GitLab

* Persistent group batchnorm added

Added persistent grouped batch norm for performance run on strong scaling case:
currently only supporting:

  1. nhwc layout
  2. fp16
  3. synchronization only within a node!

Environment variable is used to tune LAUNCH_MARGIN that limits the CTAs usage
by the persistent kernel.

Documentation and examples will follow.

* updating type().scalarType() to scalar_type()

* moving launch margin to be defined at layer creation, adding a knob cap max ctas per sm

* fixing the cta computation

* review comment:

set device_id through cudaGetDevice()
move cudaMemset to cudaMemsetAsync
updated __threadfence() to __threadfence_system() inter device write

* change .type().ScalarType() to .scalar_type() + at::ScalarType::X to at::kX

* revert scalar_type() to type() for AT_DISPATCH_FLOATING_TYPES_AND_HALF

* revert scalar_type() to type() in AT_DISPATCH_FLOATING_TYPES

* revert scalar_type() to type() for AT_DISPATCH_FLOATING_TYPES_AND_HALF in welford.cu

* revert scalar_type() to type() in layer_norm_cuda_kernel.cu

* revert at::kType  to at::ScalarType::Type

* use DISPATCH_FLOAT_AND_HALF to get rid of warnings

* add dispatch mechanisms for double+float and double+float+half

The main use of these functions (e.g.: `torch.{conv*, prelu}`) is via their `torch.nn`
wrapping layers.

The `torch.nn` layers are what contain the weights and call into these lower level
functions with the weights as a parameter in their `forward()` method.

The `torch.conv*` functions are already in the `FP16_CASTS` list due to amp's philosophy of
casting the parameters rather than the model/layer weights.

Conceptually `torch.prelu` is the same as the `torch.conv*` case, where its weight parameter
is passed in from its wrapper layer `torch.nn.PReLU`.