mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-07 00:21:07 +01:00
517c7c9861
Summary:
Anywhere we used #include "foo.h", we now say #include <foo.h>
Paths are adjusted to be rooted out of aten/src, torch/lib, or
the root level directory.
I modified CMakeLists.txt by hand to remove TH and THC from
the include paths.
I used the following script to do the canonicalization:
```
import subprocess
import re
import os.path
files = subprocess.check_output(['git', 'ls-files']).decode('utf-8').rstrip().split('\n')
for fn in files:
if not any(fn.endswith(suff) for suff in ['.cu', '.cpp', '.in', '.h', '.hpp', '.cu', '.cuh', '.cc']):
continue
if not any(fn.startswith(pref) for pref in ["aten/", "torch/"]):
continue
with open(fn, 'r') as f:
c = f.read()
def fmt(p):
return "#include <{}>".format(p)
def repl(m):
p = m.group(1)
if p in ["dlfcn.h", "unistd.h", "nvrtc.h", "cuda.h", "cuda_runtime.h", "cstdint", "cudnn.h", "Python.h", "cusparse.h", "cuda_runtime_api.h", "cuda_fp16.h", "cublas_v2.h", "stdint.h", "curand_kernel.h"]:
return fmt(p)
if any(p.startswith(pref) for pref in ["torch/csrc", "c10/", "ATen/", "caffe2/", "TH/", "THC/", "Eigen/", "gtest/", "zdl/", "gloo/", "onnx/", "miopen/"]):
return fmt(p)
for root in ["aten/src", "torch/lib", ""]:
for bad_root in [os.path.dirname(fn), "aten/src/TH", "aten/src/THC", "torch/csrc"]:
new_p = os.path.relpath(os.path.join(bad_root, p), root)
if not new_p.startswith("../") and (os.path.exists(os.path.join(root, new_p)) or os.path.exists(os.path.join(root, new_p + ".in"))):
return fmt(new_p)
print("ERROR: ", fn, p)
return m.group(0)
new_c = re.sub(r'#include "([^"]+)"', repl, c)
if new_c != c:
print(fn)
with open(fn, 'w') as f:
f.write(new_c)
```
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/14849
Reviewed By: dzhulgakov
Differential Revision: D13363445
Pulled By: ezyang
fbshipit-source-id: 52361f878a672785f9306c9e9ab2513128092b68
114 lines
4.0 KiB
C++
114 lines
4.0 KiB
C++
#ifndef TH_GENERIC_FILE
|
|
#define TH_GENERIC_FILE "torch/csrc/generic/serialization.cpp"
|
|
#else
|
|
|
|
template <class io>
|
|
void THPStorage_(writeFileRaw)(THWStorage *self, io fd)
|
|
{
|
|
scalar_t *data;
|
|
int64_t size = THWStorage_(size)(LIBRARY_STATE self);
|
|
#ifndef THC_GENERIC_FILE
|
|
data = THWStorage_(data)(LIBRARY_STATE self);
|
|
#else
|
|
std::unique_ptr<char[]> cpu_data(new char[size * sizeof(scalar_t)]);
|
|
data = (scalar_t*)cpu_data.get();
|
|
THCudaCheck(cudaMemcpy(data, THWStorage_(data)(LIBRARY_STATE self), size * sizeof(scalar_t), cudaMemcpyDeviceToHost));
|
|
#endif
|
|
doWrite(fd, &size, sizeof(int64_t));
|
|
// fast track for bytes and little endian
|
|
if (sizeof(scalar_t) == 1 || THP_nativeByteOrder() == THPByteOrder::THP_LITTLE_ENDIAN) {
|
|
doWrite(fd, data, sizeof(scalar_t) * size);
|
|
} else {
|
|
int64_t buffer_size = std::min(size, (int64_t)5000);
|
|
std::unique_ptr<uint8_t[]> le_buffer(new uint8_t[buffer_size * sizeof(scalar_t)]);
|
|
for (int64_t i = 0; i < size; i += buffer_size) {
|
|
size_t to_convert = std::min(size - i, buffer_size);
|
|
if (sizeof(scalar_t) == 2) {
|
|
THP_encodeInt16Buffer((uint8_t*)le_buffer.get(),
|
|
(const int16_t*)data + i,
|
|
THPByteOrder::THP_LITTLE_ENDIAN,
|
|
to_convert);
|
|
} else if (sizeof(scalar_t) == 4) {
|
|
THP_encodeInt32Buffer((uint8_t*)le_buffer.get(),
|
|
(const int32_t*)data + i,
|
|
THPByteOrder::THP_LITTLE_ENDIAN,
|
|
to_convert);
|
|
} else if (sizeof(scalar_t) == 8) {
|
|
THP_encodeInt64Buffer((uint8_t*)le_buffer.get(),
|
|
(const int64_t*)data + i,
|
|
THPByteOrder::THP_LITTLE_ENDIAN,
|
|
to_convert);
|
|
}
|
|
doWrite(fd, le_buffer.get(), to_convert * sizeof(scalar_t));
|
|
}
|
|
}
|
|
}
|
|
|
|
template void THPStorage_(writeFileRaw<int>)(THWStorage *self, int fd);
|
|
template void THPStorage_(writeFileRaw<PyObject*>)(THWStorage *self, PyObject* fd);
|
|
|
|
template <class io>
|
|
THWStorage * THPStorage_(readFileRaw)(io file, THWStorage *_storage)
|
|
{
|
|
scalar_t *data;
|
|
int64_t size;
|
|
doRead(file, &size, sizeof(int64_t));
|
|
THWStoragePtr storage;
|
|
if (_storage == nullptr) {
|
|
storage = THWStorage_(newWithSize)(LIBRARY_STATE size);
|
|
} else {
|
|
THPUtils_assert(THWStorage_(size)(LIBRARY_STATE _storage) == size,
|
|
"storage has wrong size: expected %ld got %ld",
|
|
size, THWStorage_(size)(LIBRARY_STATE _storage));
|
|
storage = _storage;
|
|
}
|
|
|
|
#ifndef THC_GENERIC_FILE
|
|
data = THWStorage_(data)(LIBRARY_STATE storage);
|
|
#else
|
|
std::unique_ptr<char[]> cpu_data(new char[size * sizeof(scalar_t)]);
|
|
data = (scalar_t*)cpu_data.get();
|
|
#endif
|
|
|
|
// fast track for bytes and little endian
|
|
if (sizeof(scalar_t) == 1 || THP_nativeByteOrder() == THPByteOrder::THP_LITTLE_ENDIAN) {
|
|
doRead(file, data, sizeof(scalar_t) * THWStorage_(size)(LIBRARY_STATE storage));
|
|
} else {
|
|
int64_t buffer_size = std::min(size, (int64_t)5000);
|
|
std::unique_ptr<uint8_t[]> le_buffer(new uint8_t[buffer_size * sizeof(scalar_t)]);
|
|
|
|
|
|
for (int64_t i = 0; i < size; i += buffer_size) {
|
|
size_t to_convert = std::min(size - i, buffer_size);
|
|
doRead(file, le_buffer.get(), sizeof(scalar_t) * to_convert);
|
|
|
|
if (sizeof(scalar_t) == 2) {
|
|
THP_decodeInt16Buffer((int16_t*)data + i,
|
|
le_buffer.get(),
|
|
THPByteOrder::THP_LITTLE_ENDIAN,
|
|
to_convert);
|
|
} else if (sizeof(scalar_t) == 4) {
|
|
THP_decodeInt32Buffer((int32_t*)data + i,
|
|
le_buffer.get(),
|
|
THPByteOrder::THP_LITTLE_ENDIAN,
|
|
to_convert);
|
|
} else if (sizeof(scalar_t) == 8) {
|
|
THP_decodeInt64Buffer((int64_t*)data + i,
|
|
le_buffer.get(),
|
|
THPByteOrder::THP_LITTLE_ENDIAN,
|
|
to_convert);
|
|
}
|
|
}
|
|
}
|
|
|
|
#ifdef THC_GENERIC_FILE
|
|
THCudaCheck(cudaMemcpy(THWStorage_(data)(LIBRARY_STATE storage), data, size * sizeof(scalar_t), cudaMemcpyHostToDevice));
|
|
#endif
|
|
return storage.release();
|
|
}
|
|
|
|
template THWStorage* THPStorage_(readFileRaw<int>)(int fd, THWStorage* storage);
|
|
template THWStorage* THPStorage_(readFileRaw<PyObject*>)(PyObject* fd, THWStorage* storage);
|
|
|
|
#endif
|