Joel Mathew Cherian

Improving dataset loader and preprocess script

4

There are some severe speed issues with the preprocess and data loader script and this oftens makes benchmarking a rather tedious process. I'll work on clearing up the technical debt...

Optimizations and Corrections

3

## (CORRECTION) GPMA Node labelling should start from 1 The node labelling in GPMA should start from 1, the reason is because when GPMA is initialized by default a sentinel...

Seastar - RGCN

4

Seastar's original implementation does not present a vertex centric program for RGCN, it rather uses a handwritten kernel in dgl-hack. Let's try to write a vertex-centric program for RGCN, this...

Redesigning the Seastar Tracer (Frontend)

5

The need to redesign the seastar frontend are as follows 1. Monkey-patching is considered a very hacky solution and it cannot be pushed to production 2. The current tracer design...

Handling CUDA kernel code generation effectively

1

This issue will look into the current templating approach to CUDA kernel code generation and verify its scalability and effectiveness. Additionally, we can look into good packages that handle the...

This issue will explore the need for neural network layers to be used within the vertex-centric program. If a need is seen, this issue can further discuss how this feature...