[PR #244] [MERGED] [Feature] support auto-detect topo for both GPUs and CPUs #1424

New issue

Closed

opened 2026-10-29 21:27:19 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-29 21:27:19 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/kvcache-ai/Mooncake/pull/244
Author: @alogfans
Created: 4/14/2025
Status: ✅ Merged
Merged: 4/16/2025
Merged by: @doujiang24

Base: main ← Head: detect-topo

📝 Commits (2)

cae6bac [Feature] support auto-detect topo for both GPUs and CPUs
aa323e4 Merge branch 'main' into detect-topo

📊 Changes

10 files changed (+101 additions, -46 deletions)

View changed files

📝 mooncake-integration/transfer_engine/transfer_engine_py.cpp (+51 -30)
📝 mooncake-integration/transfer_engine/transfer_engine_py.h (+1 -0)
📝 mooncake-transfer-engine/include/memory_location.h (+3 -2)
📝 mooncake-transfer-engine/include/transfer_engine.h (+2 -1)
📝 mooncake-transfer-engine/src/memory_location.cpp (+32 -3)
📝 mooncake-transfer-engine/src/topology.cpp (+3 -4)
📝 mooncake-transfer-engine/src/transport/rdma_transport/rdma_transport.cpp (+4 -2)
📝 mooncake-transfer-engine/tests/memory_location_test.cpp (+1 -1)
📝 mooncake-transfer-engine/tests/rdma_transport_test.cpp (+1 -1)
📝 mooncake-transfer-engine/tests/topology_test.cpp (+3 -2)

📄 Description

To enable fully auto detection of topology in Transfer Engine, just leave the device name empty in the Python interface.

_{🔄 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/244 **Author:** [@alogfans](https://github.com/alogfans) **Created:** 4/14/2025 **Status:** ✅ Merged **Merged:** 4/16/2025 **Merged by:** [@doujiang24](https://github.com/doujiang24) **Base:** `main` ← **Head:** `detect-topo` --- ### 📝 Commits (2) - [`cae6bac`](https://github.com/kvcache-ai/Mooncake/commit/cae6bac7d0a78cce42190f4c77dcac68ace12d45) [Feature] support auto-detect topo for both GPUs and CPUs - [`aa323e4`](https://github.com/kvcache-ai/Mooncake/commit/aa323e470604375d4e712eb46122bc1185eaaa58) Merge branch 'main' into detect-topo ### 📊 Changes **10 files changed** (+101 additions, -46 deletions) <details> <summary>View changed files</summary> 📝 `mooncake-integration/transfer_engine/transfer_engine_py.cpp` (+51 -30) 📝 `mooncake-integration/transfer_engine/transfer_engine_py.h` (+1 -0) 📝 `mooncake-transfer-engine/include/memory_location.h` (+3 -2) 📝 `mooncake-transfer-engine/include/transfer_engine.h` (+2 -1) 📝 `mooncake-transfer-engine/src/memory_location.cpp` (+32 -3) 📝 `mooncake-transfer-engine/src/topology.cpp` (+3 -4) 📝 `mooncake-transfer-engine/src/transport/rdma_transport/rdma_transport.cpp` (+4 -2) 📝 `mooncake-transfer-engine/tests/memory_location_test.cpp` (+1 -1) 📝 `mooncake-transfer-engine/tests/rdma_transport_test.cpp` (+1 -1) 📝 `mooncake-transfer-engine/tests/topology_test.cpp` (+3 -2) </details> ### 📄 Description To enable fully auto detection of topology in Transfer Engine, just leave the device name empty in the Python interface. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-29 21:27:19 +00:00