Kuo Zhang

Ways to do overlapping of computation and commucation.

1

Hello, I'm trying to use DeepEP and intergate it into Megatron (with this commit https://github.com/NVIDIA/Megatron-LM/commit/0d389f5e9935957b146c12b1dcdddc35cf9deb85), and accomplish overlapping of computation and commucation, I've noticed DeepEP is using previous_event and async_finish...

By default, in megatron, condition fais on ibgda_get_state()->num_rc_per_pe == num_channels || ibgda_get_state()->num_rc_per_pe >= num_sms

4

After updated deepep to 7f97f79bda051d991aa9681c04138007faa0366, it fails: ``` 2025-06-19 16:18 [rank5]: File "/opt/Megatron-LM/megatron/core/transformer/moe/moe_layer.py", line 156, in forward 2025-06-19 16:18 [rank5]: output, mlp_bias = custom_forward(hidden_states) 2025-06-19 16:18 [rank5]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2025-06-19 16:18...

Any plan of releases or stable tags?

1

Hello, the deepep is in rapid evolution, is there any plan of releases version or stable tags?