name string | code string | asm string | file string |
|---|---|---|---|
ncnn::ReLU_x86_fma::forward_inplace_int8(ncnn::Mat&, ncnn::Option const&) const | int ReLU_x86_fma::forward_inplace_int8(Mat& bottom_top_blob, const Option& opt) const
{
int w = bottom_top_blob.w;
int h = bottom_top_blob.h;
int d = bottom_top_blob.d;
int channels = bottom_top_blob.c;
int size = w * h * d;
int elempack = bottom_top_blob.elempack;
#if __SSE2__
if (elempack... | subq $0x2e8, %rsp # imm = 0x2E8
movq %rdi, 0x148(%rsp)
movq %rsi, 0x140(%rsp)
movq %rdx, 0x138(%rsp)
movq 0x148(%rsp), %rax
movq %rax, 0x50(%rsp)
movq 0x140(%rsp), %rax
movl 0x2c(%rax), %eax
movl %eax, 0x134(%rsp)
movq 0x140(%rsp), %rax
movl 0x30(%rax), %eax
movl %eax, 0x130(%rsp)
movq 0x140(%rsp), %rax
movl... | /Tencent[P]ncnn/build_O0/src/layer/x86/relu_x86_fma.cpp |
ncnn::Reshape::eval_shape_expr(std::vector<ncnn::Mat, std::allocator<ncnn::Mat>> const&, int&, int&, int&, int&) const | int Reshape::eval_shape_expr(const std::vector<Mat>& bottom_blobs, int& outw, int& outh, int& outd, int& outc) const
{
// [size(@0,0),size(@0,1),12,64]
std::vector<int> shape;
int er = eval_list_expression(shape_expr, bottom_blobs, shape);
if (er != 0)
return -1;
outw = 1;
outh = 1;
... | subq $0x88, %rsp
movq %rdi, 0x78(%rsp)
movq %rsi, 0x70(%rsp)
movq %rdx, 0x68(%rsp)
movq %rcx, 0x60(%rsp)
movq %r8, 0x58(%rsp)
movq %r9, 0x50(%rsp)
movq 0x78(%rsp), %rax
movq %rax, 0x8(%rsp)
leaq 0x38(%rsp), %rdi
movq %rdi, 0x10(%rsp)
callq 0x2fb20
movq 0x8(%rsp), %rdi
movq 0x10(%rsp), %rdx
addq $0xe8, %rdi
movq 0x70(%r... | /Tencent[P]ncnn/src/layer/reshape.cpp |
ncnn::Scale_x86::forward_inplace(std::vector<ncnn::Mat, std::allocator<ncnn::Mat>>&, ncnn::Option const&) const | int Scale_x86::forward_inplace(std::vector<Mat>& bottom_top_blobs, const Option& opt) const
{
Mat& bottom_top_blob = bottom_top_blobs[0];
const Mat& scale_blob = bottom_top_blobs[1];
const int w = bottom_top_blob.w;
const int h = bottom_top_blob.h;
const int d = bottom_top_blob.d;
const int cha... | subq $0x6d8, %rsp # imm = 0x6D8
movq %rdi, 0x2d0(%rsp)
movq %rsi, 0x2c8(%rsp)
movq %rdx, 0x2c0(%rsp)
movq 0x2d0(%rsp), %rax
movq %rax, 0x78(%rsp)
movq 0x2c8(%rsp), %rdi
xorl %eax, %eax
movl %eax, %esi
callq 0x6f570
movq %rax, 0x2b8(%rsp)
movq 0x2c8(%rsp), %rdi
movl $0x1, %esi
callq 0x6f570
movq %rax, %rcx
mo... | /Tencent[P]ncnn/src/layer/x86/scale_x86.cpp |
ncnn::Scale_x86_avx::forward_inplace(std::vector<ncnn::Mat, std::allocator<ncnn::Mat>>&, ncnn::Option const&) const | int Scale_x86_avx::forward_inplace(std::vector<Mat>& bottom_top_blobs, const Option& opt) const
{
Mat& bottom_top_blob = bottom_top_blobs[0];
const Mat& scale_blob = bottom_top_blobs[1];
const int w = bottom_top_blob.w;
const int h = bottom_top_blob.h;
const int d = bottom_top_blob.d;
const int... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0xe20, %rsp # imm = 0xE20
movq %rdi, 0x570(%rsp)
movq %rsi, 0x568(%rsp)
movq %rdx, 0x560(%rsp)
movq 0x570(%rsp), %rax
movq %rax, 0x128(%rsp)
movq 0x568(%rsp), %rdi
xorl %eax, %eax
movl %eax, %esi
callq 0x6f570
movq %rax, 0x558(%rsp)
movq 0x568(%rsp), %rdi
mo... | /Tencent[P]ncnn/build_O0/src/layer/x86/scale_x86_avx.cpp |
ncnn::Sigmoid_x86_avx512::forward_inplace(ncnn::Mat&, ncnn::Option const&) const | int Sigmoid_x86_avx512::forward_inplace(Mat& bottom_top_blob, const Option& opt) const
{
int w = bottom_top_blob.w;
int h = bottom_top_blob.h;
int d = bottom_top_blob.d;
int channels = bottom_top_blob.c;
int elempack = bottom_top_blob.elempack;
int size = w * h * d * elempack;
#pragma omp p... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x2480, %rsp # imm = 0x2480
movq %rdi, 0x280(%rsp)
movq %rsi, 0x278(%rsp)
movq %rdx, 0x270(%rsp)
movq 0x278(%rsp), %rax
movl 0x2c(%rax), %eax
movl %eax, 0x26c(%rsp)
movq 0x278(%rsp), %rax
movl 0x30(%rax), %eax
movl %eax, 0x268(%rsp)
movq 0x278(%rsp), %rax
mov... | /Tencent[P]ncnn/build_O0/src/layer/x86/sigmoid_x86_avx512.cpp |
ncnn::Softmax::forward_inplace(ncnn::Mat&, ncnn::Option const&) const | int Softmax::forward_inplace(Mat& bottom_top_blob, const Option& opt) const
{
// value = expf( value - global max value )
// sum all value
// value = value / sum
const int dims = bottom_top_blob.dims;
const int w = bottom_top_blob.w;
const int h = bottom_top_blob.h;
const int d = bottom_top... | subq $0x838, %rsp # imm = 0x838
movq %rdi, 0x390(%rsp)
movq %rsi, 0x388(%rsp)
movq %rdx, 0x380(%rsp)
movq 0x390(%rsp), %rax
movq %rax, 0xf0(%rsp)
movq 0x388(%rsp), %rcx
movl 0x28(%rcx), %ecx
movl %ecx, 0x37c(%rsp)
movq 0x388(%rsp), %rcx
movl 0x2c(%rcx), %ecx
movl %ecx, 0x378(%rsp)
movq 0x388(%rsp), %rcx
movl... | /Tencent[P]ncnn/src/layer/softmax.cpp |
ncnn::softmax_pack4(float*, int, int, int, float*, float*) | static void softmax_pack4(float* _ptr, int elemcount, int stride, int size1, float* _maxptr, float* _sumptr)
{
// reduce max
for (int i = 0; i < elemcount; i++)
{
const float* ptr = _ptr + i * stride;
float* maxptr = _maxptr;
int j = 0;
#if __AVX__
#if __AVX512F__
__m512i _p... | pushq %rbx
subq $0x26d0, %rsp # imm = 0x26D0
movq %rdi, 0x308(%rsp)
movl %esi, 0x304(%rsp)
movl %edx, 0x300(%rsp)
movl %ecx, 0x2fc(%rsp)
movq %r8, 0x2f0(%rsp)
movq %r9, 0x2e8(%rsp)
movl $0x0, 0x2e4(%rsp)
movl 0x2e4(%rsp), %eax
cmpl 0x304(%rsp), %eax
jge 0xee8b49
movq 0x308(%rsp), %rax
movl 0x2e4(%rsp), %ecx
i... | /Tencent[P]ncnn/src/layer/x86/softmax_x86.cpp |
ncnn::Softmax_x86_avx512::forward_inplace(ncnn::Mat&, ncnn::Option const&) const | int Softmax_x86_avx512::forward_inplace(Mat& bottom_top_blob, const Option& opt) const
{
const int dims = bottom_top_blob.dims;
const int w = bottom_top_blob.w;
const int h = bottom_top_blob.h;
const int d = bottom_top_blob.d;
const int channels = bottom_top_blob.c;
const int elempack = bottom_t... | subq $0x1068, %rsp # imm = 0x1068
movq %rdi, 0x788(%rsp)
movq %rsi, 0x780(%rsp)
movq %rdx, 0x778(%rsp)
movq 0x788(%rsp), %rax
movq %rax, 0x268(%rsp)
movq 0x780(%rsp), %rcx
movl 0x28(%rcx), %ecx
movl %ecx, 0x774(%rsp)
movq 0x780(%rsp), %rcx
movl 0x2c(%rcx), %ecx
movl %ecx, 0x770(%rsp)
movq 0x780(%rsp), %rcx
mo... | /Tencent[P]ncnn/build_O0/src/layer/x86/softmax_x86_avx512.cpp |
ncnn::softmax_pack4(float*, int, int, int, float*, float*) | static void softmax_pack4(float* _ptr, int elemcount, int stride, int size1, float* _maxptr, float* _sumptr)
{
// reduce max
for (int i = 0; i < elemcount; i++)
{
const float* ptr = _ptr + i * stride;
float* maxptr = _maxptr;
int j = 0;
#if __AVX__
#if __AVX512F__
__m512i _p... | pushq %rbp
movq %rsp, %rbp
pushq %r15
pushq %r14
pushq %rbx
andq $-0x20, %rsp
subq $0x5be0, %rsp # imm = 0x5BE0
movq %rdi, 0x700(%rsp)
movl %esi, 0x6fc(%rsp)
movl %edx, 0x6f8(%rsp)
movl %ecx, 0x6f4(%rsp)
movq %r8, 0x6e8(%rsp)
movq %r9, 0x6e0(%rsp)
movl $0x0, 0x6dc(%rsp)
movl 0x6dc(%rsp), %eax
cmpl 0x6fc(%rsp)... | /Tencent[P]ncnn/build_O0/src/layer/x86/softmax_x86_fma.cpp |
ncnn::LSTM::load_model(ncnn::ModelBin const&) | int LSTM::load_model(const ModelBin& mb)
{
int num_directions = direction == 2 ? 2 : 1;
int size = weight_data_size / num_directions / hidden_size / 4;
// raw weight data
weight_xc_data = mb.load(size, hidden_size * 4, num_directions, 0);
if (weight_xc_data.empty())
return -100;
bias_... | subq $0x5f8, %rsp # imm = 0x5F8
movq %rdi, 0x2d8(%rsp)
movq %rsi, 0x2d0(%rsp)
movq 0x2d8(%rsp), %rcx
movq %rcx, 0xf8(%rsp)
movl 0xd8(%rcx), %esi
movl $0x1, %eax
movl $0x2, %edx
cmpl $0x2, %esi
cmovel %edx, %eax
movl %eax, 0x2cc(%rsp)
movl 0xd4(%rcx), %eax
cltd
idivl 0x2cc(%rsp)
cltd
idivl 0xdc(%rcx)
movl $0x... | /Tencent[P]ncnn/src/layer/lstm.cpp |
ncnn::lstm(ncnn::Mat const&, ncnn::Mat&, int, ncnn::Mat const&, ncnn::Mat const&, ncnn::Mat const&, ncnn::Mat const&, ncnn::Mat&, ncnn::Mat&, ncnn::Option const&) | static int lstm(const Mat& bottom_blob, Mat& top_blob, int reverse, const Mat& weight_xc, const Mat& bias_c, const Mat& weight_hc, const Mat& weight_hr, Mat& hidden_state, Mat& cell_state, const Option& opt)
{
int size = bottom_blob.w;
int T = bottom_blob.h;
int num_output = top_blob.w;
int hidden_size... | subq $0x4a8, %rsp # imm = 0x4A8
movq 0x4c8(%rsp), %rax
movq 0x4c0(%rsp), %rax
movq 0x4b8(%rsp), %rax
movq 0x4b0(%rsp), %rax
movq %rdi, 0x250(%rsp)
movq %rsi, 0x248(%rsp)
movl %edx, 0x244(%rsp)
movq %rcx, 0x238(%rsp)
movq %r8, 0x230(%rsp)
movq %r9, 0x228(%rsp)
movq 0x250(%rsp), %rax
movl 0x2c(%rax), %eax
movl... | /Tencent[P]ncnn/src/layer/lstm.cpp |
ncnn::LSTM_x86::LSTM_x86() | LSTM_x86::LSTM_x86()
{
one_blob_only = false;
support_inplace = false;
} | subq $0xf8, %rsp
movq %rdi, 0x48(%rsp)
movq 0x48(%rsp), %rdi
movq %rdi, 0x28(%rsp)
callq 0xf72000
movq 0x28(%rsp), %rax
leaq 0xf931d1(%rip), %rcx # 0x1f213e8
addq $0x10, %rcx
movq %rcx, (%rax)
addq $0x298, %rax # imm = 0x298
movq %rax, 0x50(%rsp)
movq 0x50(%rsp), %rax
movq %rax, 0x30(%rsp)
movq $0x0, (%ra... | /Tencent[P]ncnn/src/layer/x86/lstm_x86.cpp |
ncnn::lstm_dynamic_quantize_scale2int8(float const*, int, float, signed char*) | static void lstm_dynamic_quantize_scale2int8(const float* ptr, int size, float scale, signed char* outptr)
{
#if NCNN_RUNTIME_CPU && NCNN_AVX512VNNI && __AVX512F__ && !__AVX512VNNI__
if (ncnn::cpu_support_x86_avx512_vnni())
{
lstm_dynamic_quantize_scale2int8_avx512vnni(ptr, size, scale, outptr);
... | subq $0x2f8, %rsp # imm = 0x2F8
movq %rdi, 0x50(%rsp)
movl %esi, 0x4c(%rsp)
movss %xmm0, 0x48(%rsp)
movq %rdx, 0x40(%rsp)
movl $0x0, 0x3c(%rsp)
movss 0x48(%rsp), %xmm0
movss %xmm0, 0x90(%rsp)
movaps 0x90(%rsp), %xmm0
shufps $0x0, %xmm0, %xmm0 # xmm0 = xmm0[0,0,0,0]
movaps %xmm0, 0x80(%rsp)
movaps 0x80(%... | /Tencent[P]ncnn/src/layer/x86/lstm_int8.h |
ncnn::lstm_dynamic_quantize(ncnn::Mat const&, ncnn::Mat&, ncnn::Mat&, ncnn::Option const&) | static void lstm_dynamic_quantize(const Mat& bottom_blob, Mat& bottom_blob_int8, Mat& bottom_blob_int8_descales, const Option& opt)
{
int size = bottom_blob.w;
int T = bottom_blob.h;
// dynamic quantize bottom_blob
bottom_blob_int8_descales.create(T, (size_t)4u, 1, opt.blob_allocator);
bottom_blob... | subq $0x78, %rsp
movq %rdi, 0x40(%rsp)
movq %rsi, 0x38(%rsp)
movq %rdx, 0x30(%rsp)
movq %rcx, 0x28(%rsp)
movq 0x40(%rsp), %rax
movl 0x2c(%rax), %eax
movl %eax, 0x24(%rsp)
movq 0x40(%rsp), %rax
movl 0x30(%rax), %eax
movl %eax, 0x20(%rsp)
movq 0x30(%rsp), %rdi
movl 0x20(%rsp), %esi
movq 0x28(%rsp), %rax
movq 0x8(%rax), %... | /Tencent[P]ncnn/build_O0/src/layer/x86/lstm_x86_avx512.cpp |
ncnn::LSTM_x86_avx::forward_int8(std::vector<ncnn::Mat, std::allocator<ncnn::Mat>> const&, std::vector<ncnn::Mat, std::allocator<ncnn::Mat>>&, ncnn::Option const&) const | int LSTM_x86_avx::forward_int8(const std::vector<Mat>& bottom_blobs, std::vector<Mat>& top_blobs, const Option& opt) const
{
const Mat& bottom_blob = bottom_blobs[0];
int T = bottom_blob.h;
int num_directions = direction == 2 ? 2 : 1;
Mat hidden;
Mat cell;
Allocator* hidden_cell_allocator = to... | subq $0x1ad8, %rsp # imm = 0x1AD8
movq %rdi, 0xb80(%rsp)
movq %rsi, 0xb78(%rsp)
movq %rdx, 0xb70(%rsp)
movq %rcx, 0xb68(%rsp)
movq 0xb80(%rsp), %rax
movq %rax, 0x3f8(%rsp)
movq 0xb78(%rsp), %rdi
xorl %eax, %eax
movl %eax, %esi
callq 0x89920
movq %rax, %rcx
movq 0x3f8(%rsp), %rax
movq %rcx, 0xb60(%rsp)
movq 0x... | /Tencent[P]ncnn/build_O0/src/layer/x86/lstm_x86_avx.cpp |
ncnn::lstm_dynamic_quantize_scale2int8(float const*, int, float, signed char*) | static void lstm_dynamic_quantize_scale2int8(const float* ptr, int size, float scale, signed char* outptr)
{
#if NCNN_RUNTIME_CPU && NCNN_AVX512VNNI && __AVX512F__ && !__AVX512VNNI__
if (ncnn::cpu_support_x86_avx512_vnni())
{
lstm_dynamic_quantize_scale2int8_avx512vnni(ptr, size, scale, outptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x7a0, %rsp # imm = 0x7A0
movq %rdi, 0xa8(%rsp)
movl %esi, 0xa4(%rsp)
vmovss %xmm0, 0xa0(%rsp)
movq %rdx, 0x98(%rsp)
callq 0x7e290
cmpl $0x0, %eax
je 0x106a963
movq 0xa8(%rsp), %rdi
movl 0xa4(%rsp), %esi
vmovss 0xa0(%rsp), %xmm0
movq 0x98(%rsp), %rdx
callq 0... | /Tencent[P]ncnn/src/layer/x86/lstm_int8.h |
ncnn::lstm_dynamic_quantize_get_absmax(float const*, int) | static float lstm_dynamic_quantize_get_absmax(const float* ptr, int size)
{
float absmax = 0.f;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _absmax_avx512 = _mm512_set1_ps(0.f);
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
_absmax_avx512 = _mm512... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x4e0, %rsp # imm = 0x4E0
movq %rdi, 0xa8(%rsp)
movl %esi, 0xa4(%rsp)
movl $0x0, 0xa0(%rsp)
movl $0x0, 0x9c(%rsp)
movl $0x0, 0xb4(%rsp)
vmovss 0xb4(%rsp), %xmm0
vmovss %xmm0, 0x4(%rsp)
vmovss %xmm0, 0x4cc(%rsp)
vmovss %xmm0, 0x4c8(%rsp)
vmovss %xmm0, 0x4c4(%... | /Tencent[P]ncnn/src/layer/x86/lstm_int8.h |
ncnn::binary_op_scalar_inplace(ncnn::Mat&, float, int, ncnn::Option const&) | static void binary_op_scalar_inplace(Mat& bottom_top_blob, float b, int op_type, const Option& opt)
{
if (op_type == BinaryOp::Operation_ADD) return binary_op_scalar_inplace<binary_op_add>(bottom_top_blob, b, opt);
if (op_type == BinaryOp::Operation_SUB) return binary_op_scalar_inplace<binary_op_sub>(bottom_top... | subq $0x18, %rsp
movq %rdi, 0x10(%rsp)
movss %xmm0, 0xc(%rsp)
movl %esi, 0x8(%rsp)
movq %rdx, (%rsp)
cmpl $0x0, 0x8(%rsp)
jne 0x10d3777
movq 0x10(%rsp), %rdi
movss 0xc(%rsp), %xmm0
movq (%rsp), %rsi
callq 0x10e0090
jmp 0x10d38c6
cmpl $0x1, 0x8(%rsp)
jne 0x10d3797
movq 0x10(%rsp), %rdi
movss 0xc(%rsp), %xmm0
movq (%rsp)... | /Tencent[P]ncnn/src/layer/binaryop.cpp |
void ncnn::binary_op_scalar_inplace<ncnn::binary_op_max>(ncnn::Mat&, float, ncnn::Option const&) | static void binary_op_scalar_inplace(Mat& a, float b, const Option& opt)
{
const Op op;
const int channels = a.c;
const int size = a.w * a.h * a.d;
#pragma omp parallel for num_threads(opt.num_threads)
for (int q = 0; q < channels; q++)
{
float* ptr = a.channel(q);
for (int i ... | subq $0x158, %rsp # imm = 0x158
movq %rdi, 0xa8(%rsp)
movss %xmm0, 0xa4(%rsp)
movq %rsi, 0x98(%rsp)
movq 0xa8(%rsp), %rax
movl 0x38(%rax), %eax
movl %eax, 0x90(%rsp)
movq 0xa8(%rsp), %rax
movl 0x2c(%rax), %eax
movq 0xa8(%rsp), %rcx
imull 0x30(%rcx), %eax
movq 0xa8(%rsp), %rcx
imull 0x34(%rcx), %eax
movl %eax... | /Tencent[P]ncnn/src/layer/binaryop.cpp |
void ncnn::binary_op_scalar_inplace<ncnn::binary_op_rsub>(ncnn::Mat&, float, ncnn::Option const&) | static void binary_op_scalar_inplace(Mat& a, float b, const Option& opt)
{
const Op op;
const int channels = a.c;
const int size = a.w * a.h * a.d;
#pragma omp parallel for num_threads(opt.num_threads)
for (int q = 0; q < channels; q++)
{
float* ptr = a.channel(q);
for (int i ... | subq $0x158, %rsp # imm = 0x158
movq %rdi, 0xa8(%rsp)
movss %xmm0, 0xa4(%rsp)
movq %rsi, 0x98(%rsp)
movq 0xa8(%rsp), %rax
movl 0x38(%rax), %eax
movl %eax, 0x90(%rsp)
movq 0xa8(%rsp), %rax
movl 0x2c(%rax), %eax
movq 0xa8(%rsp), %rcx
imull 0x30(%rcx), %eax
movq 0xa8(%rsp), %rcx
imull 0x34(%rcx), %eax
movl %eax... | /Tencent[P]ncnn/src/layer/binaryop.cpp |
ncnn::binary_op_broadcast(ncnn::Mat const&, ncnn::Mat const&, ncnn::Mat&, int, ncnn::Option const&) | static void binary_op_broadcast(const Mat& a, const Mat& b, Mat& c, int op_type, const Option& opt)
{
if (b.w * b.h * b.d * b.c * b.elempack == 1)
{
return binary_op_scalar(a, b[0], c, op_type, opt);
}
if (a.dims == b.dims && a.w == b.w && a.h == b.h && a.d == b.d && a.c == b.c && a.elempack ==... | subq $0x11e8, %rsp # imm = 0x11E8
movq %rdi, 0x750(%rsp)
movq %rsi, 0x748(%rsp)
movq %rdx, 0x740(%rsp)
movl %ecx, 0x73c(%rsp)
movq %r8, 0x730(%rsp)
movq 0x748(%rsp), %rax
movl 0x2c(%rax), %eax
movq 0x748(%rsp), %rcx
imull 0x30(%rcx), %eax
movq 0x748(%rsp), %rcx
imull 0x34(%rcx), %eax
movq 0x748(%rsp), %rcx
im... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
ncnn::binary_op_vector(float const*, float const*, float*, int, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp, int op_type)
{
using namespace BinaryOp_x86_functor;
if (op_type == BinaryOp::Operation_ADD) return binary_op_vector<binary_op_add>(ptr, ptr1, outptr, aw, bw, ap, bp);
if (op_type == BinaryOp::O... | subq $0x38, %rsp
movl 0x48(%rsp), %eax
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
cmpl $0x0, 0x48(%rsp)
jne 0x10ee2fe
movq 0x30(%rsp), %rdi
movq 0x28(%rsp), %rsi
movq 0x20(%rsp), %rdx
movl 0x1c(%rsp), %ecx
mov... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector<ncnn::BinaryOp_x86_functor::binary_op_add>(float const*, float const*, float*, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp)
{
const int w = std::max(aw, bw);
const int elempack = std::max(ap, bp);
const int size = w * elempack;
if (ap == bp)
{
if (aw == bw)
{
// no broadcast
... | subq $0x38, %rsp
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
leaq 0x1c(%rsp), %rdi
leaq 0x18(%rsp), %rsi
callq 0x53350
movl (%rax), %eax
movl %eax, 0x10(%rsp)
leaq 0x14(%rsp), %rdi
leaq 0x40(%rsp), %rsi
callq 0... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector<ncnn::BinaryOp_x86_functor::binary_op_mul>(float const*, float const*, float*, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp)
{
const int w = std::max(aw, bw);
const int elempack = std::max(ap, bp);
const int size = w * elempack;
if (ap == bp)
{
if (aw == bw)
{
// no broadcast
... | subq $0x38, %rsp
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
leaq 0x1c(%rsp), %rdi
leaq 0x18(%rsp), %rsi
callq 0x53350
movl (%rax), %eax
movl %eax, 0x10(%rsp)
leaq 0x14(%rsp), %rdi
leaq 0x40(%rsp), %rsi
callq 0... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector<ncnn::BinaryOp_x86_functor::binary_op_rsub>(float const*, float const*, float*, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp)
{
const int w = std::max(aw, bw);
const int elempack = std::max(ap, bp);
const int size = w * elempack;
if (ap == bp)
{
if (aw == bw)
{
// no broadcast
... | subq $0x38, %rsp
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
leaq 0x1c(%rsp), %rdi
leaq 0x18(%rsp), %rsi
callq 0x53350
movl (%rax), %eax
movl %eax, 0x10(%rsp)
leaq 0x14(%rsp), %rdi
leaq 0x40(%rsp), %rsi
callq 0... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_functor::binary_op_add>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | subq $0x88, %rsp
movq %rdi, -0x10(%rsp)
movq %rsi, -0x18(%rsp)
movq %rdx, -0x20(%rsp)
movl %ecx, -0x24(%rsp)
movl %r8d, -0x28(%rsp)
movq -0x18(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, -0x30(%rsp)
movl $0x0, -0x34(%rsp)
cmpl $0x4, -0x28(%rsp)
jne 0x10ef854
movq -0x18(%rsp), %rax
movq %rax, (%rsp)
movq (%rsp), %rax
m... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_functor::binary_op_sub>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | subq $0x68, %rsp
movq %rdi, -0x20(%rsp)
movq %rsi, -0x28(%rsp)
movq %rdx, -0x30(%rsp)
movl %ecx, -0x34(%rsp)
movl %r8d, -0x38(%rsp)
movl -0x34(%rsp), %eax
movl -0x38(%rsp), %ecx
imull %ecx, %eax
movl %eax, -0x40(%rsp)
movl $0x0, -0x44(%rsp)
movq -0x28(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, 0x20(%rsp)
movaps 0x20(... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_functor::binary_op_mul>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | subq $0x98, %rsp
movq %rdi, -0x18(%rsp)
movq %rsi, -0x20(%rsp)
movq %rdx, -0x28(%rsp)
movl %ecx, -0x2c(%rsp)
movl %r8d, -0x30(%rsp)
movq -0x20(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, -0x38(%rsp)
movl $0x0, -0x3c(%rsp)
cmpl $0x4, -0x30(%rsp)
jne 0x10f0956
movq -0x20(%rsp), %rax
movq %rax, -0x8(%rsp)
movq -0x8(%rsp)... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_functor::binary_op_mul>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | subq $0x68, %rsp
movq %rdi, -0x20(%rsp)
movq %rsi, -0x28(%rsp)
movq %rdx, -0x30(%rsp)
movl %ecx, -0x34(%rsp)
movl %r8d, -0x38(%rsp)
movl -0x34(%rsp), %eax
movl -0x38(%rsp), %ecx
imull %ecx, %eax
movl %eax, -0x40(%rsp)
movl $0x0, -0x44(%rsp)
movq -0x28(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, 0x20(%rsp)
movaps 0x20(... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_functor::binary_op_div>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | subq $0x58, %rsp
movq %rdi, -0x30(%rsp)
movq %rsi, -0x38(%rsp)
movq %rdx, -0x40(%rsp)
movl %ecx, -0x44(%rsp)
movl %r8d, -0x48(%rsp)
cmpl $0x4, -0x48(%rsp)
jne 0x10f161d
movl $0x0, -0x50(%rsp)
movl -0x50(%rsp), %eax
cmpl -0x44(%rsp), %eax
jge 0x10f161b
movq -0x30(%rsp), %rax
movq %rax, -0x28(%rsp)
movq -0x28(%rsp), %rax... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_functor::binary_op_max>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | subq $0x68, %rsp
movq %rdi, -0x20(%rsp)
movq %rsi, -0x28(%rsp)
movq %rdx, -0x30(%rsp)
movl %ecx, -0x34(%rsp)
movl %r8d, -0x38(%rsp)
movl -0x34(%rsp), %eax
movl -0x38(%rsp), %ecx
imull %ecx, %eax
movl %eax, -0x40(%rsp)
movl $0x0, -0x44(%rsp)
movq -0x28(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, 0x20(%rsp)
movaps 0x20(... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_functor::binary_op_min>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | subq $0x118, %rsp # imm = 0x118
movq %rdi, 0x68(%rsp)
movq %rsi, 0x60(%rsp)
movq %rdx, 0x58(%rsp)
movl %ecx, 0x54(%rsp)
movl %r8d, 0x50(%rsp)
movq 0x68(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, 0x48(%rsp)
movl $0x0, 0x44(%rsp)
cmpl $0x4, 0x50(%rsp)
jne 0x10f25b5
movq 0x68(%rsp), %rax
movq %rax, 0x78(%rsp)... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_functor::binary_op_pow>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | subq $0xdc8, %rsp # imm = 0xDC8
movq %rdi, 0x88(%rsp)
movq %rsi, 0x80(%rsp)
movq %rdx, 0x78(%rsp)
movl %ecx, 0x74(%rsp)
movl %r8d, 0x70(%rsp)
movq 0x88(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, 0x68(%rsp)
movl $0x0, 0x64(%rsp)
cmpl $0x4, 0x70(%rsp)
jne 0x10f5508
movq 0x88(%rsp), %rax
movq %rax, 0x98(%rsp)... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_functor::binary_op_rsub>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | subq $0x88, %rsp
movq %rdi, -0x18(%rsp)
movq %rsi, -0x20(%rsp)
movq %rdx, -0x28(%rsp)
movl %ecx, -0x2c(%rsp)
movl %r8d, -0x30(%rsp)
movq -0x20(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, -0x38(%rsp)
movl $0x0, -0x3c(%rsp)
cmpl $0x4, -0x30(%rsp)
jne 0x10fa796
movq -0x20(%rsp), %rax
movq %rax, -0x8(%rsp)
movq -0x8(%rsp)... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_no_broadcast<ncnn::BinaryOp_x86_functor::binary_op_rdiv>(float const*, float const*, float*, int) | static void binary_op_vector_no_broadcast(const float* ptr, const float* ptr1, float* outptr, int size)
{
const Op op;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
__m512 _b = _mm512_loadu_ps(ptr1);
__m512... | subq $0x58, %rsp
movq %rdi, -0x28(%rsp)
movq %rsi, -0x30(%rsp)
movq %rdx, -0x38(%rsp)
movl %ecx, -0x3c(%rsp)
movl $0x0, -0x44(%rsp)
movl -0x44(%rsp), %eax
addl $0x3, %eax
cmpl -0x3c(%rsp), %eax
jge 0x10faf2e
movq -0x28(%rsp), %rax
movq %rax, -0x18(%rsp)
movq -0x18(%rsp), %rax
movups (%rax), %xmm0
movaps %xmm0, -0x60(%r... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_functor::binary_op_rpow>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | subq $0xdb8, %rsp # imm = 0xDB8
movq %rdi, 0x88(%rsp)
movq %rsi, 0x80(%rsp)
movq %rdx, 0x78(%rsp)
movl %ecx, 0x74(%rsp)
movl %r8d, 0x70(%rsp)
movq 0x80(%rsp), %rax
movss (%rax), %xmm0
movss %xmm0, 0x68(%rsp)
movl $0x0, 0x64(%rsp)
cmpl $0x4, 0x70(%rsp)
jne 0x10fcbc8
movq 0x80(%rsp), %rax
movq %rax, 0x98(%rsp)... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | subq $0xd38, %rsp # imm = 0xD38
movq %rdi, -0x20(%rsp)
movq %rsi, -0x28(%rsp)
movq %rdx, -0x30(%rsp)
movl %ecx, -0x34(%rsp)
movl %r8d, -0x38(%rsp)
movl -0x34(%rsp), %eax
movl -0x38(%rsp), %ecx
imull %ecx, %eax
movl %eax, -0x40(%rsp)
movl $0x0, -0x44(%rsp)
movq -0x28(%rsp), %rax
movss (%rax), %xmm0
movss %xmm... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | subq $0xd28, %rsp # imm = 0xD28
movq %rdi, -0x30(%rsp)
movq %rsi, -0x38(%rsp)
movq %rdx, -0x40(%rsp)
movl %ecx, -0x44(%rsp)
movl %r8d, -0x48(%rsp)
cmpl $0x4, -0x48(%rsp)
jne 0x1110f01
movl $0x0, -0x50(%rsp)
movq -0x30(%rsp), %rax
movq %rax, -0x28(%rsp)
movq -0x28(%rsp), %rax
movups (%rax), %xmm0
movaps %xmm0... | /Tencent[P]ncnn/src/layer/x86/binaryop_x86.cpp |
ncnn::binary_op_scalar(ncnn::Mat const&, float, ncnn::Mat&, int, ncnn::Option const&) | static void binary_op_scalar(const Mat& a, float b, Mat& c, int op_type, const Option& opt)
{
const int channels = a.c;
const int size = a.w * a.h * a.d * a.elempack;
#pragma omp parallel for num_threads(opt.num_threads)
for (int q = 0; q < channels; q++)
{
const float* ptr = a.channel(q);
... | subq $0x2d8, %rsp # imm = 0x2D8
movq %rdi, 0x140(%rsp)
vmovss %xmm0, 0x13c(%rsp)
movq %rsi, 0x130(%rsp)
movl %edx, 0x12c(%rsp)
movq %rcx, 0x120(%rsp)
movq 0x140(%rsp), %rax
movl 0x38(%rax), %eax
movl %eax, 0x11c(%rsp)
movq 0x140(%rsp), %rax
movl 0x2c(%rax), %eax
movq 0x140(%rsp), %rcx
imull 0x30(%rcx), %eax
... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector<ncnn::BinaryOp_x86_avx512_functor::binary_op_add>(float const*, float const*, float*, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp)
{
const int w = std::max(aw, bw);
const int elempack = std::max(ap, bp);
const int size = w * elempack;
if (ap == bp)
{
if (aw == bw)
{
// no broadcast
... | subq $0x38, %rsp
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
leaq 0x1c(%rsp), %rdi
leaq 0x18(%rsp), %rsi
callq 0x53350
movl (%rax), %eax
movl %eax, 0x10(%rsp)
leaq 0x14(%rsp), %rdi
leaq 0x40(%rsp), %rsi
callq 0... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_no_broadcast<ncnn::BinaryOp_x86_avx512_functor::binary_op_add>(float const*, float const*, float*, int) | static void binary_op_vector_no_broadcast(const float* ptr, const float* ptr1, float* outptr, int size)
{
const Op op;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
__m512 _b = _mm512_loadu_ps(ptr1);
__m512... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x1d0(%rsp)
movq %rsi, 0x1c8(%rsp)
movq %rdx, 0x1c0(%rsp)
movl %ecx, 0x1bc(%rsp)
movl $0x0, 0x1b4(%rsp)
movl 0x1b4(%rsp), %eax
addl $0xf, %eax
cmpl 0x1bc(%rsp), %eax
jge 0x111c858
movq 0x1d0(%rsp), %rax
movq %rax, 0x1e0(%... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx512_functor::binary_op_add>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x500, %rsp # imm = 0x500
movq %rdi, 0x250(%rsp)
movq %rsi, 0x248(%rsp)
movq %rdx, 0x240(%rsp)
movl %ecx, 0x23c(%rsp)
movl %r8d, 0x238(%rsp)
movq 0x250(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x230(%rsp)
movl $0x0, 0x22c(%rsp)
cmpl $0x4, 0x238(%rsp)
j... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx512_functor::binary_op_sub>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x4c0, %rsp # imm = 0x4C0
movq %rdi, 0x228(%rsp)
movq %rsi, 0x220(%rsp)
movq %rdx, 0x218(%rsp)
movl %ecx, 0x214(%rsp)
movl %r8d, 0x210(%rsp)
movq 0x228(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x208(%rsp)
movl $0x0, 0x204(%rsp)
cmpl $0x4, 0x210(%rsp)
j... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_avx512_functor::binary_op_sub>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x4c0, %rsp # imm = 0x4C0
movq %rdi, 0x1b0(%rsp)
movq %rsi, 0x1a8(%rsp)
movq %rdx, 0x1a0(%rsp)
movl %ecx, 0x19c(%rsp)
movl %r8d, 0x198(%rsp)
movl 0x19c(%rsp), %eax
movl 0x198(%rsp), %ecx
imull %ecx, %eax
movl %eax, 0x190(%rsp)
movl $0x0, 0x18c(%rsp)
movq 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx512_functor::binary_op_mul>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x4c0, %rsp # imm = 0x4C0
movq %rdi, 0x228(%rsp)
movq %rsi, 0x220(%rsp)
movq %rdx, 0x218(%rsp)
movl %ecx, 0x214(%rsp)
movl %r8d, 0x210(%rsp)
movq 0x228(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x208(%rsp)
movl $0x0, 0x204(%rsp)
cmpl $0x4, 0x210(%rsp)
j... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_avx512_functor::binary_op_div>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x4c0, %rsp # imm = 0x4C0
movq %rdi, 0x228(%rsp)
movq %rsi, 0x220(%rsp)
movq %rdx, 0x218(%rsp)
movl %ecx, 0x214(%rsp)
movl %r8d, 0x210(%rsp)
movq 0x220(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x208(%rsp)
movl $0x0, 0x204(%rsp)
cmpl $0x4, 0x210(%rsp)
j... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_avx512_functor::binary_op_max>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x4c0, %rsp # imm = 0x4C0
movq %rdi, 0x1b0(%rsp)
movq %rsi, 0x1a8(%rsp)
movq %rdx, 0x1a0(%rsp)
movl %ecx, 0x19c(%rsp)
movl %r8d, 0x198(%rsp)
movl 0x19c(%rsp), %eax
movl 0x198(%rsp), %ecx
imull %ecx, %eax
movl %eax, 0x190(%rsp)
movl $0x0, 0x18c(%rsp)
movq 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_avx512_functor::binary_op_min>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x4c0, %rsp # imm = 0x4C0
movq %rdi, 0x1b0(%rsp)
movq %rsi, 0x1a8(%rsp)
movq %rdx, 0x1a0(%rsp)
movl %ecx, 0x19c(%rsp)
movl %r8d, 0x198(%rsp)
movl 0x19c(%rsp), %eax
movl 0x198(%rsp), %ecx
imull %ecx, %eax
movl %eax, 0x190(%rsp)
movl $0x0, 0x18c(%rsp)
movq 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_avx512_functor::binary_op_pow>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0xaa40, %rsp # imm = 0xAA40
movq %rdi, 0x520(%rsp)
movq %rsi, 0x518(%rsp)
movq %rdx, 0x510(%rsp)
movl %ecx, 0x50c(%rsp)
movl %r8d, 0x508(%rsp)
cmpl $0x10, 0x508(%rsp)
jne 0x113b5ee
movl $0x0, 0x500(%rsp)
movl 0x500(%rsp), %eax
cmpl 0x50c(%rsp), %eax
jge 0x113... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_avx512_functor::binary_op_rsub>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0xd00, %rsp # imm = 0xD00
movq %rdi, 0x4b0(%rsp)
movq %rsi, 0x4a8(%rsp)
movq %rdx, 0x4a0(%rsp)
movl %ecx, 0x49c(%rsp)
movl %r8d, 0x498(%rsp)
cmpl $0x10, 0x498(%rsp)
jne 0x114fd8b
movl $0x0, 0x490(%rsp)
movq 0x4b0(%rsp), %rax
movq %rax, 0x4b8(%rsp)
movq 0x4b8... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx512_functor::binary_op_atan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x26c0, %rsp # imm = 0x26C0
movq %rdi, 0x2a8(%rsp)
movq %rsi, 0x2a0(%rsp)
movq %rdx, 0x298(%rsp)
movl %ecx, 0x294(%rsp)
movl %r8d, 0x290(%rsp)
movq 0x2a8(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x288(%rsp)
movl $0x0, 0x284(%rsp)
cmpl $0x4, 0x290(%rsp)
... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_no_broadcast<ncnn::BinaryOp_x86_avx512_functor::binary_op_ratan2>(float const*, float const*, float*, int) | static void binary_op_vector_no_broadcast(const float* ptr, const float* ptr1, float* outptr, int size)
{
const Op op;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
__m512 _b = _mm512_loadu_ps(ptr1);
__m512... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x25c0, %rsp # imm = 0x25C0
movq %rdi, 0x1e8(%rsp)
movq %rsi, 0x1e0(%rsp)
movq %rdx, 0x1d8(%rsp)
movl %ecx, 0x1d4(%rsp)
movl $0x0, 0x1cc(%rsp)
movl 0x1cc(%rsp), %eax
addl $0xf, %eax
cmpl 0x1d4(%rsp), %eax
jge 0x118b24d
movq 0x1e8(%rsp), %rax
movq %rax, 0x1f8(... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx512_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x2700, %rsp # imm = 0x2700
movq %rdi, 0x2a8(%rsp)
movq %rsi, 0x2a0(%rsp)
movq %rdx, 0x298(%rsp)
movl %ecx, 0x294(%rsp)
movl %r8d, 0x290(%rsp)
movq 0x2a8(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x288(%rsp)
movl $0x0, 0x284(%rsp)
cmpl $0x4, 0x290(%rsp)
... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_avx512_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x4340, %rsp # imm = 0x4340
movq %rdi, 0x560(%rsp)
movq %rsi, 0x558(%rsp)
movq %rdx, 0x550(%rsp)
movl %ecx, 0x54c(%rsp)
movl %r8d, 0x548(%rsp)
cmpl $0x10, 0x548(%rsp)
jne 0x1194da4
movl $0x0, 0x540(%rsp)
movl 0x540(%rsp), %eax
cmpl 0x54c(%rsp), %eax
jge 0x119... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_avx512_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | pushq %rbp
movq %rsp, %rbp
andq $-0x40, %rsp
subq $0x4380, %rsp # imm = 0x4380
movq %rdi, 0x530(%rsp)
movq %rsi, 0x528(%rsp)
movq %rdx, 0x520(%rsp)
movl %ecx, 0x51c(%rsp)
movl %r8d, 0x518(%rsp)
cmpl $0x10, 0x518(%rsp)
jne 0x119d930
movl $0x0, 0x510(%rsp)
movq 0x530(%rsp), %rax
movq %rax, 0x538(%rsp)
movq 0x53... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx512.cpp |
ncnn::binary_op_broadcast(ncnn::Mat const&, ncnn::Mat const&, ncnn::Mat&, int, ncnn::Option const&) | static void binary_op_broadcast(const Mat& a, const Mat& b, Mat& c, int op_type, const Option& opt)
{
if (b.w * b.h * b.d * b.c * b.elempack == 1)
{
return binary_op_scalar(a, b[0], c, op_type, opt);
}
if (a.dims == b.dims && a.w == b.w && a.h == b.h && a.d == b.d && a.c == b.c && a.elempack ==... | subq $0x11e8, %rsp # imm = 0x11E8
movq %rdi, 0x750(%rsp)
movq %rsi, 0x748(%rsp)
movq %rdx, 0x740(%rsp)
movl %ecx, 0x73c(%rsp)
movq %r8, 0x730(%rsp)
movq 0x748(%rsp), %rax
movl 0x2c(%rax), %eax
movq 0x748(%rsp), %rcx
imull 0x30(%rcx), %eax
movq 0x748(%rsp), %rcx
imull 0x34(%rcx), %eax
movq 0x748(%rsp), %rcx
im... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
ncnn::binary_op_scalar_inplace(ncnn::Mat&, float, int, ncnn::Option const&) | static void binary_op_scalar_inplace(Mat& a, float b, int op_type, const Option& opt)
{
const int channels = a.c;
const int size = a.w * a.h * a.d * a.elempack;
#pragma omp parallel for num_threads(opt.num_threads)
for (int q = 0; q < channels; q++)
{
float* ptr = a.channel(q);
bin... | subq $0x158, %rsp # imm = 0x158
movq %rdi, 0xa8(%rsp)
vmovss %xmm0, 0xa4(%rsp)
movl %esi, 0xa0(%rsp)
movq %rdx, 0x98(%rsp)
movq 0xa8(%rsp), %rax
movl 0x38(%rax), %eax
movl %eax, 0x94(%rsp)
movq 0xa8(%rsp), %rax
movl 0x2c(%rax), %eax
movq 0xa8(%rsp), %rcx
imull 0x30(%rcx), %eax
movq 0xa8(%rsp), %rcx
imull 0x3... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector<ncnn::BinaryOp_x86_fma_functor::binary_op_sub>(float const*, float const*, float*, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp)
{
const int w = std::max(aw, bw);
const int elempack = std::max(ap, bp);
const int size = w * elempack;
if (ap == bp)
{
if (aw == bw)
{
// no broadcast
... | subq $0x38, %rsp
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
leaq 0x1c(%rsp), %rdi
leaq 0x18(%rsp), %rsi
callq 0x53350
movl (%rax), %eax
movl %eax, 0x10(%rsp)
leaq 0x14(%rsp), %rdi
leaq 0x40(%rsp), %rsi
callq 0... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector<ncnn::BinaryOp_x86_fma_functor::binary_op_rsub>(float const*, float const*, float*, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp)
{
const int w = std::max(aw, bw);
const int elempack = std::max(ap, bp);
const int size = w * elempack;
if (ap == bp)
{
if (aw == bw)
{
// no broadcast
... | subq $0x38, %rsp
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
leaq 0x1c(%rsp), %rdi
leaq 0x18(%rsp), %rsi
callq 0x53350
movl (%rax), %eax
movl %eax, 0x10(%rsp)
leaq 0x14(%rsp), %rdi
leaq 0x40(%rsp), %rsi
callq 0... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector<ncnn::BinaryOp_x86_fma_functor::binary_op_rpow>(float const*, float const*, float*, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp)
{
const int w = std::max(aw, bw);
const int elempack = std::max(ap, bp);
const int size = w * elempack;
if (ap == bp)
{
if (aw == bw)
{
// no broadcast
... | subq $0x38, %rsp
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
leaq 0x1c(%rsp), %rdi
leaq 0x18(%rsp), %rsi
callq 0x53350
movl (%rax), %eax
movl %eax, 0x10(%rsp)
leaq 0x14(%rsp), %rdi
leaq 0x40(%rsp), %rsi
callq 0... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_fma_functor::binary_op_add>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x260, %rsp # imm = 0x260
movq %rdi, 0x110(%rsp)
movq %rsi, 0x108(%rsp)
movq %rdx, 0x100(%rsp)
movl %ecx, 0xfc(%rsp)
movl %r8d, 0xf8(%rsp)
movq 0x108(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xf0(%rsp)
movl $0x0, 0xec(%rsp)
cmpl $0x4, 0xf8(%rsp)
jne 0x... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_fma_functor::binary_op_add>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x260, %rsp # imm = 0x260
movq %rdi, 0xd8(%rsp)
movq %rsi, 0xd0(%rsp)
movq %rdx, 0xc8(%rsp)
movl %ecx, 0xc4(%rsp)
movl %r8d, 0xc0(%rsp)
movl 0xc4(%rsp), %eax
movl 0xc0(%rsp), %ecx
imull %ecx, %eax
movl %eax, 0xb8(%rsp)
movl $0x0, 0xb4(%rsp)
movq 0xd0(%rsp), ... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_no_broadcast<ncnn::BinaryOp_x86_fma_functor::binary_op_sub>(float const*, float const*, float*, int) | static void binary_op_vector_no_broadcast(const float* ptr, const float* ptr1, float* outptr, int size)
{
const Op op;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
__m512 _b = _mm512_loadu_ps(ptr1);
__m512... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x200, %rsp # imm = 0x200
movq %rdi, 0xc8(%rsp)
movq %rsi, 0xc0(%rsp)
movq %rdx, 0xb8(%rsp)
movl %ecx, 0xb4(%rsp)
movl $0x0, 0xac(%rsp)
movl 0xac(%rsp), %eax
addl $0x7, %eax
cmpl 0xb4(%rsp), %eax
jge 0x11b045c
movq 0xc8(%rsp), %rax
movq %rax, 0xd8(%rsp)
movq... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_no_broadcast<ncnn::BinaryOp_x86_fma_functor::binary_op_mul>(float const*, float const*, float*, int) | static void binary_op_vector_no_broadcast(const float* ptr, const float* ptr1, float* outptr, int size)
{
const Op op;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
__m512 _b = _mm512_loadu_ps(ptr1);
__m512... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x200, %rsp # imm = 0x200
movq %rdi, 0xc8(%rsp)
movq %rsi, 0xc0(%rsp)
movq %rdx, 0xb8(%rsp)
movl %ecx, 0xb4(%rsp)
movl $0x0, 0xac(%rsp)
movl 0xac(%rsp), %eax
addl $0x7, %eax
cmpl 0xb4(%rsp), %eax
jge 0x11b1fdc
movq 0xc8(%rsp), %rax
movq %rax, 0xd8(%rsp)
movq... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_fma_functor::binary_op_div>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x260, %rsp # imm = 0x260
movq %rdi, 0xd0(%rsp)
movq %rsi, 0xc8(%rsp)
movq %rdx, 0xc0(%rsp)
movl %ecx, 0xbc(%rsp)
movl %r8d, 0xb8(%rsp)
movl 0xbc(%rsp), %eax
movl 0xb8(%rsp), %ecx
imull %ecx, %eax
movl %eax, 0xb0(%rsp)
movl $0x0, 0xac(%rsp)
movq 0xc8(%rsp), ... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_fma_functor::binary_op_div>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x170(%rsp)
movq %rsi, 0x168(%rsp)
movq %rdx, 0x160(%rsp)
movl %ecx, 0x15c(%rsp)
movl %r8d, 0x158(%rsp)
cmpl $0x8, 0x158(%rsp)
jne 0x11b5201
movl $0x0, 0x150(%rsp)
movq 0x170(%rsp), %rax
movq %rax, 0x178(%rsp)
movq 0x178(... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_no_broadcast<ncnn::BinaryOp_x86_fma_functor::binary_op_max>(float const*, float const*, float*, int) | static void binary_op_vector_no_broadcast(const float* ptr, const float* ptr1, float* outptr, int size)
{
const Op op;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
__m512 _b = _mm512_loadu_ps(ptr1);
__m512... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x200, %rsp # imm = 0x200
movq %rdi, 0xc8(%rsp)
movq %rsi, 0xc0(%rsp)
movq %rdx, 0xb8(%rsp)
movl %ecx, 0xb4(%rsp)
movl $0x0, 0xac(%rsp)
movl 0xac(%rsp), %eax
addl $0x7, %eax
cmpl 0xb4(%rsp), %eax
jge 0x11b56e0
movq 0xc8(%rsp), %rax
movq %rax, 0xd8(%rsp)
movq... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_fma_functor::binary_op_max>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x280, %rsp # imm = 0x280
movq %rdi, 0x108(%rsp)
movq %rsi, 0x100(%rsp)
movq %rdx, 0xf8(%rsp)
movl %ecx, 0xf4(%rsp)
movl %r8d, 0xf0(%rsp)
movq 0x100(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xe8(%rsp)
movl $0x0, 0xe4(%rsp)
cmpl $0x4, 0xf0(%rsp)
jne 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_fma_functor::binary_op_min>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x280, %rsp # imm = 0x280
movq %rdi, 0x108(%rsp)
movq %rsi, 0x100(%rsp)
movq %rdx, 0xf8(%rsp)
movl %ecx, 0xf4(%rsp)
movl %r8d, 0xf0(%rsp)
movq 0x100(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xe8(%rsp)
movl $0x0, 0xe4(%rsp)
cmpl $0x4, 0xf0(%rsp)
jne 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_fma_functor::binary_op_min>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x280, %rsp # imm = 0x280
movq %rdi, 0x108(%rsp)
movq %rsi, 0x100(%rsp)
movq %rdx, 0xf8(%rsp)
movl %ecx, 0xf4(%rsp)
movl %r8d, 0xf0(%rsp)
movq 0x108(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xe8(%rsp)
movl $0x0, 0xe4(%rsp)
cmpl $0x4, 0xf0(%rsp)
jne 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_fma_functor::binary_op_min>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x168(%rsp)
movq %rsi, 0x160(%rsp)
movq %rdx, 0x158(%rsp)
movl %ecx, 0x154(%rsp)
movl %r8d, 0x150(%rsp)
cmpl $0x8, 0x150(%rsp)
jne 0x11b7f89
movl $0x0, 0x148(%rsp)
movl 0x148(%rsp), %eax
cmpl 0x154(%rsp), %eax
jge 0x11b7f... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_fma_functor::binary_op_pow>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x3ba0, %rsp # imm = 0x3BA0
movq %rdi, 0x1a8(%rsp)
movq %rsi, 0x1a0(%rsp)
movq %rdx, 0x198(%rsp)
movl %ecx, 0x194(%rsp)
movl %r8d, 0x190(%rsp)
cmpl $0x8, 0x190(%rsp)
jne 0x11c3027
movl $0x0, 0x188(%rsp)
movl 0x188(%rsp), %eax
cmpl 0x194(%rsp), %eax
jge 0x11c3... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_fma_functor::binary_op_pow>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x3ba0, %rsp # imm = 0x3BA0
movq %rdi, 0x1b0(%rsp)
movq %rsi, 0x1a8(%rsp)
movq %rdx, 0x1a0(%rsp)
movl %ecx, 0x19c(%rsp)
movl %r8d, 0x198(%rsp)
cmpl $0x8, 0x198(%rsp)
jne 0x11ca4a3
movl $0x0, 0x190(%rsp)
movq 0x1b0(%rsp), %rax
movq %rax, 0x1b8(%rsp)
movq 0x1b8... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_fma_functor::binary_op_rdiv>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x168(%rsp)
movq %rsi, 0x160(%rsp)
movq %rdx, 0x158(%rsp)
movl %ecx, 0x154(%rsp)
movl %r8d, 0x150(%rsp)
cmpl $0x8, 0x150(%rsp)
jne 0x11cfc15
movl $0x0, 0x148(%rsp)
movl 0x148(%rsp), %eax
cmpl 0x154(%rsp), %eax
jge 0x11cfc... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_fma_functor::binary_op_rdiv>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x260, %rsp # imm = 0x260
movq %rdi, 0xd0(%rsp)
movq %rsi, 0xc8(%rsp)
movq %rdx, 0xc0(%rsp)
movl %ecx, 0xbc(%rsp)
movl %r8d, 0xb8(%rsp)
movl 0xbc(%rsp), %eax
movl 0xb8(%rsp), %ecx
imull %ecx, %eax
movl %eax, 0xb0(%rsp)
movl $0x0, 0xac(%rsp)
movq 0xc8(%rsp), ... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_fma_functor::binary_op_rpow>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x23c0, %rsp # imm = 0x23C0
movq %rdi, 0x110(%rsp)
movq %rsi, 0x108(%rsp)
movq %rdx, 0x100(%rsp)
movl %ecx, 0xfc(%rsp)
movl %r8d, 0xf8(%rsp)
movl 0xfc(%rsp), %eax
movl 0xf8(%rsp), %ecx
imull %ecx, %eax
movl %eax, 0xf0(%rsp)
movl $0x0, 0xec(%rsp)
movq 0x108(%r... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb_b<ncnn::BinaryOp_x86_fma_functor::binary_op_atan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_b(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
const int size = w * elempack;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
__m512 _b_avx512 = _mm512_set1_ps(*ptr1);
for (; i + 15 < size; i += 16)
{
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x24a0, %rsp # imm = 0x24A0
movq %rdi, 0x110(%rsp)
movq %rsi, 0x108(%rsp)
movq %rdx, 0x100(%rsp)
movl %ecx, 0xfc(%rsp)
movl %r8d, 0xf8(%rsp)
movl 0xfc(%rsp), %eax
movl 0xf8(%rsp), %ecx
imull %ecx, %eax
movl %eax, 0xf0(%rsp)
movl $0x0, 0xec(%rsp)
movq 0x108(%r... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_fma_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x2500, %rsp # imm = 0x2500
movq %rdi, 0x168(%rsp)
movq %rsi, 0x160(%rsp)
movq %rdx, 0x158(%rsp)
movl %ecx, 0x154(%rsp)
movl %r8d, 0x150(%rsp)
movq 0x160(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x148(%rsp)
movl $0x0, 0x144(%rsp)
cmpl $0x4, 0x150(%rsp)
... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_fma_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x3c80, %rsp # imm = 0x3C80
movq %rdi, 0x1f0(%rsp)
movq %rsi, 0x1e8(%rsp)
movq %rdx, 0x1e0(%rsp)
movl %ecx, 0x1dc(%rsp)
movl %r8d, 0x1d8(%rsp)
cmpl $0x8, 0x1d8(%rsp)
jne 0x120f199
movl $0x0, 0x1d0(%rsp)
movq 0x1f0(%rsp), %rax
movq %rax, 0x1f8(%rsp)
movq 0x1f8... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_fma.cpp |
void ncnn::binary_op_vector<ncnn::BinaryOp_x86_avx_functor::binary_op_max>(float const*, float const*, float*, int, int, int, int) | static void binary_op_vector(const float* ptr, const float* ptr1, float* outptr, int aw, int bw, int ap, int bp)
{
const int w = std::max(aw, bw);
const int elempack = std::max(ap, bp);
const int size = w * elempack;
if (ap == bp)
{
if (aw == bw)
{
// no broadcast
... | subq $0x38, %rsp
movl 0x40(%rsp), %eax
movq %rdi, 0x30(%rsp)
movq %rsi, 0x28(%rsp)
movq %rdx, 0x20(%rsp)
movl %ecx, 0x1c(%rsp)
movl %r8d, 0x18(%rsp)
movl %r9d, 0x14(%rsp)
leaq 0x1c(%rsp), %rdi
leaq 0x18(%rsp), %rsi
callq 0x53350
movl (%rax), %eax
movl %eax, 0x10(%rsp)
leaq 0x14(%rsp), %rdi
leaq 0x40(%rsp), %rsi
callq 0... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_no_broadcast<ncnn::BinaryOp_x86_avx_functor::binary_op_add>(float const*, float const*, float*, int) | static void binary_op_vector_no_broadcast(const float* ptr, const float* ptr1, float* outptr, int size)
{
const Op op;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
__m512 _b = _mm512_loadu_ps(ptr1);
__m512... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x200, %rsp # imm = 0x200
movq %rdi, 0xd0(%rsp)
movq %rsi, 0xc8(%rsp)
movq %rdx, 0xc0(%rsp)
movl %ecx, 0xbc(%rsp)
movl $0x0, 0xb4(%rsp)
movl 0xb4(%rsp), %eax
addl $0x7, %eax
cmpl 0xbc(%rsp), %eax
jge 0x121dc6c
movq 0xd0(%rsp), %rax
movq %rax, 0xe0(%rsp)
movq... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_avx_functor::binary_op_add>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x178(%rsp)
movq %rsi, 0x170(%rsp)
movq %rdx, 0x168(%rsp)
movl %ecx, 0x164(%rsp)
movl %r8d, 0x160(%rsp)
cmpl $0x8, 0x160(%rsp)
jne 0x121e965
movl $0x0, 0x158(%rsp)
movl 0x158(%rsp), %eax
cmpl 0x164(%rsp), %eax
jge 0x121e9... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_avx_functor::binary_op_sub>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x280, %rsp # imm = 0x280
movq %rdi, 0x108(%rsp)
movq %rsi, 0x100(%rsp)
movq %rdx, 0xf8(%rsp)
movl %ecx, 0xf4(%rsp)
movl %r8d, 0xf0(%rsp)
movq 0x100(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xe8(%rsp)
movl $0x0, 0xe4(%rsp)
cmpl $0x4, 0xf0(%rsp)
jne 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx_functor::binary_op_sub>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x280, %rsp # imm = 0x280
movq %rdi, 0x108(%rsp)
movq %rsi, 0x100(%rsp)
movq %rdx, 0xf8(%rsp)
movl %ecx, 0xf4(%rsp)
movl %r8d, 0xf0(%rsp)
movq 0x108(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xe8(%rsp)
movl $0x0, 0xe4(%rsp)
cmpl $0x4, 0xf0(%rsp)
jne 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_avx_functor::binary_op_mul>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x168(%rsp)
movq %rsi, 0x160(%rsp)
movq %rdx, 0x158(%rsp)
movl %ecx, 0x154(%rsp)
movl %r8d, 0x150(%rsp)
cmpl $0x8, 0x150(%rsp)
jne 0x1222065
movl $0x0, 0x148(%rsp)
movl 0x148(%rsp), %eax
cmpl 0x154(%rsp), %eax
jge 0x12220... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_avx_functor::binary_op_mul>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x170(%rsp)
movq %rsi, 0x168(%rsp)
movq %rdx, 0x160(%rsp)
movl %ecx, 0x15c(%rsp)
movl %r8d, 0x158(%rsp)
cmpl $0x8, 0x158(%rsp)
jne 0x1222a11
movl $0x0, 0x150(%rsp)
movq 0x170(%rsp), %rax
movq %rax, 0x178(%rsp)
movq 0x178(... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx_functor::binary_op_div>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x280, %rsp # imm = 0x280
movq %rdi, 0x108(%rsp)
movq %rsi, 0x100(%rsp)
movq %rdx, 0xf8(%rsp)
movl %ecx, 0xf4(%rsp)
movl %r8d, 0xf0(%rsp)
movq 0x108(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xe8(%rsp)
movl $0x0, 0xe4(%rsp)
cmpl $0x4, 0xf0(%rsp)
jne 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_avx_functor::binary_op_div>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x168(%rsp)
movq %rsi, 0x160(%rsp)
movq %rdx, 0x158(%rsp)
movl %ecx, 0x154(%rsp)
movl %r8d, 0x150(%rsp)
cmpl $0x8, 0x150(%rsp)
jne 0x1223be5
movl $0x0, 0x148(%rsp)
movl 0x148(%rsp), %eax
cmpl 0x154(%rsp), %eax
jge 0x1223b... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_avx_functor::binary_op_max>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x280, %rsp # imm = 0x280
movq %rdi, 0x108(%rsp)
movq %rsi, 0x100(%rsp)
movq %rdx, 0xf8(%rsp)
movl %ecx, 0xf4(%rsp)
movl %r8d, 0xf0(%rsp)
movq 0x100(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xe8(%rsp)
movl $0x0, 0xe4(%rsp)
cmpl $0x4, 0xf0(%rsp)
jne 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_avx_functor::binary_op_max>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x400, %rsp # imm = 0x400
movq %rdi, 0x170(%rsp)
movq %rsi, 0x168(%rsp)
movq %rdx, 0x160(%rsp)
movl %ecx, 0x15c(%rsp)
movl %r8d, 0x158(%rsp)
cmpl $0x8, 0x158(%rsp)
jne 0x1226125
movl $0x0, 0x150(%rsp)
movq 0x170(%rsp), %rax
movq %rax, 0x178(%rsp)
movq 0x178(... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx_functor::binary_op_rsub>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x2a0, %rsp # imm = 0x2A0
movq %rdi, 0x108(%rsp)
movq %rsi, 0x100(%rsp)
movq %rdx, 0xf8(%rsp)
movl %ecx, 0xf4(%rsp)
movl %r8d, 0xf0(%rsp)
movq 0x108(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0xe8(%rsp)
movl $0x0, 0xe4(%rsp)
cmpl $0x4, 0xf0(%rsp)
jne 0x1... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_no_broadcast<ncnn::BinaryOp_x86_avx_functor::binary_op_rdiv>(float const*, float const*, float*, int) | static void binary_op_vector_no_broadcast(const float* ptr, const float* ptr1, float* outptr, int size)
{
const Op op;
int i = 0;
#if __SSE2__
#if __AVX__
#if __AVX512F__
for (; i + 15 < size; i += 16)
{
__m512 _p = _mm512_loadu_ps(ptr);
__m512 _b = _mm512_loadu_ps(ptr1);
__m512... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x220, %rsp # imm = 0x220
movq %rdi, 0xc8(%rsp)
movq %rsi, 0xc0(%rsp)
movq %rdx, 0xb8(%rsp)
movl %ecx, 0xb4(%rsp)
movl $0x0, 0xac(%rsp)
movl 0xac(%rsp), %eax
addl $0x7, %eax
cmpl 0xb4(%rsp), %eax
jge 0x123f90c
movq 0xc8(%rsp), %rax
movq %rax, 0xd8(%rsp)
movq... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_avx_functor::binary_op_rpow>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x4380, %rsp # imm = 0x4380
movq %rdi, 0x1a8(%rsp)
movq %rsi, 0x1a0(%rsp)
movq %rdx, 0x198(%rsp)
movl %ecx, 0x194(%rsp)
movl %r8d, 0x190(%rsp)
cmpl $0x8, 0x190(%rsp)
jne 0x124c1cb
movl $0x0, 0x188(%rsp)
movl 0x188(%rsp), %eax
cmpl 0x194(%rsp), %eax
jge 0x124c... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_pb_a<ncnn::BinaryOp_x86_avx_functor::binary_op_rpow>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb_a(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
__m512 _p = _mm512_loadu_ps(ptr);
for (; i < w; i++)
{
__... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x4380, %rsp # imm = 0x4380
movq %rdi, 0x1b0(%rsp)
movq %rsi, 0x1a8(%rsp)
movq %rdx, 0x1a0(%rsp)
movl %ecx, 0x19c(%rsp)
movl %r8d, 0x198(%rsp)
cmpl $0x8, 0x198(%rsp)
jne 0x1253e47
movl $0x0, 0x190(%rsp)
movq 0x1b0(%rsp), %rax
movq %rax, 0x1b8(%rsp)
movq 0x1b8... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_b<ncnn::BinaryOp_x86_avx_functor::binary_op_atan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_b(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float b = *ptr1;
int i = 0;
#if __SSE2__
__m128 _b_128 = (elempack == 4) ? _mm_loadu_ps(ptr1) : _mm_set1_ps(b);
#if __AVX__
__m256 _b_256 = (elempack == 8) ? _m... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x26e0, %rsp # imm = 0x26E0
movq %rdi, 0x168(%rsp)
movq %rsi, 0x160(%rsp)
movq %rdx, 0x158(%rsp)
movl %ecx, 0x154(%rsp)
movl %r8d, 0x150(%rsp)
movq 0x160(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x148(%rsp)
movl $0x0, 0x144(%rsp)
cmpl $0x4, 0x150(%rsp)
... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_a<ncnn::BinaryOp_x86_avx_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_a(const float* ptr, const float* ptr1, float* outptr, int size, int elempack)
{
const Op op;
const float a = *ptr;
int i = 0;
#if __SSE2__
__m128 _a_128 = (elempack == 4) ? _mm_loadu_ps(ptr) : _mm_set1_ps(a);
#if __AVX__
__m256 _a_256 = (elempack == 8) ? _mm2... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x2700, %rsp # imm = 0x2700
movq %rdi, 0x168(%rsp)
movq %rsi, 0x160(%rsp)
movq %rdx, 0x158(%rsp)
movl %ecx, 0x154(%rsp)
movl %r8d, 0x150(%rsp)
movq 0x168(%rsp), %rax
vmovss (%rax), %xmm0
vmovss %xmm0, 0x148(%rsp)
movl $0x0, 0x144(%rsp)
cmpl $0x4, 0x150(%rsp)
... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
void ncnn::binary_op_vector_broadcast_pb<ncnn::BinaryOp_x86_avx_functor::binary_op_ratan2>(float const*, float const*, float*, int, int) | static void binary_op_vector_broadcast_pb(const float* ptr, const float* ptr1, float* outptr, int w, int elempack)
{
const Op op;
#if __SSE2__
#if __AVX__
#if __AVX512F__
if (elempack == 16)
{
int i = 0;
for (; i < w; i++)
{
__m512 _p = _mm512_loadu_ps(ptr);
... | pushq %rbp
movq %rsp, %rbp
andq $-0x20, %rsp
subq $0x3fc0, %rsp # imm = 0x3FC0
movq %rdi, 0x1e8(%rsp)
movq %rsi, 0x1e0(%rsp)
movq %rdx, 0x1d8(%rsp)
movl %ecx, 0x1d4(%rsp)
movl %r8d, 0x1d0(%rsp)
cmpl $0x8, 0x1d0(%rsp)
jne 0x1279d09
movl $0x0, 0x1c8(%rsp)
movl 0x1c8(%rsp), %eax
cmpl 0x1d4(%rsp), %eax
jge 0x1279... | /Tencent[P]ncnn/build_O0/src/layer/x86/binaryop_x86_avx.cpp |
ncnn::UnaryOp::forward_inplace(ncnn::Mat&, ncnn::Option const&) const | int UnaryOp::forward_inplace(Mat& bottom_top_blob, const Option& opt) const
{
if (op_type == Operation_ABS)
return unary_op_inplace<unary_op_abs>(bottom_top_blob, opt);
if (op_type == Operation_NEG)
return unary_op_inplace<unary_op_neg>(bottom_top_blob, opt);
if (op_type == Operation_FLOOR... | subq $0x38, %rsp
movq %rdi, 0x28(%rsp)
movq %rsi, 0x20(%rsp)
movq %rdx, 0x18(%rsp)
movq 0x28(%rsp), %rax
movq %rax, 0x8(%rsp)
cmpl $0x0, 0xd0(%rax)
jne 0x12856fe
movq 0x20(%rsp), %rdi
movq 0x18(%rsp), %rsi
callq 0x1285a00
movl %eax, 0x34(%rsp)
jmp 0x12859f0
movq 0x8(%rsp), %rax
cmpl $0x1, 0xd0(%rax)
jne 0x1285724
movq ... | /Tencent[P]ncnn/src/layer/unaryop.cpp |
int ncnn::unary_op_inplace<ncnn::unary_op_neg>(ncnn::Mat&, ncnn::Option const&) | static int unary_op_inplace(Mat& a, const Option& opt)
{
Op op;
int size = static_cast<int>(a.total());
#pragma omp parallel for num_threads(opt.num_threads)
for (int i = 0; i < size; i++)
{
a[i] = op(a[i]);
}
return 0;
} | subq $0x48, %rsp
movq %rdi, 0x18(%rsp)
movq %rsi, 0x10(%rsp)
movq 0x18(%rsp), %rax
movq %rax, 0x20(%rsp)
movq 0x20(%rsp), %rcx
movq 0x40(%rcx), %rax
movslq 0x38(%rcx), %rcx
imulq %rcx, %rax
movl %eax, 0x8(%rsp)
movl $0x0, 0x4(%rsp)
movl 0x4(%rsp), %eax
cmpl 0x8(%rsp), %eax
jge 0x1285b54
movq 0x18(%rsp), %rcx
movslq 0x4... | /Tencent[P]ncnn/src/layer/unaryop.cpp |
int ncnn::unary_op_inplace<ncnn::unary_op_square>(ncnn::Mat&, ncnn::Option const&) | static int unary_op_inplace(Mat& a, const Option& opt)
{
Op op;
int size = static_cast<int>(a.total());
#pragma omp parallel for num_threads(opt.num_threads)
for (int i = 0; i < size; i++)
{
a[i] = op(a[i]);
}
return 0;
} | subq $0x48, %rsp
movq %rdi, 0x18(%rsp)
movq %rsi, 0x10(%rsp)
movq 0x18(%rsp), %rax
movq %rax, 0x20(%rsp)
movq 0x20(%rsp), %rcx
movq 0x40(%rcx), %rax
movslq 0x38(%rcx), %rcx
imulq %rcx, %rax
movl %eax, 0x8(%rsp)
movl $0x0, 0x4(%rsp)
movl 0x4(%rsp), %eax
cmpl 0x8(%rsp), %eax
jge 0x1285d64
movq 0x18(%rsp), %rcx
movslq 0x4... | /Tencent[P]ncnn/src/layer/unaryop.cpp |
Subsets and Splits
SQL Console for LLM4Binary/decompile-bench
Filters out entries with file names ending in .cpp, providing a basic subset of the dataset that excludes C++ files.