pycuda._ driver. LogicError: explicit_ context_ dependent failed: invalid device context - no currently

In execution tensorRT There was a mistake in reasoning , The details are as follows ：

D:\cv\ConNext_demo\testtrt.py:32: DeprecationWarning: Use set_optimization_profile_async instead.
  context.active_optimization_profile = 0
Traceback (most recent call last):
  File "D:\cv\ConNext_demo\testtrt.py", line 37, in <module>
    d_input = cuda.mem_alloc(image.nbytes)  #  Allocate input memory .
pycuda._driver.LogicError: explicit_context_dependent failed: invalid device context - no currently active context?

This is a pycuda Not initialized , Just execute initialization ：

import pycuda.driver as cuda
import pycuda.autoinit

import pycuda.autoinit This line is gray , It feels useless , But if there is no such sentence, the above error will be reported .