[PR #381] [MERGED] Fix duplicate calculation of num_tokens_per_rank #544

New issue

Closed

opened 2026-10-27 21:03:50 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-27 21:03:50 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/DeepEP/pull/381
Author: @tianchongchong
Created: 8/28/2025
Status: ✅ Merged
Merged: 9/1/2025
Merged by: @LyricZhao

Base: main ← Head: patch-1

📝 Commits (1)

7ccb9be Update intranode.cu

📊 Changes

1 file changed (+1 additions, -3 deletions)

View changed files

📝 csrc/kernels/intranode.cu (+1 -3)

📄 Description

Each thread is responsible for one target rank. Replace each thread repeat the calculation kNumRanks times

_{🔄 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/381 **Author:** [@tianchongchong](https://github.com/tianchongchong) **Created:** 8/28/2025 **Status:** ✅ Merged **Merged:** 9/1/2025 **Merged by:** [@LyricZhao](https://github.com/LyricZhao) **Base:** `main` ← **Head:** `patch-1` --- ### 📝 Commits (1) - [`7ccb9be`](https://github.com/deepseek-ai/DeepEP/commit/7ccb9bec3510048eb6129603a5ccb8ce17ca78fa) Update intranode.cu ### 📊 Changes **1 file changed** (+1 additions, -3 deletions) <details> <summary>View changed files</summary> 📝 `csrc/kernels/intranode.cu` (+1 -3) </details> ### 📄 Description Each thread is responsible for one target rank. Replace each thread repeat the calculation kNumRanks times --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-27 21:03:50 +00:00