[PR #740] fix: get_nvlink_gbs may return 0 as a divisor #745

New issue

Open

opened 2026-10-27 21:05:49 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-27 21:05:49 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/DeepEP/pull/740
Author: @cxxlink
Created: 8/27/2026
Status: 🔄 Open

Base: main ← Head: fix_div_0

📝 Commits (1)

7f6280c fix: get_nvlink_gbs may return 0 as a divisor

📊 Changes

1 file changed (+9 additions, -1 deletions)

View changed files

📝 deep_ep/buffers/elastic.py (+9 -1)

📄 Description

Problem

Function get_nvlink_gbs returns zero and unprotected is used as a divisor.

Root cause

The result of command "nvidia-smi nvlink -s" execution is empty, if the environment does not support nvink.

Fix

Add protection before division.

Validation

Protection takes effect before division.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/deepseek-ai/DeepEP/pull/740 **Author:** [@cxxlink](https://github.com/cxxlink) **Created:** 8/27/2026 **Status:** 🔄 Open **Base:** `main` ← **Head:** `fix_div_0` --- ### 📝 Commits (1) - [`7f6280c`](https://github.com/deepseek-ai/DeepEP/commit/7f6280c845ccfb50af7d2a6c966f6d1d1c52fdef) fix: get_nvlink_gbs may return 0 as a divisor ### 📊 Changes **1 file changed** (+9 additions, -1 deletions) <details> <summary>View changed files</summary> 📝 `deep_ep/buffers/elastic.py` (+9 -1) </details> ### 📄 Description # Problem Function get_nvlink_gbs returns zero and unprotected is used as a divisor. # Root cause The result of command "**nvidia-smi nvlink -s**" execution is empty, if the environment does not support nvink. # Fix Add protection before division. # Validation Protection takes effect before division. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi added the

pull-request

label

2026-10-27 21:05:49 +00:00

bsakaguchi referenced this issue

2026-10-27 21:05:52 +00:00

[PR #745] [MERGED] [HybridEP] Fix aarch64 inter-node JIT, CUDA graph capture, and test EP topology #750