#include <compilation_cache.h>
|
| ManualCudaCache ()=default |
|
| ManualCudaCache (const ManualCudaCacheProto &buf) |
|
ManualCudaCacheProto | toProtobuf () const |
|
void | cacheKernel (const std::string &id, const std::vector< const DLTensor * > &inputs, const std::vector< const DLTensor * > &outputs, const std::string &kernelSpecializedName, const std::vector< int > &kernelParameters, const std::string &cudaSource, const Grid &grid, const Block &block) |
|
std::unique_ptr
< CudaCache::RetrievalResult > | retrieveKernel (const std::string &id, const std::vector< const DLTensor * > &inputs, const std::vector< const DLTensor * > &outputs) const |
|
template<typename C , typename TensorTy > |
auto | searchKernelImpl (C &c, const std::string &id, const std::vector< TensorTy > &inputs, const std::vector< TensorTy > &outputs) -> decltype(c.searchKernel(id, inputs, outputs)) |
|
size_t | size () const |
|
void | clear () |
|
|
static std::shared_ptr
< ManualCudaCache > & | getGlobalSharedCache () |
|
template<typename C , typename InputTy > |
static auto | searchKernelImpl (C &c, const std::string &id, const std::vector< InputTy > &inputs, const std::vector< InputTy > &outputs) -> decltype(c.searchKernel(id, inputs, outputs)) |
|
tc::ManualCudaCache::ManualCudaCache |
( |
| ) |
|
|
default |
tc::ManualCudaCache::ManualCudaCache |
( |
const ManualCudaCacheProto & |
buf | ) |
|
void tc::ManualCudaCache::cacheKernel |
( |
const std::string & |
id, |
|
|
const std::vector< const DLTensor * > & |
inputs, |
|
|
const std::vector< const DLTensor * > & |
outputs, |
|
|
const std::string & |
kernelSpecializedName, |
|
|
const std::vector< int > & |
kernelParameters, |
|
|
const std::string & |
cudaSource, |
|
|
const Grid & |
grid, |
|
|
const Block & |
block |
|
) |
| |
static std::shared_ptr<ManualCudaCache>& tc::ManualCudaCache::getGlobalSharedCache |
( |
| ) |
|
|
staticprivate |
std::unique_ptr<CudaCache::RetrievalResult> tc::ManualCudaCache::retrieveKernel |
( |
const std::string & |
id, |
|
|
const std::vector< const DLTensor * > & |
inputs, |
|
|
const std::vector< const DLTensor * > & |
outputs |
|
) |
| const |
CachedEntry* tc::ManualCudaCache::searchKernel |
( |
const std::string & |
id, |
|
|
const std::vector< const DLTensor * > & |
inputs, |
|
|
const std::vector< const DLTensor * > & |
outputs |
|
) |
| |
|
private |
const CachedEntry* tc::ManualCudaCache::searchKernel |
( |
const std::string & |
id, |
|
|
const std::vector< const DLTensor * > & |
inputs, |
|
|
const std::vector< const DLTensor * > & |
outputs |
|
) |
| const |
|
private |
template<typename C , typename TensorTy >
auto tc::ManualCudaCache::searchKernelImpl |
( |
C & |
c, |
|
|
const std::string & |
id, |
|
|
const std::vector< TensorTy > & |
inputs, |
|
|
const std::vector< TensorTy > & |
outputs |
|
) |
| -> decltype(c.searchKernel(id, inputs, outputs)) |
template<typename C , typename InputTy >
static auto tc::ManualCudaCache::searchKernelImpl |
( |
C & |
c, |
|
|
const std::string & |
id, |
|
|
const std::vector< InputTy > & |
inputs, |
|
|
const std::vector< InputTy > & |
outputs |
|
) |
| -> decltype(c.searchKernel(id, inputs, outputs)) |
|
staticprivate |
ManualCudaCacheProto tc::ManualCudaCache::toProtobuf |
( |
| ) |
const |
The documentation for this class was generated from the following files: