Search results

deepEP run two nodes, each nodes equipped 1 A100+1CX6, hit assert.

Status: Open.

#104 In deepseek-ai/DeepEP;

· Knight-Cai opened on Mar 31, 2025

python test_low_latency.py blocked in H20

Status: Open.

#103 In deepseek-ai/DeepEP;

· cll24 opened on Mar 31, 2025

Why normal mode does not support AR?

Status: Open.

#100 In deepseek-ai/DeepEP;

· yejunhong1 opened on Mar 31, 2025

Suggestions for Improving the Readability of DeepEP Code

Status: Open.

#98 In deepseek-ai/DeepEP;

· zuochunwei opened on Mar 28, 2025

Question about the global synchronization/barrier in internode notify_dispatch

Status: Open.

#94 In deepseek-ai/DeepEP;

· songhexiang opened on Mar 26, 2025

Will the Normal Kernel (dispatch kernel) exclusively occupy the SM of the GPU?

Status: Open.

#92 In deepseek-ai/DeepEP;

· PhilipJohnson2 opened on Mar 25, 2025

"RuntimeError: Failed: CUDA error /opt/DeepEP/csrc/kernels/internode.cu:133 ' named symbol not found" in container when running test_intranode.py

Status: Open.

#87 In deepseek-ai/DeepEP;

· yuwenjingsei opened on Mar 24, 2025

socketStartConnect: exceeded timeouts (3)socketStartConnect: exceeded timeouts (3)

Status: Open.

#86 In deepseek-ai/DeepEP;

· zhoutianzi666 opened on Mar 21, 2025

Benchmark test over RoCE network

Status: Open.

#82 In deepseek-ai/DeepEP;

· yanminjia opened on Mar 19, 2025

About the number of messages chunked in IBGDA

Status: Open.

#80 In deepseek-ai/DeepEP;

· GitAlice123 opened on Mar 19, 2025

Could normal kernel and ll kernel be executed in same process?

Status: Open.

#76 In deepseek-ai/DeepEP;

· lowintelligence opened on Mar 15, 2025

curious about the difference between latency and throughput

Status: Open.

#75 In deepseek-ai/DeepEP;

· Eutenacity opened on Mar 15, 2025