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

Updated issue in opalgolhs

Browse source
This commit is contained in:
Alejandro Saucedo 2020-09-19 16:07:55 +01:00
parent f8f9a38fdf
commit 71a0b24e42
2 changed files with 6 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

3
single_include/kompute/Kompute.hpp
View file

@ -1442,11 +1442,13 @@ OpAlgoLhsRhsOut<tX, tY, tZ>::record()
// Barrier to ensure the data is finished writing to buffer memory
this->mTensorLHS->recordBufferMemoryBarrier(
this->mCommandBuffer,
vk::AccessFlagBits::eHostWrite,
vk::AccessFlagBits::eShaderRead,
vk::PipelineStageFlagBits::eHost,
vk::PipelineStageFlagBits::eComputeShader);
this->mTensorRHS->recordBufferMemoryBarrier(
this->mCommandBuffer,
vk::AccessFlagBits::eHostWrite,
vk::AccessFlagBits::eShaderRead,
vk::PipelineStageFlagBits::eHost,
@ -1456,6 +1458,7 @@ OpAlgoLhsRhsOut<tX, tY, tZ>::record()
// Barrier to ensure the shader code is executed before buffer read
this->mTensorOutput->recordBufferMemoryBarrier(
this->mCommandBuffer,
vk::AccessFlagBits::eShaderWrite,
vk::AccessFlagBits::eTransferRead,
vk::PipelineStageFlagBits::eComputeShader,

3
src/include/kompute/operations/OpAlgoLhsRhsOut.hpp
View file

@ -177,11 +177,13 @@ OpAlgoLhsRhsOut<tX, tY, tZ>::record()
// Barrier to ensure the data is finished writing to buffer memory
this->mTensorLHS->recordBufferMemoryBarrier(
this->mCommandBuffer,
vk::AccessFlagBits::eHostWrite,
vk::AccessFlagBits::eShaderRead,
vk::PipelineStageFlagBits::eHost,
vk::PipelineStageFlagBits::eComputeShader);
this->mTensorRHS->recordBufferMemoryBarrier(
this->mCommandBuffer,
vk::AccessFlagBits::eHostWrite,
vk::AccessFlagBits::eShaderRead,
vk::PipelineStageFlagBits::eHost,
@ -191,6 +193,7 @@ OpAlgoLhsRhsOut<tX, tY, tZ>::record()
// Barrier to ensure the shader code is executed before buffer read
this->mTensorOutput->recordBufferMemoryBarrier(
this->mCommandBuffer,
vk::AccessFlagBits::eShaderWrite,
vk::AccessFlagBits::eTransferRead,
vk::PipelineStageFlagBits::eComputeShader,