[GitHub] [incubator-tvm] comaniac commented on a change in pull request #6560: [BYOC][ETHOSN] Fix tests for new module API

[GitBox <gi...@apache.org>]

comaniac commented on a change in pull request #6560:
URL: https://github.com/apache/incubator-tvm/pull/6560#discussion_r495130579



##########
File path: tests/python/contrib/test_ethosn/test_networks.py
##########
@@ -78,24 +78,16 @@ def get_model():
             )
         return _get_tflite_model(model_path, input_dict, "uint8")
 
-    outputs = []
     inputs = {}
     for input_name in input_dict:
         input_shape = input_dict[input_name]
         inputs[input_name] = tei.get_real_image(input_shape[1], input_shape[2])
 
-    for npu in [False, True]:
-        mod, params = get_model()
-        graph, lib, params = tei.build(
-            mod, params, npu=npu, expected_host_ops=host_ops, npu_partitions=npu_partitions
-        )
-        if npu:
-            tei.assert_lib_hash(lib, compile_hash)
-        if run:
-            outputs.append(tei.run(graph, lib, params, inputs, output_count, npu=npu))
-
+    mod, params = get_model()
+    m = tei.build(mod, params, npu=True, expected_host_ops=host_ops, npu_partitions=npu_partitions)

Review comment:
       After looking into `build`, it seems like it just doesn't go through the Ethos-N partition when `npu=False` but just equivalent to `relay.build`. Accordingly,
   
   - Looks like `npu` will never be False. Do we need to keep this argument?
   - If we need to keep this argument, maybe we need better naming such as `with_npu`. `npu` is somewhat confusing.




----------------------------------------------------------------
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:
users@infra.apache.org