srkreddy1238 commented on code in PR #13670:
URL: https://github.com/apache/tvm/pull/13670#discussion_r1059571173
##########
tests/python/contrib/test_clml/test_ops.py:
##########
@@ -54,11 +62,8 @@ def _get_conv_model(
shape = (shape[0], shape[1], shape[2] + padding[0] * 2, shape[3] +
padding[1] * 2)
is_depthwise = shape[1] == channels == groups
- weight_format = "OIHW" if is_depthwise else "OIHW"
- if weight_format == "IOHW":
- weight_shape = (shape[1] // groups, channels, kernel_h, kernel_w)
- else:
- weight_shape = (channels, shape[1] // groups, kernel_h, kernel_w)
+ weight_format = "OIHW"
Review Comment:
This was the issue we had earlier with depthwise.
Weight layout transform for depthwise is handled internally by CLML
compilation process internally.
https://github.com/apache/tvm/blob/6f6bf1912e08f5113c1ac438ea600270b479055d/python/tvm/relay/op/contrib/clml.py#L79
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
