AnastasiaStulova opened a new pull request #8361: URL: https://github.com/apache/tvm/pull/8361
When dilation is larger than value 1 in conv2d with NHWC layout, the ordering of indexes when accessing data array in computation of convolution appears to be incorrect. 'data_vec' is defined as lambda n, oho, owo, kh, kw, ic, ohi, owi: but accessed as data_vec[n, oho, owo, kh, kw, ohi, owi, ic] This patch fixes the order of indexes. ============================================= **Question:** While the bug is easily observable I wonder if there is some way to improve testing of this even if it might not be possible to add in CI straight away. I used `tests/python/topi/python/test_topi_conv2d_nhwc.py` to catch and test the issue. If it makes sense I could prepare another change adding `arm_cpu` in the list of test devices and/or even running using RPC that can be activated by passing a certain flag. -- 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]
