mkolod commented on issue #11380: Add ability to query cuDNN BatchNorm min. epsilon. Allow ONNX importer to use cuDNN BN if chosen eps >= cuDNN min. eps. URL: https://github.com/apache/incubator-mxnet/pull/11380#issuecomment-402290583 @marcoabreu Note that all tests succeed except for [MKLDNN-GPU](http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/incubator-mxnet/detail/PR-11380/15/pipeline), which gives: ```test_operator_gpu.test_sparse_hybrid_block_grad ... [19:36:39] src/operator/tensor/./.././../common/../operator/mxnet_op.h:576: Check failed: (err) == (cudaSuccess) Name: mxnet_generic_kernel ErrStr:an illegal memory access was encountered ``` No other GPU build reports such an issue. 48 tests fail this way, but they don't fail for any other GPU build. Seems like that runner is having issues. Note that my code only pertains to the ONNX importer (see [here](https://github.com/apache/incubator-mxnet/pull/11380/files)), and these tests aren't even related to the ONNX importer.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
