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

Working implementation with tests

Browse source
This commit is contained in:
Alejandro Saucedo 2021-03-07 08:00:19 +00:00
parent cf7d46cd23
commit f02b9d6915
21 changed files with 297 additions and 216 deletions
Download patch file Download diff file Expand all files Collapse all files

2
src/OpTensorCopy.cpp
View file

@ -61,7 +61,7 @@ OpTensorCopy::postEval(const vk::CommandBuffer& commandBuffer)
uint32_t size = this->mTensors[0]->size();
uint32_t dataTypeMemSize = this->mTensors[0]->dataTypeMemorySize();
uint32_t memSize = size * dataTypeMemSize;
const void* data = this->mTensors[0]->getRawData();
void* data = this->mTensors[0]->rawData();
// Copy the data from the first tensor into all the tensors
for (size_t i = 1; i < this->mTensors.size(); i++) {