[PR #315] enhance warp copy efficiency in cached_notify() #517

New issue

Open

opened 2026-10-27 21:03:35 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-27 21:03:35 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/DeepEP/pull/315
Author: @ZhiyiHu1999
Created: 7/18/2025
Status: 🔄 Open

Base: main ← Head: feature/enhance_warp_copy

📝 Commits (1)

4a039a7 enhance warp copy efficiency and remove num_channels <= num_warps restriction in cached_notify()

📊 Changes

1 file changed (+19 additions, -19 deletions)

View changed files

📝 csrc/kernels/internode.cu (+19 -19)

📄 Description

This PR implements a similar approach to #266,

_{🔄 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/315 **Author:** [@ZhiyiHu1999](https://github.com/ZhiyiHu1999) **Created:** 7/18/2025 **Status:** 🔄 Open **Base:** `main` ← **Head:** `feature/enhance_warp_copy` --- ### 📝 Commits (1) - [`4a039a7`](https://github.com/deepseek-ai/DeepEP/commit/4a039a7a8f054c607f1f3a07c21628e9703b386a) enhance warp copy efficiency and remove num_channels <= num_warps restriction in cached_notify() ### 📊 Changes **1 file changed** (+19 additions, -19 deletions) <details> <summary>View changed files</summary> 📝 `csrc/kernels/internode.cu` (+19 -19) </details> ### 📄 Description This PR implements a similar approach to [#266](https://github.com/deepseek-ai/DeepEP/pull/266 ), --- <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 21:03:35 +00:00

bsakaguchi referenced this issue

2026-10-27 21:04:29 +00:00

[PR #517] Add internode Support for the SM80 architecture #606