + vmovsd -72(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movq -216(%ebp), %rax + vmovsd -64(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + shrq $32, %rax + vmovsd -56(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + addl $232, %esp + popq %r10 + .cfi_def_cfa 10, 0 + popq %rbp + leal -8(%r10), %esp + .cfi_def_cfa 7, 8 + ret +#endif +.endm + +ENTRY (_ZGVeN8vvv_sincos_knl) +WRAPPER_AVX512_vvv_vl8l8 _ZGVeN8vl8l8_sincos_knl +END (_ZGVeN8vvv_sincos_knl) + +ENTRY (_ZGVeN8vvv_sincos_skx) +WRAPPER_AVX512_vvv_vl8l8 _ZGVeN8vl8l8_sincos_skx END (_ZGVeN8vvv_sincos_skx) .section .rodata, "a" 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 758aeeaeed..f6ed609b5f 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 @@ -49,9 +49,9 @@ R2 = XOR( RC, SC ). */ .text -ENTRY (_ZGVeN16vvv_sincosf_knl) +ENTRY (_ZGVeN16vl4l4_sincosf_knl) #ifndef HAVE_AVX512_ASM_SUPPORT -WRAPPER_IMPL_AVX512_fFF _ZGVdN8vvv_sincosf +WRAPPER_IMPL_AVX512_fFF _ZGVdN8vl4l4_sincosf #else pushq %rbp cfi_adjust_cfa_offset (8) @@ -267,9 +267,10 @@ WRAPPER_IMPL_AVX512_fFF _ZGVdN8vvv_sincosf vmovss %xmm0, 1280(%rsp,%r15,8) jmp .LBL_1_7 #endif -END (_ZGVeN16vvv_sincosf_knl) +END (_ZGVeN16vl4l4_sincosf_knl) +libmvec_hidden_def(_ZGVeN16vl4l4_sincosf_knl) -ENTRY (_ZGVeN16vvv_sincosf_skx) +ENTRY (_ZGVeN16vl4l4_sincosf_skx) #ifndef HAVE_AVX512_ASM_SUPPORT WRAPPER_IMPL_AVX512_fFF _ZGVdN8vvv_sincosf #else @@ -496,6 +497,307 @@ WRAPPER_IMPL_AVX512_fFF _ZGVdN8vvv_sincosf vmovss %xmm0, 1280(%rsp,%r15,8) jmp .LBL_2_7 #endif +END (_ZGVeN16vl4l4_sincosf_skx) +libmvec_hidden_def(_ZGVeN16vl4l4_sincosf_skx) + +/* Wrapper between vvv and vl4l4 vector variants. */ +.macro WRAPPER_AVX512_vvv_vl4l4 callee +#ifndef __ILP32__ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + 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 + 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 + lea 64(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + movq 128(%rsp), %rdx + movq 136(%rsp), %rsi + movq 144(%rsp), %r8 + movq 152(%rsp), %r10 + movl (%rsp), %eax + movl 4(%rsp), %ecx + movl 8(%rsp), %edi + movl 12(%rsp), %r9d + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 160(%rsp), %rax + movq 168(%rsp), %rcx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 176(%rsp), %rdi + movq 184(%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 192(%rsp), %r11 + movq 200(%rsp), %rdx + movl %esi, (%rdi) + movl %r8d, (%r9) + movq 208(%rsp), %rsi + movq 216(%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 224(%rsp), %r10 + movq 232(%rsp), %rax + movl %ecx, (%rsi) + movl %edi, (%r8) + movq 240(%rsp), %rcx + movq 248(%rsp), %rdi + movl 48(%rsp), %r9d + movl 52(%rsp), %r11d + movl 56(%rsp), %edx + movl 60(%rsp), %esi + movl %r9d, (%r10) + movl %r11d, (%rax) + movq 256(%rsp), %r9 + movq 264(%rsp), %r11 + movl %edx, (%rcx) + movl %esi, (%rdi) + movq 272(%rsp), %rdx + movq 280(%rsp), %rsi + movl 64(%rsp), %r8d + movl 68(%rsp), %r10d + movl 72(%rsp), %eax + movl 76(%rsp), %ecx + movl %r8d, (%r9) + movl %r10d, (%r11) + movq 288(%rsp), %r8 + movq 296(%rsp), %r10 + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 304(%rsp), %rax + movq 312(%rsp), %rcx + movl 80(%rsp), %edi + movl 84(%rsp), %r9d + movl 88(%rsp), %r11d + movl 92(%rsp), %edx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 320(%rsp), %rdi + movq 328(%rsp), %r9 + movl %r11d, (%rax) + movl %edx, (%rcx) + movq 336(%rsp), %r11 + movq 344(%rsp), %rdx + movl 96(%rsp), %esi + movl 100(%rsp), %r8d + movl 104(%rsp), %r10d + movl 108(%rsp), %eax + movl %esi, (%rdi) + movl %r8d, (%r9) + movq 352(%rsp), %rsi + movq 360(%rsp), %r8 + movl %r10d, (%r11) + movl %eax, (%rdx) + movq 368(%rsp), %r10 + movq 376(%rsp), %rax + movl 112(%rsp), %ecx + movl 116(%rsp), %edi + movl 120(%rsp), %r9d + movl 124(%rsp), %r11d + movl %ecx, (%rsi) + movl %edi, (%r8) + movl %r9d, (%r10) + movl %r11d, (%rax) + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%rbp) + ret +#else + leal 8(%rsp), %r10d + .cfi_def_cfa 10, 0 + andl $-64, %esp + pushq -8(%r10d) + pushq %rbp + .cfi_escape 0x10,0x6,0x2,0x76,0 + movl %esp, %ebp + pushq %r10 + .cfi_escape 0xf,0x3,0x76,0x78,0x6 + 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 + call HIDDEN_JUMPTARGET(\callee) + movl -240(%ebp), %eax + vmovss -176(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -236(%ebp), %eax + vmovss -172(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -232(%ebp), %eax + vmovss -168(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -228(%ebp), %eax + vmovss -164(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -224(%ebp), %eax + vmovss -160(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -220(%ebp), %eax + vmovss -156(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -216(%ebp), %eax + vmovss -152(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -212(%ebp), %eax + vmovss -148(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -208(%ebp), %eax + vmovss -144(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -204(%ebp), %eax + vmovss -140(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -200(%ebp), %eax + vmovss -136(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -196(%ebp), %eax + vmovss -132(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -192(%ebp), %eax + vmovss -128(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -188(%ebp), %eax + vmovss -124(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -184(%ebp), %eax + vmovss -120(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -180(%ebp), %eax + vmovss -116(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -304(%ebp), %eax + vmovss -112(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -300(%ebp), %eax + vmovss -108(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -296(%ebp), %eax + vmovss -104(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -292(%ebp), %eax + vmovss -100(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -288(%ebp), %eax + vmovss -96(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -284(%ebp), %eax + vmovss -92(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -280(%ebp), %eax + vmovss -88(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -276(%ebp), %eax + vmovss -84(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -272(%ebp), %eax + vmovss -80(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -268(%ebp), %eax + vmovss -76(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -264(%ebp), %eax + vmovss -72(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -260(%ebp), %eax + vmovss -68(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -256(%ebp), %eax + vmovss -64(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -252(%ebp), %eax + vmovss -60(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -248(%ebp), %eax + vmovss -56(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -244(%ebp), %eax + vmovss -52(%ebp), %xmm0 + vmovss %xmm0, (%eax) + addl $296, %esp + popq %r10 + .cfi_def_cfa 10, 0 + popq %rbp + leal -8(%r10), %esp + .cfi_def_cfa 7, 8 + ret +#endif +.endm + +ENTRY (_ZGVeN16vvv_sincosf_knl) +WRAPPER_AVX512_vvv_vl4l4 _ZGVeN16vl4l4_sincosf_knl +END (_ZGVeN16vvv_sincosf_knl) + +ENTRY (_ZGVeN16vvv_sincosf_skx) +WRAPPER_AVX512_vvv_vl4l4 _ZGVeN16vl4l4_sincosf_skx END (_ZGVeN16vvv_sincosf_skx) .section .rodata, "a" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf4_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf4_core_sse4.S index 643fc0ca3b..16993aaffc 100644 --- a/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf4_core_sse4.S +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf4_core_sse4.S @@ -20,7 +20,7 @@ #include "svml_s_trig_data.h" .text -ENTRY (_ZGVbN4vvv_sincosf_sse4) +ENTRY (_ZGVbN4vl4l4_sincosf_sse4) /* ALGORITHM DESCRIPTION: @@ -265,4 +265,82 @@ ENTRY (_ZGVbN4vvv_sincosf_sse4) movss %xmm0, 256(%rsp,%r15,8) jmp .LBL_1_7 +END (_ZGVbN4vl4l4_sincosf_sse4) +libmvec_hidden_def(_ZGVbN4vl4l4_sincosf_sse4) + +/* vvv version implemented with wrapper to vl4l4 variant. */ +ENTRY (_ZGVbN4vvv_sincosf_sse4) +#ifndef __ILP32__ + subq $104, %rsp + .cfi_def_cfa_offset 112 + movdqu %xmm1, 32(%rsp) + lea (%rsp), %rdi + movdqu %xmm2, 48(%rdi) + lea 16(%rsp), %rsi + movdqu %xmm3, 48(%rsi) + movdqu %xmm4, 64(%rsi) + call HIDDEN_JUMPTARGET(_ZGVbN4vl4l4_sincosf_sse4) + movq 32(%rsp), %rdx + movq 40(%rsp), %rsi + movq 48(%rsp), %r8 + movq 56(%rsp), %r10 + movl (%rsp), %eax + movl 4(%rsp), %ecx + movl 8(%rsp), %edi + movl 12(%rsp), %r9d + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 64(%rsp), %rax + movq 72(%rsp), %rcx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 80(%rsp), %rdi + movq 88(%rsp), %r9 + movl 16(%rsp), %r11d + movl 20(%rsp), %edx + movl 24(%rsp), %esi + movl 28(%rsp), %r8d + movl %r11d, (%rax) + movl %edx, (%rcx) + movl %esi, (%rdi) + movl %r8d, (%r9) + addq $104, %rsp + .cfi_def_cfa_offset 8 + ret +#else + subl $72, %esp + .cfi_def_cfa_offset 80 + leal 48(%rsp), %esi + movaps %xmm1, 16(%esp) + leal 32(%rsp), %edi + movaps %xmm2, (%esp) + call HIDDEN_JUMPTARGET(_ZGVbN4vl4l4_sincosf_sse4) + movl 16(%esp), %eax + movss 32(%esp), %xmm0 + movss %xmm0, (%eax) + movl 20(%esp), %eax + movss 36(%esp), %xmm0 + movss %xmm0, (%eax) + movl 24(%esp), %eax + movss 40(%esp), %xmm0 + movss %xmm0, (%eax) + movl 28(%esp), %eax + movss 44(%esp), %xmm0 + movss %xmm0, (%eax) + movl (%esp), %eax + movss 48(%esp), %xmm0 + movss %xmm0, (%eax) + movl 4(%esp), %eax + movss 52(%esp), %xmm0 + movss %xmm0, (%eax) + movl 8(%esp), %eax + movss 56(%esp), %xmm0 + movss %xmm0, (%eax) + movl 12(%esp), %eax + movss 60(%esp), %xmm0 + movss %xmm0, (%eax) + addl $72, %esp + .cfi_def_cfa_offset 8 + ret +#endif END (_ZGVbN4vvv_sincosf_sse4) diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf8_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf8_core_avx2.S index f2a0ba7116..75aae00a46 100644 --- a/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf8_core_avx2.S +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_sincosf8_core_avx2.S @@ -20,7 +20,7 @@ #include "svml_s_trig_data.h" .text -ENTRY(_ZGVdN8vvv_sincosf_avx2) +ENTRY (_ZGVdN8vl4l4_sincosf_avx2) /* ALGORITHM DESCRIPTION: @@ -238,4 +238,152 @@ ENTRY(_ZGVdN8vvv_sincosf_avx2) vmovss %xmm0, 384(%rsp,%r15,8) jmp .LBL_1_7 -END(_ZGVdN8vvv_sincosf_avx2) +END (_ZGVdN8vl4l4_sincosf_avx2) +libmvec_hidden_def(_ZGVdN8vl4l4_sincosf_avx2) + +/* vvv version implemented with wrapper to vl4l4 variant. */ +ENTRY (_ZGVdN8vvv_sincosf_avx2) +#ifndef __ILP32__ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-32, %rsp + subq $192, %rsp + vmovdqu %ymm1, 64(%rsp) + lea (%rsp), %rdi + vmovdqu %ymm2, 96(%rdi) + vmovdqu %ymm3, 128(%rdi) + vmovdqu %ymm4, 160(%rdi) + lea 32(%rsp), %rsi + call HIDDEN_JUMPTARGET(_ZGVdN8vl4l4_sincosf_avx2) + movq 64(%rsp), %rdx + movq 72(%rsp), %rsi + movq 80(%rsp), %r8 + movq 88(%rsp), %r10 + movl (%rsp), %eax + movl 4(%rsp), %ecx + movl 8(%rsp), %edi + movl 12(%rsp), %r9d + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 96(%rsp), %rax + movq 104(%rsp), %rcx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 112(%rsp), %rdi + movq 120(%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 128(%rsp), %r11 + movq 136(%rsp), %rdx + movl %esi, (%rdi) + movl %r8d, (%r9) + movq 144(%rsp), %rsi + movq 152(%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 160(%rsp), %r10 + movq 168(%rsp), %rax + movl %ecx, (%rsi) + movl %edi, (%r8) + movq 176(%rsp), %rcx + movq 184(%rsp), %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 + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%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 %r10 + .cfi_escape 0xf,0x3,0x76,0x78,0x6 + leal -48(%rbp), %esi + leal -80(%rbp), %edi + subl $136, %esp + vmovdqa %ymm1, -112(%ebp) + vmovdqa %ymm2, -144(%ebp) + call HIDDEN_JUMPTARGET(_ZGVdN8vl4l4_sincosf_avx2) + vmovdqa -112(%ebp), %xmm0 + vmovq %xmm0, %rax + vmovss -80(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -76(%ebp), %xmm0 + shrq $32, %rax + vmovss %xmm0, (%eax) + movq -104(%ebp), %rax + vmovss -72(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -68(%ebp), %xmm0 + shrq $32, %rax + vmovss %xmm0, (%eax) + movq -96(%ebp), %rax + vmovss -64(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -60(%ebp), %xmm0 + shrq $32, %rax + vmovss %xmm0, (%eax) + movq -88(%ebp), %rax + vmovss -56(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -52(%ebp), %xmm0 + shrq $32, %rax + vmovss %xmm0, (%eax) + vmovdqa -144(%ebp), %xmm0 + vmovq %xmm0, %rax + vmovss -48(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -44(%ebp), %xmm0 + shrq $32, %rax + vmovss %xmm0, (%eax) + movq -136(%ebp), %rax + vmovss -40(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -36(%ebp), %xmm0 + shrq $32, %rax + vmovss %xmm0, (%eax) + movq -128(%ebp), %rax + vmovss -32(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -28(%ebp), %xmm0 + shrq $32, %rax + vmovss %xmm0, (%eax) + movq -120(%ebp), %rax + vmovss -24(%ebp), %xmm0 + vmovss %xmm0, (%eax) + vmovss -20(%ebp), %xmm0 + shrq $32, %rax + vmovss %xmm0, (%eax) + addl $136, %esp + popq %r10 + .cfi_def_cfa 10, 0 + popq %rbp + leal -8(%r10), %esp + .cfi_def_cfa 7, 8 + ret +#endif +END (_ZGVdN8vvv_sincosf_avx2) diff --git a/sysdeps/x86_64/fpu/svml_d_sincos2_core.S b/sysdeps/x86_64/fpu/svml_d_sincos2_core.S index 74afa0a677..96ab726f79 100644 --- a/sysdeps/x86_64/fpu/svml_d_sincos2_core.S +++ b/sysdeps/x86_64/fpu/svml_d_sincos2_core.S @@ -20,8 +20,89 @@ #include "svml_d_wrapper_impl.h" .text -ENTRY (_ZGVbN2vvv_sincos) +ENTRY (_ZGVbN2vl8l8_sincos) WRAPPER_IMPL_SSE2_fFF sincos +END (_ZGVbN2vl8l8_sincos) +libmvec_hidden_def (_ZGVbN2vl8l8_sincos) + +/* SSE2 ISA version as wrapper to scalar (for vector + function declared with #pragma omp declare simd notinbranch). */ +.macro WRAPPER_IMPL_SSE2_fFF_vvv callee +#ifndef __ILP32__ + subq $88, %rsp + cfi_adjust_cfa_offset(88) + movaps %xmm0, 64(%rsp) + lea (%rsp), %rdi + movdqa %xmm1, 32(%rdi) + lea 16(%rsp), %rsi + movdqa %xmm2, 32(%rsi) + call JUMPTARGET(\callee) + movsd 72(%rsp), %xmm0 + lea 8(%rsp), %rdi + lea 24(%rsp), %rsi + call JUMPTARGET(\callee) + movq 32(%rsp), %rdx + movq 48(%rsp), %rsi + movq 40(%rsp), %r8 + movq 56(%rsp), %r10 + movq (%rsp), %rax + movq 16(%rsp), %rcx + movq 8(%rsp), %rdi + movq 24(%rsp), %r9 + movq %rax, (%rdx) + movq %rcx, (%rsi) + movq %rdi, (%r8) + movq %r9, (%r10) + addq $88, %rsp + cfi_adjust_cfa_offset(-88) + ret +#else + pushq %rbp + .cfi_def_cfa_offset 16 + .cfi_offset 6, -16 + pushq %rbx + .cfi_def_cfa_offset 24 + .cfi_offset 3, -24 + subl $88, %esp + .cfi_def_cfa_offset 112 + leal 64(%rsp), %esi + movaps %xmm1, 32(%esp) + leal 48(%rsp), %edi + movaps %xmm2, 16(%esp) + movq %rsi, %rbp + movq %rdi, %rbx + movaps %xmm0, (%esp) + call JUMPTARGET(\callee) + movupd 8(%esp), %xmm0 + leal 8(%rbp), %esi + leal 8(%rbx), %edi + call JUMPTARGET(\callee) + movdqa 32(%esp), %xmm1 + movsd 48(%esp), %xmm0 + movq %xmm1, %rax + movdqa 16(%esp), %xmm2 + movsd %xmm0, (%eax) + movsd 56(%esp), %xmm0 + pextrd $1, %xmm1, %eax + movsd %xmm0, (%eax) + movsd 64(%esp), %xmm0 + movq %xmm2, %rax + movsd %xmm0, (%eax) + movsd 72(%esp), %xmm0 + pextrd $1, %xmm2, %eax + movsd %xmm0, (%eax) + addl $88, %esp + .cfi_def_cfa_offset 24 + popq %rbx + .cfi_def_cfa_offset 16 + popq %rbp + .cfi_def_cfa_offset 8 + ret +#endif +.endm + +ENTRY (_ZGVbN2vvv_sincos) +WRAPPER_IMPL_SSE2_fFF_vvv sincos END (_ZGVbN2vvv_sincos) #ifndef USE_MULTIARCH diff --git a/sysdeps/x86_64/fpu/svml_d_sincos4_core.S b/sysdeps/x86_64/fpu/svml_d_sincos4_core.S index 2c0b011fb3..088d5ad917 100644 --- a/sysdeps/x86_64/fpu/svml_d_sincos4_core.S +++ b/sysdeps/x86_64/fpu/svml_d_sincos4_core.S @@ -20,8 +20,131 @@ #include "svml_d_wrapper_impl.h" .text +ENTRY (_ZGVdN4vl8l8_sincos) +WRAPPER_IMPL_AVX_fFF _ZGVbN2vl8l8_sincos +END (_ZGVdN4vl8l8_sincos) +libmvec_hidden_def (_ZGVdN4vl8l8_sincos) + +/* AVX2 ISA version as wrapper to SSE ISA version (for vector + function declared with #pragma omp declare simd notinbranch). */ +.macro WRAPPER_IMPL_AVX2_fFF_vvv callee +#ifndef __ILP32__ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-32, %rsp + subq $160, %rsp + vmovupd %ymm0, 128(%rsp) + lea (%rsp), %rdi + vmovdqu %ymm1, 64(%rdi) + vmovdqu %ymm2, 96(%rdi) + lea 32(%rsp), %rsi + vzeroupper + call HIDDEN_JUMPTARGET(\callee) + vmovupd 144(%rsp), %xmm0 + lea 16(%rsp), %rdi + lea 48(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + movq 64(%rsp), %rdx + movq 96(%rsp), %rsi + movq 72(%rsp), %r8 + movq 104(%rsp), %r10 + movq (%rsp), %rax + movq 32(%rsp), %rcx + movq 8(%rsp), %rdi + movq 40(%rsp), %r9 + movq %rax, (%rdx) + movq %rcx, (%rsi) + movq 80(%rsp), %rax + movq 112(%rsp), %rcx + movq %rdi, (%r8) + movq %r9, (%r10) + movq 88(%rsp), %rdi + movq 120(%rsp), %r9 + movq 16(%rsp), %r11 + movq 48(%rsp), %rdx + movq 24(%rsp), %rsi + movq 56(%rsp), %r8 + movq %r11, (%rax) + movq %rdx, (%rcx) + movq %rsi, (%rdi) + movq %r8, (%r9) + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%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 + vmovapd -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 (_ZGVdN4vvv_sincos) -WRAPPER_IMPL_AVX_fFF _ZGVbN2vvv_sincos +WRAPPER_IMPL_AVX2_fFF_vvv _ZGVbN2vl8l8_sincos END (_ZGVdN4vvv_sincos) #ifndef USE_MULTIARCH diff --git a/sysdeps/x86_64/fpu/svml_d_sincos4_core_avx.S b/sysdeps/x86_64/fpu/svml_d_sincos4_core_avx.S index e4320a97c7..a60a524eeb 100644 --- a/sysdeps/x86_64/fpu/svml_d_sincos4_core_avx.S +++ b/sysdeps/x86_64/fpu/svml_d_sincos4_core_avx.S @@ -20,6 +20,124 @@ #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 _ZGVbN2vvv_sincos +WRAPPER_IMPL_AVX_fFF_vvv _ZGVbN2vl8l8_sincos END (_ZGVcN4vvv_sincos) diff --git a/sysdeps/x86_64/fpu/svml_d_sincos8_core.S b/sysdeps/x86_64/fpu/svml_d_sincos8_core.S index 68d490e5bc..7f51ed5c8c 100644 --- a/sysdeps/x86_64/fpu/svml_d_sincos8_core.S +++ b/sysdeps/x86_64/fpu/svml_d_sincos8_core.S @@ -20,6 +20,205 @@ #include "svml_d_wrapper_impl.h" .text +ENTRY (_ZGVeN8vl8l8_sincos) +WRAPPER_IMPL_AVX512_fFF _ZGVdN4vl8l8_sincos +END (_ZGVeN8vl8l8_sincos) + +/* AVX512 ISA version as wrapper to AVX2 ISA version (for vector + function declared with #pragma omp declare simd notinbranch). */ +.macro WRAPPER_IMPL_AVX512_fFF_vvv callee +#ifndef __ILP32__ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-64, %rsp + subq $320, %rsp + /* Encoding for vmovups %zmm0, 256(%rsp). */ + .byte 0x62 + .byte 0xf1 + .byte 0x7c + .byte 0x48 + .byte 0x11 + .byte 0x44 + .byte 0x24 + .byte 0x04 + lea (%rsp), %rdi + /* Encoding for vmovups %zmm1, 128(%rdi). */ + .byte 0x62 + .byte 0xf1 + .byte 0x7c + .byte 0x48 + .byte 0x11 + .byte 0x4f + .byte 0x02 + /* Encoding for vmovups %zmm2, 192(%rdi). */ + .byte 0x62 + .byte 0xf1 + .byte 0x7c + .byte 0x48 + .byte 0x11 + .byte 0x57 + .byte 0x03 + lea 64(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + vmovdqu 288(%rsp), %ymm0 + lea 32(%rsp), %rdi + lea 96(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + movq 128(%rsp), %rdx + movq 192(%rsp), %rsi + movq 136(%rsp), %r8 + movq 200(%rsp), %r10 + movq (%rsp), %rax + movq 64(%rsp), %rcx + movq 8(%rsp), %rdi + movq 72(%rsp), %r9 + movq %rax, (%rdx) + movq %rcx, (%rsi) + movq 144(%rsp), %rax + movq 208(%rsp), %rcx + movq %rdi, (%r8) + movq %r9, (%r10) + movq 152(%rsp), %rdi + movq 216(%rsp), %r9 + movq 16(%rsp), %r11 + movq 80(%rsp), %rdx + movq 24(%rsp), %rsi + movq 88(%rsp), %r8 + movq %r11, (%rax) + movq %rdx, (%rcx) + movq 160(%rsp), %r11 + movq 224(%rsp), %rdx + movq %rsi, (%rdi) + movq %r8, (%r9) + movq 168(%rsp), %rsi + movq 232(%rsp), %r8 + movq 32(%rsp), %r10 + movq 96(%rsp), %rax + movq 40(%rsp), %rcx + movq 104(%rsp), %rdi + movq %r10, (%r11) + movq %rax, (%rdx) + movq 176(%rsp), %r10 + movq 240(%rsp), %rax + movq %rcx, (%rsi) + movq %rdi, (%r8) + movq 184(%rsp), %rcx + movq 248(%rsp), %rdi + movq 48(%rsp), %r9 + movq 112(%rsp), %r11 + movq 56(%rsp), %rdx + movq 120(%rsp), %rsi + movq %r9, (%r10) + movq %r11, (%rax) + movq %rdx, (%rcx) + movq %rsi, (%rdi) + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%rbp) + ret +#else + leal 8(%rsp), %r10d + .cfi_def_cfa 10, 0 + andl $-64, %esp + pushq -8(%r10d) + pushq %rbp + .cfi_escape 0x10,0x6,0x2,0x76,0 + movl %esp, %ebp + pushq %r12 + leal -112(%rbp), %esi + pushq %r10 + .cfi_escape 0xf,0x3,0x76,0x70,0x6 + .cfi_escape 0x10,0xc,0x2,0x76,0x78 + leal -176(%rbp), %edi + movq %rsi, %r12 + pushq %rbx + .cfi_escape 0x10,0x3,0x2,0x76,0x68 + movq %rdi, %rbx + subl $280, %esp + vmovdqa %ymm1, -208(%ebp) + vmovdqa %ymm2, -240(%ebp) + /* Encoding for vmovapd %zmm0, -304(%ebp). */ + .byte 0x67 + .byte 0x62 + .byte 0xf1 + .byte 0xfd + .byte 0x48 + .byte 0x29 + .byte 0x85 + .byte 0xd0 + .byte 0xfe + .byte 0xff + .byte 0xff + call HIDDEN_JUMPTARGET(\callee) + leal 32(%r12), %esi + vmovupd -272(%ebp), %ymm0 + leal 32(%rbx), %edi + call HIDDEN_JUMPTARGET(\callee) + movl -208(%ebp), %eax + vmovsd -176(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -204(%ebp), %eax + vmovsd -168(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -200(%ebp), %eax + vmovsd -160(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -196(%ebp), %eax + vmovsd -152(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -192(%ebp), %eax + vmovsd -144(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -188(%ebp), %eax + vmovsd -136(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -184(%ebp), %eax + vmovsd -128(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -180(%ebp), %eax + vmovsd -120(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -240(%ebp), %eax + vmovsd -112(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -236(%ebp), %eax + vmovsd -104(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -232(%ebp), %eax + vmovsd -96(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -228(%ebp), %eax + vmovsd -88(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -224(%ebp), %eax + vmovsd -80(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -220(%ebp), %eax + vmovsd -72(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -216(%ebp), %eax + vmovsd -64(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + movl -212(%ebp), %eax + vmovsd -56(%ebp), %xmm0 + vmovsd %xmm0, (%eax) + addl $280, %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 (_ZGVeN8vvv_sincos) -WRAPPER_IMPL_AVX512_fFF _ZGVdN4vvv_sincos +WRAPPER_IMPL_AVX512_fFF_vvv _ZGVdN4vl8l8_sincos END (_ZGVeN8vvv_sincos) diff --git a/sysdeps/x86_64/fpu/svml_s_sincosf16_core.S b/sysdeps/x86_64/fpu/svml_s_sincosf16_core.S index 5cbf10b8da..aae1adb8d7 100644 --- a/sysdeps/x86_64/fpu/svml_s_sincosf16_core.S +++ b/sysdeps/x86_64/fpu/svml_s_sincosf16_core.S @@ -20,6 +20,339 @@ #include "svml_s_wrapper_impl.h" .text +ENTRY (_ZGVeN16vl4l4_sincosf) +WRAPPER_IMPL_AVX512_fFF _ZGVdN8vl4l4_sincosf +END (_ZGVeN16vl4l4_sincosf) + +/* AVX512 ISA version as wrapper to AVX2 ISA version (for vector + function declared with #pragma omp declare simd notinbranch). */ +.macro WRAPPER_IMPL_AVX512_fFF_vvv callee +#ifndef __ILP32__ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-64, %rsp + subq $448, %rsp + /* Encoding for vmovups %zmm0, 384(%rsp). */ + .byte 0x62 + .byte 0xf1 + .byte 0x7c + .byte 0x48 + .byte 0x11 + .byte 0x44 + .byte 0x24 + .byte 0x06 + lea (%rsp), %rdi + /* Encoding for vmovups %zmm1, 128(%rdi). */ + .byte 0x62 + .byte 0xf1 + .byte 0x7c + .byte 0x48 + .byte 0x11 + .byte 0x4f + .byte 0x02 + /* 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 + lea 64(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + vmovdqu 416(%rsp), %ymm0 + lea 32(%rsp), %rdi + lea 96(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + movq 128(%rsp), %rdx + movq 136(%rsp), %rsi + movq 144(%rsp), %r8 + movq 152(%rsp), %r10 + movl (%rsp), %eax + movl 4(%rsp), %ecx + movl 8(%rsp), %edi + movl 12(%rsp), %r9d + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 160(%rsp), %rax + movq 168(%rsp), %rcx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 176(%rsp), %rdi + movq 184(%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 192(%rsp), %r11 + movq 200(%rsp), %rdx + movl %esi, (%rdi) + movl %r8d, (%r9) + movq 208(%rsp), %rsi + movq 216(%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 224(%rsp), %r10 + movq 232(%rsp), %rax + movl %ecx, (%rsi) + movl %edi, (%r8) + movq 240(%rsp), %rcx + movq 248(%rsp), %rdi + movl 48(%rsp), %r9d + movl 52(%rsp), %r11d + movl 56(%rsp), %edx + movl 60(%rsp), %esi + movl %r9d, (%r10) + movl %r11d, (%rax) + movq 256(%rsp), %r9 + movq 264(%rsp), %r11 + movl %edx, (%rcx) + movl %esi, (%rdi) + movq 272(%rsp), %rdx + movq 280(%rsp), %rsi + movl 64(%rsp), %r8d + movl 68(%rsp), %r10d + movl 72(%rsp), %eax + movl 76(%rsp), %ecx + movl %r8d, (%r9) + movl %r10d, (%r11) + movq 288(%rsp), %r8 + movq 296(%rsp), %r10 + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 304(%rsp), %rax + movq 312(%rsp), %rcx + movl 80(%rsp), %edi + movl 84(%rsp), %r9d + movl 88(%rsp), %r11d + movl 92(%rsp), %edx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 320(%rsp), %rdi + movq 328(%rsp), %r9 + movl %r11d, (%rax) + movl %edx, (%rcx) + movq 336(%rsp), %r11 + movq 344(%rsp), %rdx + movl 96(%rsp), %esi + movl 100(%rsp), %r8d + movl 104(%rsp), %r10d + movl 108(%rsp), %eax + movl %esi, (%rdi) + movl %r8d, (%r9) + movq 352(%rsp), %rsi + movq 360(%rsp), %r8 + movl %r10d, (%r11) + movl %eax, (%rdx) + movq 368(%rsp), %r10 + movq 376(%rsp), %rax + movl 112(%rsp), %ecx + movl 116(%rsp), %edi + movl 120(%rsp), %r9d + movl 124(%rsp), %r11d + movl %ecx, (%rsi) + movl %edi, (%r8) + movl %r9d, (%r10) + movl %r11d, (%rax) + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%rbp) + ret +#else + leal 8(%rsp), %r10d + .cfi_def_cfa 10, 0 + andl $-64, %esp + pushq -8(%r10d) + pushq %rbp + .cfi_escape 0x10,0x6,0x2,0x76,0 + movl %esp, %ebp + pushq %r12 + leal -112(%rbp), %esi + pushq %r10 + .cfi_escape 0xf,0x3,0x76,0x70,0x6 + .cfi_escape 0x10,0xc,0x2,0x76,0x78 + leal -176(%rbp), %edi + movq %rsi, %r12 + pushq %rbx + .cfi_escape 0x10,0x3,0x2,0x76,0x68 + movq %rdi, %rbx + subl $344, %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 + /* Encoding for vmovaps %zmm0, -368(%ebp). */ + .byte 0x67 + .byte 0x62 + .byte 0xf1 + .byte 0x7c + .byte 0x48 + .byte 0x29 + .byte 0x85 + .byte 0x90 + .byte 0xfe + .byte 0xff + .byte 0xff + call HIDDEN_JUMPTARGET(\callee) + leal 32(%r12), %esi + vmovups -336(%ebp), %ymm0 + leal 32(%rbx), %edi + call HIDDEN_JUMPTARGET(\callee) + movl -240(%ebp), %eax + vmovss -176(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -236(%ebp), %eax + vmovss -172(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -232(%ebp), %eax + vmovss -168(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -228(%ebp), %eax + vmovss -164(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -224(%ebp), %eax + vmovss -160(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -220(%ebp), %eax + vmovss -156(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -216(%ebp), %eax + vmovss -152(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -212(%ebp), %eax + vmovss -148(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -208(%ebp), %eax + vmovss -144(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -204(%ebp), %eax + vmovss -140(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -200(%ebp), %eax + vmovss -136(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -196(%ebp), %eax + vmovss -132(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -192(%ebp), %eax + vmovss -128(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -188(%ebp), %eax + vmovss -124(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -184(%ebp), %eax + vmovss -120(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -180(%ebp), %eax + vmovss -116(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -304(%ebp), %eax + vmovss -112(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -300(%ebp), %eax + vmovss -108(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -296(%ebp), %eax + vmovss -104(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -292(%ebp), %eax + vmovss -100(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -288(%ebp), %eax + vmovss -96(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -284(%ebp), %eax + vmovss -92(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -280(%ebp), %eax + vmovss -88(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -276(%ebp), %eax + vmovss -84(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -272(%ebp), %eax + vmovss -80(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -268(%ebp), %eax + vmovss -76(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -264(%ebp), %eax + vmovss -72(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -260(%ebp), %eax + vmovss -68(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -256(%ebp), %eax + vmovss -64(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -252(%ebp), %eax + vmovss -60(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -248(%ebp), %eax + vmovss -56(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -244(%ebp), %eax + vmovss -52(%ebp), %xmm0 + vmovss %xmm0, (%eax) + addl $344, %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 (_ZGVeN16vvv_sincosf) -WRAPPER_IMPL_AVX512_fFF _ZGVdN8vvv_sincosf +WRAPPER_IMPL_AVX512_fFF_vvv _ZGVdN8vl4l4_sincosf END (_ZGVeN16vvv_sincosf) diff --git a/sysdeps/x86_64/fpu/svml_s_sincosf4_core.S b/sysdeps/x86_64/fpu/svml_s_sincosf4_core.S index 1a7d2733af..0963c391ff 100644 --- a/sysdeps/x86_64/fpu/svml_s_sincosf4_core.S +++ b/sysdeps/x86_64/fpu/svml_s_sincosf4_core.S @@ -16,13 +16,135 @@ License along with the GNU C Library; if not, see <http://www.gnu.org/licenses/>. */ - #include <sysdep.h> #include "svml_s_wrapper_impl.h" .text -ENTRY (_ZGVbN4vvv_sincosf) +ENTRY (_ZGVbN4vl4l4_sincosf) WRAPPER_IMPL_SSE2_fFF sincosf +END (_ZGVbN4vl4l4_sincosf) +libmvec_hidden_def (_ZGVbN4vl4l4_sincosf) + +/* SSE2 ISA version as wrapper to scalar (for vector + function declared with #pragma omp declare simd notinbranch). */ +.macro WRAPPER_IMPL_SSE2_fFF_vvv callee +#ifndef __ILP32__ + subq $120, %rsp + cfi_adjust_cfa_offset(120) + movaps %xmm0, 96(%rsp) + lea (%rsp), %rdi + movdqa %xmm1, 32(%rdi) + lea 16(%rsp), %rsi + movdqa %xmm2, 32(%rsi) + movdqa %xmm3, 48(%rsi) + movdqa %xmm4, 64(%rsi) + call JUMPTARGET(\callee) + movss 100(%rsp), %xmm0 + lea 4(%rsp), %rdi + lea 20(%rsp), %rsi + call JUMPTARGET(\callee) + movss 104(%rsp), %xmm0 + lea 8(%rsp), %rdi + lea 24(%rsp), %rsi + call JUMPTARGET(\callee) + movss 108(%rsp), %xmm0 + lea 12(%rsp), %rdi + lea 28(%rsp), %rsi + call JUMPTARGET(\callee) + movq 32(%rsp), %rdx + movq 40(%rsp), %rsi + movq 48(%rsp), %r8 + movq 56(%rsp), %r10 + movl (%rsp), %eax + movl 4(%rsp), %ecx + movl 8(%rsp), %edi + movl 12(%rsp), %r9d + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 64(%rsp), %rax + movq 72(%rsp), %rcx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 80(%rsp), %rdi + movq 88(%rsp), %r9 + movl 16(%rsp), %r11d + movl 20(%rsp), %edx + movl 24(%rsp), %esi + movl 28(%rsp), %r8d + movl %r11d, (%rax) + movl %edx, (%rcx) + movl %esi, (%rdi) + movl %r8d, (%r9) + addq $120, %rsp + cfi_adjust_cfa_offset(-120) + ret +#else + pushq %rbp + .cfi_def_cfa_offset 16 + .cfi_offset 6, -16 + pushq %rbx + .cfi_def_cfa_offset 24 + .cfi_offset 3, -24 + subl $88, %esp + .cfi_def_cfa_offset 112 + leal 64(%rsp), %esi + movaps %xmm1, (%esp) + leal 48(%rsp), %edi + movaps %xmm2, 16(%esp) + movq %rsi, %rbp + movq %rdi, %rbx + movaps %xmm0, 32(%esp) + call JUMPTARGET(\callee) + movups 36(%esp), %xmm0 + leal 4(%rbp), %esi + leal 4(%rbx), %edi + call JUMPTARGET(\callee) + movups 40(%esp), %xmm0 + leal 8(%rbp), %esi + leal 8(%rbx), %edi + call JUMPTARGET(\callee) + movups 44(%esp), %xmm0 + leal 12(%rbp), %esi + leal 12(%rbx), %edi + call JUMPTARGET(\callee) + movq (%esp), %rax + movss 48(%esp), %xmm0 + movdqa (%esp), %xmm4 + movdqa 16(%esp), %xmm7 + movss %xmm0, (%eax) + movss 52(%esp), %xmm0 + pextrd $1, %xmm4, %eax + movss %xmm0, (%eax) + movq 8(%esp), %rax + movss 56(%esp), %xmm0 + movss %xmm0, (%eax) + movss 60(%esp), %xmm0 + pextrd $3, %xmm4, %eax + movss %xmm0, (%eax) + movq 16(%esp), %rax + movss 64(%esp), %xmm0 + movss %xmm0, (%eax) + movss 68(%esp), %xmm0 + pextrd $1, %xmm7, %eax + movss %xmm0, (%eax) + movq 24(%esp), %rax + movss 72(%esp), %xmm0 + movss %xmm0, (%eax) + movss 76(%esp), %xmm0 + pextrd $3, %xmm7, %eax + movss %xmm0, (%eax) + addl $88, %esp + .cfi_def_cfa_offset 24 + popq %rbx + .cfi_def_cfa_offset 16 + popq %rbp + .cfi_def_cfa_offset 8 + ret +#endif +.endm + +ENTRY (_ZGVbN4vvv_sincosf) +WRAPPER_IMPL_SSE2_fFF_vvv sincosf END (_ZGVbN4vvv_sincosf) #ifndef USE_MULTIARCH diff --git a/sysdeps/x86_64/fpu/svml_s_sincosf8_core.S b/sysdeps/x86_64/fpu/svml_s_sincosf8_core.S index 74d1dfd1a8..93ac91608f 100644 --- a/sysdeps/x86_64/fpu/svml_s_sincosf8_core.S +++ b/sysdeps/x86_64/fpu/svml_s_sincosf8_core.S @@ -20,8 +20,179 @@ #include "svml_s_wrapper_impl.h" .text +ENTRY (_ZGVdN8vl4l4_sincosf) +WRAPPER_IMPL_AVX_fFF _ZGVbN4vl4l4_sincosf +END (_ZGVdN8vl4l4_sincosf) +libmvec_hidden_def (_ZGVdN8vl4l4_sincosf) + +/* AVX2 ISA version as wrapper to SSE ISA version (for vector + function declared with #pragma omp declare simd notinbranch). */ +.macro WRAPPER_IMPL_AVX2_fFF_vvv callee +#ifndef __ILP32__ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-32, %rsp + subq $224, %rsp + vmovups %ymm0, 192(%rsp) + lea (%rsp), %rdi + vmovdqu %ymm1, 64(%rdi) + vmovdqu %ymm2, 96(%rdi) + vmovdqu %ymm3, 128(%rdi) + vmovdqu %ymm4, 160(%rdi) + lea 32(%rsp), %rsi + vzeroupper + call HIDDEN_JUMPTARGET(\callee) + vmovups 208(%rsp), %xmm0 + lea 16(%rsp), %rdi + lea 48(%rsp), %rsi + call HIDDEN_JUMPTARGET(\callee) + movq 64(%rsp), %rdx + movq 72(%rsp), %rsi + movq 80(%rsp), %r8 + movq 88(%rsp), %r10 + movl (%rsp), %eax + movl 4(%rsp), %ecx + movl 8(%rsp), %edi + movl 12(%rsp), %r9d + movl %eax, (%rdx) + movl %ecx, (%rsi) + movq 96(%rsp), %rax + movq 104(%rsp), %rcx + movl %edi, (%r8) + movl %r9d, (%r10) + movq 112(%rsp), %rdi + movq 120(%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 128(%rsp), %r11 + movq 136(%rsp), %rdx + movl %esi, (%rdi) + movl %r8d, (%r9) + movq 144(%rsp), %rsi + movq 152(%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 160(%rsp), %r10 + movq 168(%rsp), %rax + movl %ecx, (%rsi) + movl %edi, (%r8) + movq 176(%rsp), %rcx + movq 184(%rsp), %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 + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%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 + vmovdqa %ymm1, -144(%ebp) + vmovdqa %ymm2, -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) + movl -144(%ebp), %eax + vmovss -112(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -140(%ebp), %eax + vmovss -108(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -136(%ebp), %eax + vmovss -104(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -132(%ebp), %eax + vmovss -100(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -128(%ebp), %eax + vmovss -96(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -124(%ebp), %eax + vmovss -92(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -120(%ebp), %eax + vmovss -88(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -116(%ebp), %eax + vmovss -84(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -176(%ebp), %eax + vmovss -80(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -172(%ebp), %eax + vmovss -76(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -168(%ebp), %eax + vmovss -72(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -164(%ebp), %eax + vmovss -68(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -160(%ebp), %eax + vmovss -64(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -156(%ebp), %eax + vmovss -60(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -152(%ebp), %eax + vmovss -56(%ebp), %xmm0 + vmovss %xmm0, (%eax) + movl -148(%ebp), %eax + vmovss -52(%ebp), %xmm0 + 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 (_ZGVdN8vvv_sincosf) -WRAPPER_IMPL_AVX_fFF _ZGVbN4vvv_sincosf +WRAPPER_IMPL_AVX2_fFF_vvv _ZGVbN4vl4l4_sincosf END (_ZGVdN8vvv_sincosf) #ifndef USE_MULTIARCH 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) diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx.c b/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx.c new file mode 100644 index 0000000000..896f1bcbaf --- /dev/null +++ b/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx.c @@ -0,0 +1 @@ +#include "test-double-libmvec-sincos.c" diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx2.c b/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx2.c new file mode 100644 index 0000000000..896f1bcbaf --- /dev/null +++ b/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx2.c @@ -0,0 +1 @@ +#include "test-double-libmvec-sincos.c" diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx512.c b/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx512.c new file mode 100644 index 0000000000..896f1bcbaf --- /dev/null +++ b/sysdeps/x86_64/fpu/test-double-libmvec-sincos-avx512.c @@ -0,0 +1 @@ +#include "test-double-libmvec-sincos.c" diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-sincos.c b/sysdeps/x86_64/fpu/test-double-libmvec-sincos.c new file mode 100644 index 0000000000..80348a260e --- /dev/null +++ b/sysdeps/x86_64/fpu/test-double-libmvec-sincos.c @@ -0,0 +1,69 @@ +/* Test for vector sincos ABI. + Copyright (C) 2016 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 <math.h> +#include <math-tests-arch.h> + +#define N 1000 +double x[N], s[N], c[N]; +double* s_ptrs[N]; +double* c_ptrs[N]; +int arch_check = 1; + +static void +init_arg (void) +{ + int i; + + CHECK_ARCH_EXT; + + arch_check = 0; + + for(i = 0; i < N; i++) + { + x[i] = i / 3; + s_ptrs[i] = &s[i]; + c_ptrs[i] = &c[i]; + } +} + +static int +test_sincos_abi (void) +{ + int i; + + init_arg (); + + if (arch_check) + return 77; + +#pragma omp simd + for(i = 0; i < N; i++) + sincos (x[i], s_ptrs[i], c_ptrs[i]); + + return 0; +} + +static int +do_test (void) +{ + return test_sincos_abi (); +} + +#define TEST_FUNCTION do_test () +#include "../../../test-skeleton.c" diff --git a/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c index a9d15979aa..375582e6d3 100644 --- a/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c +++ b/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c @@ -17,13 +17,17 @@ <http://www.gnu.org/licenses/>. */ #include "test-double-vlen2.h" +#include "test-math-vector-sincos.h" #include <immintrin.h> #define VEC_TYPE __m128d VECTOR_WRAPPER (WRAPPER_NAME (cos), _ZGVbN2v_cos) VECTOR_WRAPPER (WRAPPER_NAME (sin), _ZGVbN2v_sin) -VECTOR_WRAPPER_fFF (WRAPPER_NAME (sincos), _ZGVbN2vvv_sincos) VECTOR_WRAPPER (WRAPPER_NAME (log), _ZGVbN2v_log) VECTOR_WRAPPER (WRAPPER_NAME (exp), _ZGVbN2v_exp) VECTOR_WRAPPER_ff (WRAPPER_NAME (pow), _ZGVbN2vv_pow) + +#define VEC_INT_TYPE __m128i + +VECTOR_WRAPPER_fFF_2 (WRAPPER_NAME (sincos), _ZGVbN2vvv_sincos) diff --git a/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c index eb6a531502..00b7d4ef26 100644 --- a/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c +++ b/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c @@ -17,6 +17,7 @@ <http://www.gnu.org/licenses/>. */ #include "test-double-vlen4.h" +#include "test-math-vector-sincos.h" #include <immintrin.h> #undef VEC_SUFF @@ -26,7 +27,14 @@ VECTOR_WRAPPER (WRAPPER_NAME (cos), _ZGVdN4v_cos) VECTOR_WRAPPER (WRAPPER_NAME (sin), _ZGVdN4v_sin) -VECTOR_WRAPPER_fFF (WRAPPER_NAME (sincos), _ZGVdN4vvv_sincos) VECTOR_WRAPPER (WRAPPER_NAME (log), _ZGVdN4v_log) VECTOR_WRAPPER (WRAPPER_NAME (exp), _ZGVdN4v_exp) VECTOR_WRAPPER_ff (WRAPPER_NAME (pow), _ZGVdN4vv_pow) + +#ifndef __ILP32__ +# define VEC_INT_TYPE __m256i +#else +# define VEC_INT_TYPE __m128i +#endif + +VECTOR_WRAPPER_fFF_2 (WRAPPER_NAME (sincos), _ZGVdN4vvv_sincos) diff --git a/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c index 52b81da3ee..51ddbfadf5 100644 --- a/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c +++ b/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c @@ -17,13 +17,21 @@ <http://www.gnu.org/licenses/>. */ #include "test-double-vlen4.h" +#include "test-math-vector-sincos.h" #include <immintrin.h> #define VEC_TYPE __m256d VECTOR_WRAPPER (WRAPPER_NAME (cos), _ZGVcN4v_cos) VECTOR_WRAPPER (WRAPPER_NAME (sin), _ZGVcN4v_sin) -VECTOR_WRAPPER_fFF (WRAPPER_NAME (sincos), _ZGVcN4vvv_sincos) VECTOR_WRAPPER (WRAPPER_NAME (log), _ZGVcN4v_log) VECTOR_WRAPPER (WRAPPER_NAME (exp), _ZGVcN4v_exp) VECTOR_WRAPPER_ff (WRAPPER_NAME (pow), _ZGVcN4vv_pow) + +#define VEC_INT_TYPE __m128i + +#ifndef __ILP32__ +VECTOR_WRAPPER_fFF_3 (WRAPPER_NAME (sincos), _ZGVcN4vvv_sincos) +#else +VECTOR_WRAPPER_fFF_2 (WRAPPER_NAME (sincos), _ZGVcN4vvv_sincos) +#endif diff --git a/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c index c10bb9cb4a..5460b6b26b 100644 --- a/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c +++ b/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c @@ -17,13 +17,21 @@ <http://www.gnu.org/licenses/>. */ #include "test-double-vlen8.h" +#include "test-math-vector-sincos.h" #include <immintrin.h> #define VEC_TYPE __m512d VECTOR_WRAPPER (WRAPPER_NAME (cos), _ZGVeN8v_cos) VECTOR_WRAPPER (WRAPPER_NAME (sin), _ZGVeN8v_sin) -VECTOR_WRAPPER_fFF (WRAPPER_NAME (sincos), _ZGVeN8vvv_sincos) VECTOR_WRAPPER (WRAPPER_NAME (log), _ZGVeN8v_log) VECTOR_WRAPPER (WRAPPER_NAME (exp), _ZGVeN8v_exp) VECTOR_WRAPPER_ff (WRAPPER_NAME (pow), _ZGVeN8vv_pow) + +#ifndef __ILP32__ +# define VEC_INT_TYPE __m512i +#else +# define VEC_INT_TYPE __m256i +#endif + +VECTOR_WRAPPER_fFF_2 (WRAPPER_NAME (sincos), _ZGVeN8vvv_sincos) diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx.c b/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx.c new file mode 100644 index 0000000000..5b45f0a055 --- /dev/null +++ b/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx.c @@ -0,0 +1 @@ +#include "test-float-libmvec-sincosf.c" diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx2.c b/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx2.c new file mode 100644 index 0000000000..5b45f0a055 --- /dev/null +++ b/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx2.c @@ -0,0 +1 @@ +#include "test-float-libmvec-sincosf.c" diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx512.c b/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx512.c new file mode 100644 index 0000000000..5b45f0a055 --- /dev/null +++ b/sysdeps/x86_64/fpu/test-float-libmvec-sincosf-avx512.c @@ -0,0 +1 @@ +#include "test-float-libmvec-sincosf.c" diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-sincosf.c b/sysdeps/x86_64/fpu/test-float-libmvec-sincosf.c new file mode 100644 index 0000000000..3b7aad877b --- /dev/null +++ b/sysdeps/x86_64/fpu/test-float-libmvec-sincosf.c @@ -0,0 +1,69 @@ +/* Test for vector sincosf ABI. + Copyright (C) 2016 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 <math.h> +#include <math-tests-arch.h> + +#define N 1000 +float x[N], s[N], c[N]; +float *s_ptrs[N]; +float *c_ptrs[N]; +int arch_check = 1; + +static void +init_arg (void) +{ + int i; + + CHECK_ARCH_EXT; + + arch_check = 0; + + for(i = 0; i < N; i++) + { + x[i] = i / 3; + s_ptrs[i] = &s[i]; + c_ptrs[i] = &c[i]; + } +} + +static int +test_sincosf_abi (void) +{ + int i; + + init_arg (); + + if (arch_check) + return 77; + +#pragma omp simd + for(i = 0; i < N; i++) + sincosf (x[i], s_ptrs[i], c_ptrs[i]); + + return 0; +} + +static int +do_test (void) +{ + return test_sincosf_abi (); +} + +#define TEST_FUNCTION do_test () +#include "../../../test-skeleton.c" diff --git a/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c index dc09e4a338..f3bf7dcc3e 100644 --- a/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c +++ b/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c @@ -17,13 +17,21 @@ <http://www.gnu.org/licenses/>. */ #include "test-float-vlen16.h" +#include "test-math-vector-sincos.h" #include <immintrin.h> #define VEC_TYPE __m512 VECTOR_WRAPPER (WRAPPER_NAME (cosf), _ZGVeN16v_cosf) VECTOR_WRAPPER (WRAPPER_NAME (sinf), _ZGVeN16v_sinf) -VECTOR_WRAPPER_fFF (WRAPPER_NAME (sincosf), _ZGVeN16vvv_sincosf) VECTOR_WRAPPER (WRAPPER_NAME (logf), _ZGVeN16v_logf) VECTOR_WRAPPER (WRAPPER_NAME (expf), _ZGVeN16v_expf) VECTOR_WRAPPER_ff (WRAPPER_NAME (powf), _ZGVeN16vv_powf) + +#define VEC_INT_TYPE __m512i + +#ifndef __ILP32__ +VECTOR_WRAPPER_fFF_3 (WRAPPER_NAME (sincosf), _ZGVeN16vvv_sincosf) +#else +VECTOR_WRAPPER_fFF_2 (WRAPPER_NAME (sincosf), _ZGVeN16vvv_sincosf) +#endif diff --git a/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c index 0bb9818146..4060f944c5 100644 --- a/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c +++ b/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c @@ -17,13 +17,21 @@ <http://www.gnu.org/licenses/>. */ #include "test-float-vlen4.h" +#include "test-math-vector-sincos.h" #include <immintrin.h> #define VEC_TYPE __m128 VECTOR_WRAPPER (WRAPPER_NAME (cosf), _ZGVbN4v_cosf) VECTOR_WRAPPER (WRAPPER_NAME (sinf), _ZGVbN4v_sinf) -VECTOR_WRAPPER_fFF (WRAPPER_NAME (sincosf), _ZGVbN4vvv_sincosf) VECTOR_WRAPPER (WRAPPER_NAME (logf), _ZGVbN4v_logf) VECTOR_WRAPPER (WRAPPER_NAME (expf), _ZGVbN4v_expf) VECTOR_WRAPPER_ff (WRAPPER_NAME (powf), _ZGVbN4vv_powf) + +#define VEC_INT_TYPE __m128i + +#ifndef __ILP32__ +VECTOR_WRAPPER_fFF_3 (WRAPPER_NAME (sincosf), _ZGVbN4vvv_sincosf) +#else +VECTOR_WRAPPER_fFF_2 (WRAPPER_NAME (sincosf), _ZGVbN4vvv_sincosf) +#endif diff --git a/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c index 4985ac2379..d1fc43225c 100644 --- a/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c +++ b/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c @@ -17,6 +17,7 @@ <http://www.gnu.org/licenses/>. */ #include "test-float-vlen8.h" +#include "test-math-vector-sincos.h" #include <immintrin.h> #undef VEC_SUFF @@ -26,7 +27,17 @@ VECTOR_WRAPPER (WRAPPER_NAME (cosf), _ZGVdN8v_cosf) VECTOR_WRAPPER (WRAPPER_NAME (sinf), _ZGVdN8v_sinf) -VECTOR_WRAPPER_fFF (WRAPPER_NAME (sincosf), _ZGVdN8vvv_sincosf) VECTOR_WRAPPER (WRAPPER_NAME (logf), _ZGVdN8v_logf) VECTOR_WRAPPER (WRAPPER_NAME (expf), _ZGVdN8v_expf) VECTOR_WRAPPER_ff (WRAPPER_NAME (powf), _ZGVdN8vv_powf) + +/* Redefinition of wrapper to be compatible with _ZGVdN8vvv_sincosf. */ +#undef VECTOR_WRAPPER_fFF + +#define VEC_INT_TYPE __m256i + +#ifndef __ILP32__ +VECTOR_WRAPPER_fFF_3 (WRAPPER_NAME (sincosf), _ZGVdN8vvv_sincosf) +#else +VECTOR_WRAPPER_fFF_2 (WRAPPER_NAME (sincosf), _ZGVdN8vvv_sincosf) +#endif diff --git a/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c index 9cc2883399..99b462afeb 100644 --- a/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c +++ b/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c @@ -17,13 +17,21 @@ <http://www.gnu.org/licenses/>. */ #include "test-float-vlen8.h" +#include "test-math-vector-sincos.h" #include <immintrin.h> #define VEC_TYPE __m256 VECTOR_WRAPPER (WRAPPER_NAME (cosf), _ZGVcN8v_cosf) VECTOR_WRAPPER (WRAPPER_NAME (sinf), _ZGVcN8v_sinf) -VECTOR_WRAPPER_fFF (WRAPPER_NAME (sincosf), _ZGVcN8vvv_sincosf) VECTOR_WRAPPER (WRAPPER_NAME (logf), _ZGVcN8v_logf) VECTOR_WRAPPER (WRAPPER_NAME (expf), _ZGVcN8v_expf) VECTOR_WRAPPER_ff (WRAPPER_NAME (powf), _ZGVcN8vv_powf) + +#define VEC_INT_TYPE __m128i + +#ifndef __ILP32__ +VECTOR_WRAPPER_fFF_4 (WRAPPER_NAME (sincosf), _ZGVcN8vvv_sincosf) +#else +VECTOR_WRAPPER_fFF_3 (WRAPPER_NAME (sincosf), _ZGVcN8vvv_sincosf) +#endif |