FranklandJack commented on PR #14519: URL: https://github.com/apache/tvm/pull/14519#issuecomment-1514455412
> > I think at present though we don't do any testing for Arm Cortex A targets (see [here](https://github.com/apache/tvm/blob/main/tests/python/relay/strategy/arm_cpu/test_generalized_conv2d.py#L149)), so perhaps this is outside the scope of this PR? > > I thought that this schedule is a common depthwise conv2d schedule for ARM and doesn't matter is it Cortex A or any other target, am I right? But anyway, I'm ok if we merge this PR w/o test, because it is pretty simple and probably writing a good test which will check generated tir, will require too much efford. Sorry, yes you are correct, other arm CPU targets will also go through this path. Although for the test we currently only test the cortex-M target with the aot cornerstone 300 runner, which I think means we will always go the CMSIS-NN path. I think you are probably right and we can merge this without a test, although you've identified we are missing test coverage for these target schedules, so I think after this is merged I'll follow up with another PR that fixes that. -- 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]
