-
Notifications
You must be signed in to change notification settings - Fork 204
/
elu.cu
164 lines (148 loc) · 6.93 KB
/
elu.cu
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
#include <stdio.h>
#include <stdlib.h>
#include <float.h>
#include <vector>
#include <algorithm>
#include <cuda_runtime.h>
#include <cuda_fp16.h>
#include <torch/types.h>
#include <torch/extension.h>
#define WARP_SIZE 32
#define INT4(value) (reinterpret_cast<int4*>(&(value))[0])
#define FLOAT4(value) (reinterpret_cast<float4*>(&(value))[0])
#define HALF2(value) (reinterpret_cast<half2*>(&(value))[0])
#define BFLOAT2(value) (reinterpret_cast<__nv_bfloat162*>(&(value))[0])
#define LDST128BITS(value) (reinterpret_cast<float4*>(&(value))[0])
// 定义全局 alpha 值
#define ALPHA 1.0f
// 定义 CHECK_TORCH_TENSOR_DTYPE 宏
#define CHECK_TORCH_TENSOR_DTYPE(T, th_type) \
if (((T).options().dtype() != (th_type))) { \
std::cout << "Tensor Info:" << (T).options() << std::endl; \
throw std::runtime_error("Tensor dtype must be " #th_type); \
}
// 定义 TORCH_BINDING_COMMON_EXTENSION 宏
#define STRINGFY(str) #str
#define TORCH_BINDING_COMMON_EXTENSION(func) \
m.def(STRINGFY(func), &func, STRINGFY(func));
// ELU 计算函数
// -------------------------------------- FP32 --------------------------------------
__device__ __forceinline__ float elu(float x) {
return x > 0.f ? x : ALPHA * (expf(x) - 1.f);
}
// -------------------------------------- FP16 --------------------------------------
__device__ __forceinline__ half elu_half(half x) {
return __hgt(x, __float2half(0.f)) ? x : __hmul(__float2half(ALPHA), __hsub(hexp(x), __float2half(1.f)));
}
// CUDA 核函数
// -------------------------------------- FP32 --------------------------------------
__global__ void elu_f32_kernel(float* x, float* y, int N) {
int idx = blockIdx.x * blockDim.x + threadIdx.x;
if (idx < N) y[idx] = elu(x[idx]);
}
__global__ void elu_f32x4_kernel(float* x, float* y, int N) {
int idx = (blockIdx.x * blockDim.x + threadIdx.x) * 4;
if (idx < N) {
float4 reg_x = FLOAT4(x[idx]);
float4 reg_y;
reg_y.x = elu(reg_x.x);
reg_y.y = elu(reg_x.y);
reg_y.z = elu(reg_x.z);
reg_y.w = elu(reg_x.w);
FLOAT4(y[idx]) = reg_y;
}
}
// -------------------------------------- FP16 --------------------------------------
__global__ void elu_f16_kernel(half* x, half* y, int N) {
int idx = blockIdx.x * blockDim.x + threadIdx.x;
if (idx < N) y[idx] = elu_half(x[idx]);
}
__global__ void elu_f16x2_kernel(half* x, half* y, int N) {
int idx = 2 * (blockIdx.x * blockDim.x + threadIdx.x);
if (idx < N) {
half2 reg_x = HALF2(x[idx]);
half2 reg_y;
reg_y.x = elu_half(reg_x.x);
reg_y.y = elu_half(reg_x.y);
HALF2(y[idx]) = reg_y;
}
}
__global__ void elu_f16x8_kernel(half* x, half* y, int N) {
int idx = 8 * (blockIdx.x * blockDim.x + threadIdx.x);
half2 reg_x_0 = HALF2(x[idx + 0]);
half2 reg_x_1 = HALF2(x[idx + 2]);
half2 reg_x_2 = HALF2(x[idx + 4]);
half2 reg_x_3 = HALF2(x[idx + 6]);
half2 reg_y_0, reg_y_1, reg_y_2, reg_y_3;
reg_y_0.x = elu_half(reg_x_0.x);
reg_y_0.y = elu_half(reg_x_0.y);
reg_y_1.x = elu_half(reg_x_1.x);
reg_y_1.y = elu_half(reg_x_1.y);
reg_y_2.x = elu_half(reg_x_2.x);
reg_y_2.y = elu_half(reg_x_2.y);
reg_y_3.x = elu_half(reg_x_3.x);
reg_y_3.y = elu_half(reg_x_3.y);
if ((idx + 0) < N) { HALF2(y[idx + 0]) = reg_y_0; }
if ((idx + 2) < N) { HALF2(y[idx + 2]) = reg_y_1; }
if ((idx + 4) < N) { HALF2(y[idx + 4]) = reg_y_2; }
if ((idx + 6) < N) { HALF2(y[idx + 6]) = reg_y_3; }
}
__global__ void elu_f16x8_pack_kernel(half* x, half* y, int N) {
int idx = 8 * (blockIdx.x * blockDim.x + threadIdx.x);
half pack_x[8], pack_y[8];
LDST128BITS(pack_x[0]) = LDST128BITS(x[idx]);
#pragma unroll
for (int i = 0; i < 8; i++) {
pack_y[i] = elu_half(pack_x[i]);
}
if ((idx + 7) < N) { LDST128BITS(y[idx]) = LDST128BITS(pack_y[0]); }
}
// PyTorch 绑定代码
#define TORCH_BINDING_ELU(packed_type, th_type, element_type, n_elements) \
void elu_##packed_type(torch::Tensor x, torch::Tensor y) { \
CHECK_TORCH_TENSOR_DTYPE(x, (th_type)) \
CHECK_TORCH_TENSOR_DTYPE(y, (th_type)) \
const int ndim = x.dim(); \
if (ndim != 2) { \
int N = 1; \
for (int i = 0; i < ndim; ++i) { N *= x.size(i); } \
dim3 block(256 / (n_elements)); \
dim3 grid((N + 256 - 1) / 256); \
elu_##packed_type##_kernel<<<grid, block>>>( \
reinterpret_cast<element_type*>(x.data_ptr()), \
reinterpret_cast<element_type*>(y.data_ptr()), N); \
} else { \
const int S = x.size(0); \
const int K = x.size(1); \
const int N = S * K; \
if ((K/(n_elements)) <= 1024) { \
dim3 block(K/(n_elements)); \
dim3 grid(S); \
elu_##packed_type##_kernel<<<grid, block>>>( \
reinterpret_cast<element_type*>(x.data_ptr()), \
reinterpret_cast<element_type*>(y.data_ptr()), N); \
} else { \
int N = 1; \
for (int i = 0; i < ndim; ++i) { N *= x.size(i); } \
dim3 block(256 / (n_elements)); \
dim3 grid((N + 256 - 1) / 256); \
elu_##packed_type##_kernel<<<grid, block>>>( \
reinterpret_cast<element_type*>(x.data_ptr()), \
reinterpret_cast<element_type*>(y.data_ptr()), N); \
} \
} \
}
TORCH_BINDING_ELU(f32, torch::kFloat32, float, 1)
TORCH_BINDING_ELU(f32x4, torch::kFloat32, float, 4)
TORCH_BINDING_ELU(f16, torch::kHalf, half, 1)
TORCH_BINDING_ELU(f16x2, torch::kHalf, half, 2)
TORCH_BINDING_ELU(f16x8, torch::kHalf, half, 8)
TORCH_BINDING_ELU(f16x8_pack, torch::kHalf, half, 8)
PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
TORCH_BINDING_COMMON_EXTENSION(elu_f32)
TORCH_BINDING_COMMON_EXTENSION(elu_f32x4)
TORCH_BINDING_COMMON_EXTENSION(elu_f16)
TORCH_BINDING_COMMON_EXTENSION(elu_f16x2)
TORCH_BINDING_COMMON_EXTENSION(elu_f16x8)
TORCH_BINDING_COMMON_EXTENSION(elu_f16x8_pack)
}