owensgroup / ml_perf_model

ML performance model for GPU training of DLRM and more.

owensgroup/ml_perf_model Issues

Convolution perf model generalizatoin
Closed 2 years ago1
Study of overhead stats
Closed 2 years ago1
Performance model for variable embedding table size
Closed 3 years ago3
Multi-node multi-GPU support
Updated 3 years ago2
Migrate from nvprof to nsys/ncu
Closed 3 years ago
PyTorch custom operator inputs fix (emb op)
Closed 3 years ago
Any other models that have active time proportion (ATP) as low as DLRM?
Closed 3 years ago1
Get rid of profiler overheads in E2E
Closed 3 years ago1
Get input shapes of batched embedding lookup when extracting execution graph
Closed 3 years ago1
Code refactor
Closed 3 years ago1
Integrate with PARAM
Updated 3 years ago