[v1.4.x] Update MKL-DNN to fix the OSX build issue

[TaoLv]

Description

Find the issue description at: https://lists.apache.org/thread.html/1d5883444907f0a06d711d1f482feeadcbe1d523e9944e953a26cf51@%3Cdev.mxnet.apache.org%3E
and
oneapi-src/oneDNN#405

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

• The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant JIRA issue created (except PRs with tiny changes)
• Changes are complete (i.e. I finished coding on this PR)
• All changes have test coverage:
• Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
• Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
• Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
• Code is well-documented:
• For user-facing API changes, API doc string has been updated.
• For new C++ functions in header files, their functionalities and arguments are documented.
• For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
• Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
• To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

Changes

• Feature1, tests, (and when applicable, API doc)
• Feature2, tests, (and when applicable, API doc)

Comments

• If this change is a backward incompatible change, why must this change be made.
• Interesting edge cases to note here

[TaoLv]

@szha @srochel @pengzhao-intel

[TaoLv]

Weird CI failures.. :(

[ankkhedia]

@TaoLv Thanks for your contribution! Could you please look into CI failures?

@mxnet-label-bot add [pr-awaiting-testing, MKLDNN, Build]

[lanking520]

Restarted :), @marcoabreu could you also grant CI access to @TaoLv so he will be able to retrigger the test next time

[piyushghai]

@TaoLv
A unit test seems to be failing on this PR :

test_operator_gpu.test_stn
ValueError: threshold must be numeric and non-NAN, try sys.maxsize for untruncated representation

Not sure if this is related to your changes in this PR.

[lanking520]

Retrigger the test again, not sure if there is an issue with Cent OS. But I do saw multiple failure in the Cent OS testing

[TaoLv]

Thank you for your attention @lanking520, @piyushghai. The test_stn issue was fixed on master branch several day ago: #14063. But the fix is not ported to v1.4.x branch.

[szha]

@TaoLv I cherry-picked the fix. I will merge this PR if there's no other issue.

[szha]

Errors:
http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/incubator-mxnet/detail/PR-14141/5/pipeline#step-1982-log-1537 (test_stn)
http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/mxnet-validation%2Fcentos-gpu/detail/PR-14141/4/pipeline/#step-97-log-1536 (test_stn)
http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/incubator-mxnet/detail/PR-14141/5/pipeline/1121#step-1540-log-2421 (R package coverage, seems transient)
http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/mxnet-validation%2Fwebsite/detail/PR-14141/4/pipeline/#step-118-log-1420 (clojure, @gigasquid does it look like the problem you've fixed on master?) (update: I cherry-picked the clojure doc fixes to v1.4.x branch)

[szha]

None looks related to mkldnn.

[lanking520]

@TaoLv Contribute back to master?

[TaoLv]

MKL-DNN 0.18 is coming soon: /~https://github.com/intel/mkl-dnn/releases/tag/v0.18-rc. We plan to update master directly to 0.18 release. Internal validation is WIP.

[szha]

@TaoLv master pip build for mac mkl is still failing, so I'd recommend patching master too, ASAP.

[piyushghai]

@TaoLv The OSX build from source is not working still. It's broken.

I tried a simple command : make -j4

[piyushghai]

Update --> Ignore my previous comment. I resolved the issue thanks to @lanking520