zhuwenxi commented on issue #7596:
URL: https://github.com/apache/tvm/issues/7596#issuecomment-795033720


   @leeexyz , I looked into the test cases for tvmc 
(https://github.com/apache/tvm/blob/main/tests/python/unittest/test_target_codegen_c_host.py#L36)
 and found that all the cases export their module as library and then load the 
module, before actually run the module, like this:
   <pre>
   mhost = tvm.build(s, [A, B, C], "c", name="test_fadd")
   **mhost.export_library(path_dso)
   m = tvm.runtime.load_module(path_dso)**
   fadd = m["test_fadd"]
   fadd(a, b, c)
   </pre> 
   
   My question is, is this a desired behavior, rather than a bug that tvmc does 
not support to run a module directly?
   <pre>
   # Following code pattern is not desired in tvmc?
   f = tvm.build(s, [A, B, D, bias], "c")
   f(a, b, d, bb)
   </pre>


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to