- Ask a related questionWhat is a related question?A related question is a question created from another question. When the related question is created, it will be automatically linked to the original question.
if i have a tensorflow model, and there is an unspport op on C7x/MMA,
so ,i use tvm to optimize and generate the artifacts .
the model could be split two partion,
a. ops are not supported by TIDL ---- neo-ai-dlr runtime load the deploy_lib.so ,deploy_graph.json, deploy_param.params ,run on ARM core.
b. ops are supported by TIDL, using openVX load the subgraph0_net.bin , and run on the C7x/MMA
When I am confused, how to link the results of the two parts, do you need the customer to rewrite code by themselves？
example ,the data flow as below
input data --> subgraph0_net.bin --> deploy_lib.so -->out
look at the deploy_graph.json file ,seem it contains the tidl_0(subgraph0_net), is right?
in other words ,the neo-ai-dlr runtime could load and run the subgraph0_net subgraph0_net automatically ？
input data--> deploy_lib.so --->out
