[Auto-TVM] Could we train use multi cpu cores?

Questions
1

It seems that we can only use one cpu core during auto tvm training. I tried Cython as ffi, but seems that no effect. @merrymercy do you train use one cpu core too? It seems very slow. Could we use multi cpu cores?

2

What is the num_threads (https://docs.tvm.ai/api/python/autotvm.html#tvm.autotvm.tuner.XGBTuner) that you are using?

3

By default it will use all CPU cores.
Maybe you can try to set num_threads explicitly.

4

Thanks. I think should be my modified code to result in this. Because my mac can be multi-cores, but linux server can not.