openxla / xla

A machine learning compiler for GPUs, CPUs, and ML accelerators
Apache License 2.0
2.55k stars 394 forks source link

[XLA:MSA] Implement an auxiliary function to simulate the async memory request overhead. #14874

Open copybara-service[bot] opened 1 month ago

copybara-service[bot] commented 1 month ago

[XLA:MSA] Implement an auxiliary function to simulate the async memory request overhead.

The off-the-shelf runtime predictor does not include the overhead of asynchronous copies. We need to simulate the memory system to get the overhead of these copies. We implement an auxiliary function is this patch to simulate the bandwidth sharing model. This function can be further used to support end2end simulator that involves async copies.