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

Further tests added to new structure

Browse source
This commit is contained in:
Alejandro Saucedo 2021-02-25 22:33:08 +00:00
parent 3f1288271d
commit 6378583a23
17 changed files with 636 additions and 514 deletions
Download patch file Download diff file Expand all files Collapse all files

6
src/OpAlgoDispatch.cpp
View file

@ -4,12 +4,10 @@
namespace kp {
OpAlgoDispatch::OpAlgoDispatch(const std::vector<std::shared_ptr<Tensor>>& tensors,
const std::shared_ptr<kp::Algorithm>& algorithm)
OpAlgoDispatch::OpAlgoDispatch(const std::shared_ptr<kp::Algorithm>& algorithm)
{
KP_LOG_DEBUG("Kompute OpAlgoDispatch constructor");
this->mTensors = tensors;
this->mAlgorithm = algorithm;
}
@ -24,7 +22,7 @@ OpAlgoDispatch::record(std::shared_ptr<vk::CommandBuffer> commandBuffer)
KP_LOG_DEBUG("Kompute OpAlgoDispatch record called");
// Barrier to ensure the data is finished writing to buffer memory
for (std::shared_ptr<Tensor> tensor : this->mTensors) {
for (const std::shared_ptr<Tensor>& tensor : this->mAlgorithm->getTensors()) {
tensor->recordBufferMemoryBarrier(
commandBuffer,
vk::AccessFlagBits::eHostWrite,