* Add device support in hetero_rgcn
Signed-off-by: Xue, Chendi <chendi.xue@intel.com>
* use num_workers instead of hard code and enable cpu_affinity for pytorch > 1.12
Signed-off-by: Xue, Chendi <chendi.xue@intel.com>
* Remove hard code num_workers and use dgl version to check if cpu_affinity is supported
Signed-off-by: Xue, Chendi <chendi.xue@intel.com>
* Remove specified dl_cores and computer_cores
Add error print if num_workers are miss set
Signed-off-by: Xue, Chendi <chendi.xue@intel.com>
* expected num_workers should be less than num_physical_cores
Signed-off-by: Chendi Xue <chendi.xue@intel.com>
* Update examples/pytorch/ogb/ogbn-mag/hetero_rgcn.py
Co-authored-by: Chang Liu <chang.liu@utexas.edu>
* Remove dgl version and num_workers print
Signed-off-by: Xue, Chendi <chendi.xue@intel.com>
* add comment to explain is_support_affinity
* Fix typo
Signed-off-by: Xue, Chendi <chendi.xue@intel.com>
Signed-off-by: Xue, Chendi <chendi.xue@intel.com>
Signed-off-by: Chendi Xue <chendi.xue@intel.com>
Co-authored-by: Chang Liu <chang.liu@utexas.edu>
* rgcn-ogbn-mag
* Add link in README.md
* correct code-format,add the reset_parameters function to the HeteroEmbedding module
* add the annotation in hetero.py
* add a unit test
* modify format
* Update
Co-authored-by: Mufei Li <mufeili1996@gmail.com>
Co-authored-by: Ubuntu <ubuntu@ip-172-31-50-143.us-west-2.compute.internal>
* Adding initial files of example
* Removing old timing code
* Improving doc strings and fixing some minor bugs
* Merging from upstream and addressing PR comments
Co-authored-by: zakjost <jostza@amazon.com>
Co-authored-by: Mufei Li <mufeili1996@gmail.com>