Apache MXNet 1.8.0 is a minor release that includes backward compatible changes.
As MXNet follows semver, this release includes backwards-compatible fixes and new features only.

Discussion on the roadmap: https://github.com/apache/incubator-mxnet/issues/18800

Release Managers:

Scope

Reference: MXNet 1.8.0 Roadmap

IssuePRsContributor(s)Notes
Upgrade oneDNN version to v1.6https://github.com/apache/incubator-mxnet/pull/18867@bartekkuncerDone
BatchNorm performance improvement for axis!=1https://github.com/apache/incubator-mxnet/pull/18676@stu1130Done
Users unable to create convolutional layers with large number of filters and large spatial kernel due to upper boundhttps://github.com/apache/incubator-mxnet/pull/18910@wkcnDone
Backward inference supporthttps://github.com/apache/incubator-mxnet/pull/18895@Kh4LDone
TensorRT-Gluon Partition API & TRT 7 Suppoerhttps://github.com/apache/incubator-mxnet/pull/18916@Kh4LDone
ONNX v1.7 supporthttps://github.com/apache/incubator-mxnet/pull/19017@josephevansDone
Stop packaging GPL libquadmath.sohttps://github.com/apache/incubator-mxnet/pull/19055Done
Change Partition API's options_map to std::unordered_maphttps://github.com/apache/incubator-mxnet/pull/18964@Kh4LDone
Support extra inputs for subgraph opshttps://github.com/apache/incubator-mxnet/pull/18894@samskalickyDone
Fix race condition in NaiveEngine::PushAsynchttps://github.com/apache/incubator-mxnet/pull/19122@szhaDone
Update CUB and include it only for CUDA < 11https://github.com/apache/incubator-mxnet/pull/18975@Kh4L
TensorRT: add INT8 with calibrationhttps://github.com/apache/incubator-mxnet/pull/19011@Kh4L
Fix the error of gradient of np.padhttps://github.com/apache/incubator-mxnet/pull/19044@CassiniXuDone
Intgemm support (8-bit matrix multiply library for x86 CPUs)https://github.com/apache/incubator-mxnet/pull/19099kpuatamazon
Update to thrust 1.9.8 on Windowshttps://github.com/apache/incubator-mxnet/pull/18977@stu1130
Fix for duplicate subgraph inputs/outputshttps://github.com/apache/incubator-mxnet/pull/16131@samskalicky
Add cmake flag USE_FATBIN_COMPRESSION, ON by default for CUDA >= 11https://github.com/apache/incubator-mxnet/pull/19123Done








Release Timeline

Following timeline is based on everything goes well.(Added some buffer time)

StepTasksGoalActualComments
Code Freeze and release startTrack ongoing PRsSeptember 18th9/18
Cut the release branchCheck license headersSeptember 19th9/19

Make code changes with necessary version updates on v1.xSeptember 20th9/21#19196

Cut the release branchSeptember 21st9/19
Test the release and tag the releaseNightly test, Jenkins CISeptember 25th9/24

RAT checkSeptember 25th9/24

Tag RC0September 26th






Package artifacts and validateCreate release artifactsSeptember 26th


Validate release packageSeptember 26th


Test release packageSeptember 26th






Begin apache votingStart vote on dev@September 26th


Start vote on general@October 15th






Finalizing and posting the releaseCreate the final release tag on githubOctober 30th


Rename, resign and upload the src tar to final dirOctober 30th


Update the website using tagOctober 31st


Release the official pip packageNovember 1st


Release the official docker imagesNovember 1st


After 24 hrs, validate the packages are uploadedNovember 3rd


Draft the offical announce email and reviewNovember 3rd


Send out the email on announce@November 4th


Update the apache blogNovember 4th


Update the aws blogNovember 4th


Send internal announcementNovember 4th

  • No labels