Maozhou Ge comments

Repositories
Issues
Comments

Results 13 comments of


                                            Maozhou Ge

[DOC] Example of RMM Python API for DDP distributed training?

@wence- I try to reproduce the error I came across, Can you help on that? Thanks! - test_rmm.py ```python import os import rmm from rmm.allocators.torch import rmm_torch_allocator import torch local_rank...

[DOC] Example of RMM Python API for DDP distributed training?

> Can you try removing the lines: > > ``` > local_rank = int(os.environ.get("LOCAL_RANK", 0)) > rmm.reinitialize(pool_allocator=True, devices=local_rank) > ``` > > And in `demo_basic` function, move the `reinitialize` call...

[QUESTION] how and where to use multi-node trace profiler in paper of megascale

> @oliverYoung2001 Hi, the cuda event monitor tool will be released by the end of July, as said in the README.md. any update?