[WIP] typedef for MKL_INT to point to int64_t when building with Large Tensor

[access2rohit]

Description

This PR makes MKL_INT a typedef for int64_t when using large tensor thus avoiding the following error when building with MKL as blas and the compiler flag -Werror=narrowing
Error:

[2020-06-27T06:00:01.634Z] /work/mxnet/src/operator/contrib/transformer.cc: In function 'void mxnet::op::strided_batch_sgemm(bool, bool, mxnet::index_t, mxnet::index_t, mxnet::index_t, float, const float*, mxnet::index_t, mxnet::index_t, const float*, mxnet::index_t, mxnet::index_t, float, float*, mxnet::index_t, mxnet::index_t, int32_t)':

[2020-06-27T06:00:01.634Z] /work/mxnet/src/operator/contrib/transformer.cc:143:31: error: narrowing conversion of 'm' from 'mxnet::index_t {aka long int}' to 'int' inside { } [-Werror=narrowing]

[2020-06-27T06:00:01.634Z]    MKL_INT p_m[GROUP_SIZE] = {m};

[2020-06-27T06:00:01.634Z]                                ^

[2020-06-27T06:00:01.634Z] /work/mxnet/src/operator/contrib/transformer.cc:144:31: error: narrowing conversion of 'n' from 'mxnet::index_t {aka long int}' to 'int' inside { } [-Werror=narrowing]

[2020-06-27T06:00:01.634Z]    MKL_INT p_n[GROUP_SIZE] = {n};

[2020-06-27T06:00:01.634Z]                                ^

[2020-06-27T06:00:01.634Z] /work/mxnet/src/operator/contrib/transformer.cc:145:31: error: narrowing conversion of 'k' from 'mxnet::index_t {aka long int}' to 'int' inside { } [-Werror=narrowing]

[2020-06-27T06:00:01.634Z]    MKL_INT p_k[GROUP_SIZE] = {k};

[2020-06-27T06:00:01.634Z]                                ^

[2020-06-27T06:00:01.634Z] /work/mxnet/src/operator/contrib/transformer.cc:146:35: error: narrowing conversion of 'lda' from 'mxnet::index_t {aka long int}' to 'int' inside { } [-Werror=narrowing]

[2020-06-27T06:00:01.634Z]    MKL_INT p_lda[GROUP_SIZE] = {lda};

[2020-06-27T06:00:01.634Z]                                    ^

[2020-06-27T06:00:01.634Z] /work/mxnet/src/operator/contrib/transformer.cc:147:35: error: narrowing conversion of 'ldb' from 'mxnet::index_t {aka long int}' to 'int' inside { } [-Werror=narrowing]

[2020-06-27T06:00:01.634Z]    MKL_INT p_ldb[GROUP_SIZE] = {ldb};

[2020-06-27T06:00:01.634Z]                                    ^

[2020-06-27T06:00:01.634Z] /work/mxnet/src/operator/contrib/transformer.cc:148:35: error: narrowing conversion of 'ldc' from 'mxnet::index_t {aka long int}' to 'int' inside { } [-Werror=narrowing]

[2020-06-27T06:00:01.634Z]    MKL_INT p_ldc[GROUP_SIZE] = {ldc};

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

• 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)
• Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
• To the best of my knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

[mxnet-bot]

Hey @access2rohit , Thanks for submitting the PR
All tests are already queued to run once. If tests fail, you can trigger one or more tests again with the following commands:

• To trigger all jobs: @mxnet-bot run ci [all]
• To trigger specific jobs: @mxnet-bot run ci [job1, job2]

---

CI supported jobs: [miscellaneous, windows-gpu, clang, website, sanity, windows-cpu, edge, unix-gpu, unix-cpu, centos-gpu, centos-cpu]

---

Note:
Only following 3 categories can trigger CI :PR Author, MXNet Committer, Jenkins Admin.
All CI tests must pass before the PR can be merged.

[TaoLv]

Please refer to https://software.intel.com/content/www/us/en/develop/documentation/mkl-macos-developer-guide/top/linking-your-application-with-the-intel-math-kernel-library/linking-in-detail/linking-with-interface-libraries/using-the-ilp64-interface-vs-lp64-interface.html

The cmake flag MKL_USE_ILP64 needs to be set to enable MKL ILP64 interface.

[szha]

Thanks @access2rohit for the change and thank you @TaoLv for the context. I think we should also turn on MKL_USE_ILP64 automatically in cmake when both MKL and large tensor is used.

[access2rohit]

Not required as its enabled by default now :)