thek0tyara/llama-cpp-turboquant
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 4

Fully working end to end flow

Browse source
This commit is contained in:
Alejandro Saucedo 2020-08-23 06:50:04 +01:00
parent 68ba2d84d0
commit 88944de156
2 changed files with 8 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

1
src/OpMult.cpp
View file

@ -36,6 +36,7 @@ template<uint32_t tX, uint32_t tY, uint32_t tZ>
OpMult<tX, tY, tZ>::~OpMult() OpMult<tX, tY, tZ>::~OpMult()
{ {
SPDLOG_DEBUG("Kompute OpMult destructor started"); SPDLOG_DEBUG("Kompute OpMult destructor started");
} }
template<uint32_t tX, uint32_t tY, uint32_t tZ> template<uint32_t tX, uint32_t tY, uint32_t tZ>

8
src/OpMult.hpp
View file

@ -38,11 +38,17 @@ class OpMult : public OpBase
void postSubmit() override; void postSubmit() override;
private: private:
// Always owned resources
std::shared_ptr<Tensor> mTensorOutputStaging;
// Optionally owned resources
std::shared_ptr<Algorithm> mAlgorithm; std::shared_ptr<Algorithm> mAlgorithm;
bool mFreeAlgorithm = false;
// Never owned resources
std::shared_ptr<Tensor> mTensorLHS; std::shared_ptr<Tensor> mTensorLHS;
std::shared_ptr<Tensor> mTensorRHS; std::shared_ptr<Tensor> mTensorRHS;
std::shared_ptr<Tensor> mTensorOutput; std::shared_ptr<Tensor> mTensorOutput;
std::shared_ptr<Tensor> mTensorOutputStaging;
uint32_t mX; uint32_t mX;
uint32_t mY; uint32_t mY;