Confusing warning message when mkldnn is turned off

[eric-haibin-lin]

MKLDNN does not expose any special storage type to users, the warning message will make ordinary user confused.
MXNET_MKLDNN_ENABLED=0 python3

Python 3.6.9 (default, Nov  7 2019, 10:44:02)
[GCC 8.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import mxnet as mx
>>> x = mx.nd.ones((1))
>>> net = mx.gluon.nn.Dense(10)
>>> net.initialize(mx.init.Uniform())
>>> net(x)
[16:50:26] src/operator/contrib/../tensor/./../../common/utils.h:473:
Storage type fallback detected:
operator = FullyConnected
input storage types = [default, default, default, ]
output storage types = [default, ]
params = {"no_bias" : False, "__profiler_scope__" : <unk>:, "flatten" : True, "num_hidden" : 10, }
context.dev_mask = cpu
The operator with default storage type will be dispatched for execution. You're seeing this warning message because the operator above is unable to process the given ndarrays with specified storage types, context and parameter. Temporary dense ndarrays are generated in order to execute the operator. This does not affect the correctness of the programme. You can set environment variable MXNET_STORAGE_FALLBACK_LOG_VERBOSE to 0 to suppress this warning.
[16:50:26] src/operator/contrib/../tensor/./../../common/utils.h:473: MXNET_MKLDNN_ENABLED flag is off. You can re-enable by setting MXNET_MKLDNN_ENABLED=1

[[ 0.0068339   0.01299825  0.0301265   0.04819721  0.01438687  0.05011239
   0.00628365  0.04861524 -0.01068833  0.01729892]]
<NDArray 1x10 @cpu(0)>
>>>
>>>

@TaoLv

[TaoLv]

@eric-haibin-lin The storage fallback logic is reused for disabling MKL-DNN at runtime. See #12058 (comment).

[TaoLv]

It's mainly due to InferStorageType is not only used for sparse/dense dispatching, but also for MKL-DNN operator dispatching.

[eric-haibin-lin]

yes.. but it's still confusing to other users

[pengzhao-intel]

@anko-intel