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
|
#include <c10/cuda/CUDAGuard.h>
#include <torch/csrc/distributed/c10d/Utils.hpp>
#include <torch/csrc/distributed/c10d/quantization/quantization_gpu.h>
#include <torch/csrc/distributed/c10d/quantization/quantization_utils.h>
#include <torch/library.h>
// TODO: The kernels are copied from fbgemm_gpu, we should dedup them later
// FP32 -> BF16 kernel
__global__ void _float_to_bfloat16_cuda_kernel(
const float* __restrict__ input,
const size_t nrows,
const size_t ncols,
uint16_t* __restrict__ output) {
const auto row_incre = blockDim.y * gridDim.y;
const auto col_incre = blockDim.x * gridDim.x;
for (auto row = blockIdx.y * blockDim.y + threadIdx.y; row < nrows;
row += row_incre) {
const float* input_row = input + row * ncols;
uint16_t* output_row = output + row * ncols;
for (auto col = blockIdx.x * blockDim.x + threadIdx.x; col < ncols;
col += col_incre) {
// Add 2^15 and right shift 16 to do round-nearest
output_row[col] =
(*reinterpret_cast<const uint32_t*>(input_row + col) + (1 << 15)) >>
16;
}
}
}
// BF16 -> FP32 kernel
__global__ void _bfloat16_to_float_cuda_kernel(
const uint16_t* __restrict__ input,
const size_t nrows,
const size_t ncols,
float* __restrict__ output) {
const auto row_incre = blockDim.y * gridDim.y;
const auto col_incre = blockDim.x * gridDim.x;
for (auto row = blockIdx.y * blockDim.y + threadIdx.y; row < nrows;
row += row_incre) {
for (auto col = blockIdx.x * blockDim.x + threadIdx.x; col < ncols;
col += col_incre) {
const uint16_t* input_row = input + row * ncols;
float* output_row = output + row * ncols;
uint32_t val_fp32 = static_cast<uint32_t>(
reinterpret_cast<const uint16_t*>(input_row)[col])
<< 16;
reinterpret_cast<uint32_t*>(output_row)[col] = val_fp32;
}
}
}
namespace torch::distributed::c10d::quantization {
at::Tensor _float_to_bfloat16_cuda(const at::Tensor& input) {
TENSOR_ON_CUDA_GPU(input);
// Currently it supports 2D inputs
TENSOR_NDIM_EQUALS(input, 2);
at::cuda::OptionalCUDAGuard device_guard;
device_guard.set_index(input.get_device());
const auto nrows = input.size(0);
const auto ncols = input.size(1);
const size_t output_columns = ncols;
auto output = at::empty(
{nrows, ncols},
#if HAS_NCCL_BF16_DATATYPE
input.options().dtype(at::kBFloat16));
#else
input.options().dtype(at::kHalf));
#endif
if (nrows == 0 || ncols == 0) {
return output;
}
constexpr size_t threads_per_block = 256;
const auto blockDim_x = std::min(output_columns, threads_per_block);
dim3 blockDim(blockDim_x, threads_per_block / blockDim_x);
const auto gridDim_x = (output_columns + blockDim.x - 1) / blockDim.x;
const auto gridDim_y =
std::min<size_t>((nrows + blockDim.y - 1) / blockDim.y, 65535u);
dim3 gridDim(gridDim_x, gridDim_y);
_float_to_bfloat16_cuda_kernel<<<
gridDim,
blockDim,
0,
at::cuda::getCurrentCUDAStream()>>>(
input.const_data_ptr<float>(),
nrows,
ncols,
#if HAS_NCCL_BF16_DATATYPE
reinterpret_cast<uint16_t*>(output.mutable_data_ptr<at::BFloat16>())
#else
reinterpret_cast<uint16_t*>(output.mutable_data_ptr<at::Half>())
#endif
);
C10_CUDA_KERNEL_LAUNCH_CHECK();
return output;
}
at::Tensor _bfloat16_to_float_cuda(const at::Tensor& input) {
TENSOR_ON_CUDA_GPU(input);
// Currently it supports 2D inputs
TENSOR_NDIM_EQUALS(input, 2);
at::cuda::OptionalCUDAGuard device_guard;
device_guard.set_index(input.get_device());
const auto nrows = input.size(0);
const auto ncols = input.size(1);
const size_t output_columns = ncols;
auto output = at::empty(
{nrows, ncols}, // 4 = sizeof(float)
input.options().dtype(at::kFloat)); // at::kBytes for uint8_t
if (nrows == 0 || ncols == 0) {
return output;
}
constexpr size_t threads_per_block = 256;
const auto blockDim_x = std::min(output_columns, threads_per_block);
dim3 blockDim(blockDim_x, threads_per_block / blockDim_x);
const auto gridDim_x = (output_columns + blockDim.x - 1) / blockDim.x;
const auto gridDim_y =
std::min<size_t>((nrows + blockDim.y - 1) / blockDim.y, 65535u);
dim3 gridDim(gridDim_x, gridDim_y);
_bfloat16_to_float_cuda_kernel<<<
gridDim,
blockDim,
0,
at::cuda::getCurrentCUDAStream()>>>(
#if HAS_NCCL_BF16_DATATYPE
reinterpret_cast<const uint16_t*>(input.const_data_ptr<at::BFloat16>()),
#else
reinterpret_cast<const uint16_t*>(input.const_data_ptr<at::Half>()),
#endif
nrows,
ncols,
output.mutable_data_ptr<float>());
C10_CUDA_KERNEL_LAUNCH_CHECK();
return output;
}
#define DISPATCH_TO_CUDA(name, function) \
m.impl(name, torch::dispatch(c10::DispatchKey::CUDA, TORCH_FN(function)))
TORCH_LIBRARY_IMPL(quantization, CUDA, m) {
DISPATCH_TO_CUDA("_Bfloat16QuantizedToFloat", _bfloat16_to_float_cuda);
DISPATCH_TO_CUDA("_FloatToBfloat16Quantized", _float_to_bfloat16_cuda);
}
} // namespace torch::distributed::c10d::quantization
|
