Hi,
Is it possible to have the UDO custom layers run on CPU while the main model runs on GPU ? This was possible while using UDL. However while running the alexnet model UDO example with --use_gpu mode via snpe-net-run, I get the followng error -
error_code=1010; error_message=UDO Layer failed network validation. error_code=1010; error_message=UDO Layer failed network validation. UDO Constraint Validation Failed; error_component=Model Validation; line_no=2550; thread_id=3912205280; error_component=Model Validation; line_no=271; thread_id=4096339984
Can I get some clarification if this is possible or not.
Thanks,
Anirudh
Dear customer,
You need to write the code of OpenCL into UDO for GPU. Regarding the OpenCL implementation inside SNPE UDO, you can prepare the env for the OpenCL compiling.
BR.
Wei