• Wu Zhiwen's avatar
    dnn/Vulkan: skip heavy convolution task · 33c9d57c
    Wu Zhiwen authored
    This is a workaround for GPU hang on heavy convolution workload (> 10 GFLOPS).
    e.g. ResNet101_DUC_HDC
    
    For the long time task, vkWaitForFences() return without error but next call on
    vkQueueSubmit() return -4, i.e. "VK_ERROR_DEVICE_LOST" and driver reports GPU hang.
    
    Need more investigation on root cause of GPU hang and need to optimize convolution shader
    to reduce process time.
    33c9d57c
Name
Last commit
Last update
..
caffe Loading commit data...
darknet Loading commit data...
layers Loading commit data...
ocl4dnn Loading commit data...
onnx Loading commit data...
opencl Loading commit data...
tensorflow Loading commit data...
torch Loading commit data...
vkcom Loading commit data...
dnn.cpp Loading commit data...
halide_scheduler.cpp Loading commit data...
halide_scheduler.hpp Loading commit data...
init.cpp Loading commit data...
nms.cpp Loading commit data...
nms.inl.hpp Loading commit data...
op_halide.cpp Loading commit data...
op_halide.hpp Loading commit data...
op_inf_engine.cpp Loading commit data...
op_inf_engine.hpp Loading commit data...
op_vkcom.cpp Loading commit data...
op_vkcom.hpp Loading commit data...
precomp.hpp Loading commit data...