[PR #1098] [MERGED] [EP] Support multiple torch versions

No labels

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

kvcache-ai/Mooncake#2010

bsakaguchi commented

2026-10-29 21:32:06 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/kvcache-ai/Mooncake/pull/1098
Author: @UNIDY2002
Created: 11/24/2025
Status: ✅ Merged
Merged: 11/28/2025
Merged by: @ShangmingCai

Base: main ← Head: sunxun/torch-version-fix

📝 Commits (10+)

83d81e3 Debug
7dcaf49 Debug
ca3c750 Debug
8c99d18 Debug
f013611 Debug
115cada Debug
122980e Debug
b4cfb10 Debug
ed4fd4c Debug
9e89267 Debug

📊 Changes

9 files changed (+89 additions, -60 deletions)

View changed files

📝 .github/workflows/ci.yml (+2 -1)
📝 .github/workflows/release-non-cuda.yaml (+0 -1)
📝 .github/workflows/release.yaml (+2 -1)
📝 CMakeLists.txt (+1 -3)
➕ mooncake-ep/setup.py (+46 -0)
📝 mooncake-integration/ep/ep_py.cpp (+1 -6)
➕ mooncake-wheel/mooncake/ep.py (+16 -0)
📝 mooncake-wheel/setup.py (+4 -38)
📝 scripts/build_wheel.sh (+17 -10)

📄 Description

Description

Build the Mooncake EP Backend's dynamic library against various torch versions.

Switch to setuptools to build Mooncake EP.

In addition, strip debug info for Mooncake EP to optimize wheel size (each version consumes 1.9 MB, which is acceptable).

Type of Change

Types
- Bug fix
- New feature
  - Transfer Engine
  - Mooncake Store
  - Mooncake EP
  - Integration
  - P2P Store
  - Python Wheel
- Breaking change
- CI/CD
- Documentation update
- Other

How Has This Been Tested?

Build the library and use it with different torch versions. The unit tests related to EP & Backend should always pass. E2E test should also pass.

Checklist

I have performed a self-review of my own code.
I have updated the documentation.
I have added tests to prove my changes are effective.

_{🔄 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/kvcache-ai/Mooncake/pull/1098 **Author:** [@UNIDY2002](https://github.com/UNIDY2002) **Created:** 11/24/2025 **Status:** ✅ Merged **Merged:** 11/28/2025 **Merged by:** [@ShangmingCai](https://github.com/ShangmingCai) **Base:** `main` ← **Head:** `sunxun/torch-version-fix` --- ### 📝 Commits (10+) - [`83d81e3`](https://github.com/kvcache-ai/Mooncake/commit/83d81e302541924aa6903ecd104ceee2a357488b) Debug - [`7dcaf49`](https://github.com/kvcache-ai/Mooncake/commit/7dcaf49f8d9d288c31aaec4b02bc3438c8a5f5f3) Debug - [`ca3c750`](https://github.com/kvcache-ai/Mooncake/commit/ca3c7501c72677c066096ade1c5bbd015b0a8591) Debug - [`8c99d18`](https://github.com/kvcache-ai/Mooncake/commit/8c99d181e3b30d8b2269fdfdce35b3b4f098b55b) Debug - [`f013611`](https://github.com/kvcache-ai/Mooncake/commit/f0136114d92e7d7cf41c0ea74e156dc20dc805f1) Debug - [`115cada`](https://github.com/kvcache-ai/Mooncake/commit/115cada3c6bbdaefff3a4f1a9ec49ff3246b6093) Debug - [`122980e`](https://github.com/kvcache-ai/Mooncake/commit/122980eda28f07f111fab6ed291d9fa9a778aec1) Debug - [`b4cfb10`](https://github.com/kvcache-ai/Mooncake/commit/b4cfb108a82b15e89a783a1ef0211de689a1a88b) Debug - [`ed4fd4c`](https://github.com/kvcache-ai/Mooncake/commit/ed4fd4c910a7bc55bd8f8402f7b6d727c4366ca4) Debug - [`9e89267`](https://github.com/kvcache-ai/Mooncake/commit/9e89267b4e85333480bd1de6d153b9c6126f342c) Debug ### 📊 Changes **9 files changed** (+89 additions, -60 deletions) <details> <summary>View changed files</summary> 📝 `.github/workflows/ci.yml` (+2 -1) 📝 `.github/workflows/release-non-cuda.yaml` (+0 -1) 📝 `.github/workflows/release.yaml` (+2 -1) 📝 `CMakeLists.txt` (+1 -3) ➕ `mooncake-ep/setup.py` (+46 -0) 📝 `mooncake-integration/ep/ep_py.cpp` (+1 -6) ➕ `mooncake-wheel/mooncake/ep.py` (+16 -0) 📝 `mooncake-wheel/setup.py` (+4 -38) 📝 `scripts/build_wheel.sh` (+17 -10) </details> ### 📄 Description ## Description  Build the Mooncake EP Backend's dynamic library against various torch versions. Switch to setuptools to build Mooncake EP. In addition, strip debug info for Mooncake EP to optimize wheel size (each version consumes 1.9 MB, which is acceptable). ## Type of Change * Types - [x] Bug fix - [ ] New feature - [ ] Transfer Engine - [ ] Mooncake Store - [ ] Mooncake EP - [ ] Integration - [ ] P2P Store - [ ] Python Wheel - [ ] Breaking change - [ ] CI/CD - [ ] Documentation update - [ ] Other ## How Has This Been Tested?  Build the library and use it with different torch versions. The unit tests related to EP & Backend should always pass. E2E test should also pass. ## Checklist - [x] I have performed a self-review of my own code. - [ ] I have updated the documentation. - [ ] I have added tests to prove my changes are effective. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

closed this issue
added the
pull-request
label

Rows
Columns

[PR #1098] [MERGED] [EP] Support multiple torch versions #2010