what's the final solution for this question?
@junrushao1994
---
[Visit
Can we make node-level parallelism on gpu?
---
[Visit
Here are two methods to convert model, what's the differences?
1.first method:
```bash
mod,params
=relay.frontend.from_mxnet(sym,shape_dict,arg_params=arg_params,aux_params=aux_params)
with relay.build_config(opt_level=opt_level):
graph, lib, params = relay.build(mod, target=target,
I try to set the TVM_NUM_THREADS on my laptop (the server has run over 15 hours
and I don't want to affect it). It seems that the program still only occupies 1
physical core.
CPU: Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz 6C12T
GPU: GeForce GTX 1060 Mobile