masahi commented on a change in pull request #8669:
URL: https://github.com/apache/tvm/pull/8669#discussion_r685574682
##########
File path: python/tvm/topi/x86/dense_alter_op.py
##########
@@ -39,6 +40,17 @@ def _alter_dense_layout(attrs, inputs, tinfos, out_type):
relay.op.get("nn.dense"), attrs, tinfos, out_type, target
)
workload = autotvm.task.get_workload(outs)
+
+ data_type =
relay.transform.InferType()(IRModule.from_expr(inputs[0]))["main"].body.checked_type
Review comment:
Also I want to mention that, normally I would expect that
`InferCorrectLayout` would specify the desired layout and layout transform
would be inserted automatically. But I got an impression that
`InferCorrectLayout` doesn't really work with `AlterOpLayout` (unlike
`ConvertLayout`), in that `AlterOpLayout` requires shape information. So we
need to 1. get shape of the transformed input and 2. insert layout transform
inside this call back. I thought this design is weird but it seems this is the
only way.
--
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]
