[GitHub] [incubator-tvm] comaniac edited a comment on pull request #5656: [BYOC] Pattern Language MergeComposite
comaniac edited a comment on pull request #5656: URL: https://github.com/apache/incubator-tvm/pull/5656#issuecomment-634171984 All unit tests including DNNL are passed in my local after rebasing with #5663. @masahi @zhiics @mbaret this is ready for review. 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: us...@infra.apache.org
[GitHub] [incubator-tvm] comaniac edited a comment on pull request #5656: [BYOC] Pattern Language MergeComposite
comaniac edited a comment on pull request #5656: URL: https://github.com/apache/incubator-tvm/pull/5656#issuecomment-632938583 @masahi I found that `test_dnnl_fuse` in `test_pass_partition_graph` is actually failed at the DNNL code compilation, but it won't fail the CI because we don't have DNNL enabled (we should do it soon). The failure in that case is not related to this PR but `bind_params_by_name`. The DNNL function template such as `dnnl_fused_conv2d_bias_relu` expecting 3 inputs, but we generate the function call based on the call node argument, which would be just 1 after `bind_params_by_name`. 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: us...@infra.apache.org
[GitHub] [incubator-tvm] comaniac edited a comment on pull request #5656: [BYOC] Pattern Language MergeComposite
comaniac edited a comment on pull request #5656: URL: https://github.com/apache/incubator-tvm/pull/5656#issuecomment-632938583 @masahi I found that `test_dnnl_fuse` in `test_pass_partition_graph` is actually failed at the DNNL code compilation, but it won't fail the CI because we don't have DNNL enabled (we should do it soon). The failure in that case is not related to this PR but `bind_params_by_name`. The DNNL function template such as `dnnl_fused_conv2d_bias_relu` expecting 3 inputs, but you generate the function based on the call node argument, which would be just 1 after `bind_params_by_name`. 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: us...@infra.apache.org
