DickJC123 commented on issue #14329: [Flaky] flaky test in
test_operator_gpu.test_convolution_multiple_streams
URL:
https://github.com/apache/incubator-mxnet/issues/14329#issuecomment-470630937
I need to point out that a recently merged "op bulking" PR saw a segfault in
either the
DickJC123 commented on issue #14329: [Flaky] flaky test in
test_operator_gpu.test_convolution_multiple_streams
URL:
https://github.com/apache/incubator-mxnet/issues/14329#issuecomment-469914392
Rather than open up a new issue to track any follow-up PR from @arcadiaphy
I thought it best
DickJC123 commented on issue #14329: [Flaky] flaky test in
test_operator_gpu.test_convolution_multiple_streams
URL:
https://github.com/apache/incubator-mxnet/issues/14329#issuecomment-469885039
To debug this further, I checked out the ASAN PR commit, reverted my
dual-stream PR, then
DickJC123 commented on issue #14329: [Flaky] flaky test in
test_operator_gpu.test_convolution_multiple_streams
URL:
https://github.com/apache/incubator-mxnet/issues/14329#issuecomment-469531735
I believe I've isolated the problem. I started by checking out the commit
that introduced the
DickJC123 commented on issue #14329: [Flaky] flaky test in
test_operator_gpu.test_convolution_multiple_streams
URL:
https://github.com/apache/incubator-mxnet/issues/14329#issuecomment-469494252
Looking at this now. The problem occurs during shutdown of the system when