[MXNET-33] SSD example not working with mkl-dnn (#10021)

* use mkl-dnn for 'valid' pooling_convention only

* pooling convention full not supported by current mkl-dnn impl

* disable unreachable code

* add sample model test for mkldnn

* fix review feedback

* add jira link to comment

* fix lint issue

* rename python test for mkl

* enable python tests for mkldnn in CI

* use vgg16 with convention full

* fix unittest

Jenkinsfile

@@ -107,6 +107,12 @@ def python3_ut(docker_container_name) {
   }
 }
 
+def python3_ut_mkldnn(docker_container_name) {
+  timeout(time: max_time, unit: 'MINUTES') {
+    sh "ci/build.py --build --platform ${docker_container_name} /work/runtime_functions.sh unittest_ubuntu_python3_cpu_mkldnn"
+  }
+}
+
 // GPU test has two parts. 1) run unittest on GPU, 2) compare the results on
 // both CPU and GPU
 // Python 2
@@ -438,7 +444,7 @@ try {
         ws('workspace/ut-python3-mkldnn-cpu') {
           init_git()
           unpack_lib('mkldnn_cpu', mx_mkldnn_lib)
-          python3_ut('ubuntu_cpu')
+          python3_ut_mkldnn('ubuntu_cpu')
         }
       }
     },

ci/docker/runtime_functions.sh

@@ -375,6 +375,18 @@ unittest_ubuntu_python3_cpu() {
     nosetests-3.4 --verbose tests/python/quantization
 }
 
+unittest_ubuntu_python3_cpu_mkldnn() {
+    set -ex
+    export PYTHONPATH=./python/ 
+    # MXNET_MKLDNN_DEBUG is buggy and produces false positives
+    # https://github.com/apache/incubator-mxnet/issues/10026
+    #export MXNET_MKLDNN_DEBUG=1  # Ignored if not present
+    export MXNET_STORAGE_FALLBACK_LOG_VERBOSE=0
+    nosetests-3.4 --verbose tests/python/unittest
+    nosetests-3.4 --verbose tests/python/quantization
+    nosetests-3.4 --verbose tests/python/mkl
+}
+
 unittest_ubuntu_python2_gpu() {
     set -ex
     export PYTHONPATH=./python/

src/operator/nn/mkldnn/mkldnn_pooling-inl.h

@@ -92,12 +92,18 @@ inline bool SupportMKLDNNPooling(const PoolingParam &param,
 
   if (param.pooling_convention == pool_enum::kValid)
     return true;
+  else
+    return false;
 
+// need to support pooling convention full
+// https://issues.apache.org/jira/browse/MXNET-33
+#if 0
   if (((dshape[2] + 2 * param.pad[0] - param.kernel[0]) % param.stride[0] == 0) &&
       ((dshape[3] + 2 * param.pad[1] - param.kernel[1]) % param.stride[1] == 0))
     return true;
   else
     return false;
+#endif
 }
 
 inline bool MKLDNNRequireWorkspace(const PoolingParam &param) {