[PR #106] [MERGED] [TransferEngine] feature: introduce USE_NVMEOF to enable NVMe-oF separately. #1325

New issue

Closed

opened 2026-10-29 21:26:26 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-29 21:26:26 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/kvcache-ai/Mooncake/pull/106
Author: @doujiang24
Created: 2/14/2025
Status: ✅ Merged
Merged: 2/20/2025
Merged by: @stmatengss

Base: main ← Head: cufile

📝 Commits (1)

97888e4 [TransferEngine] feature: introduce USE_NVMEOF to enable NVMe-oF separately.

📊 Changes

9 files changed (+33 additions, -15 deletions)

View changed files

📝 CMakeLists.txt (+10 -0)
📝 mooncake-transfer-engine/example/transfer_engine_bench.cpp (+5 -3)
📝 mooncake-transfer-engine/src/CMakeLists.txt (+4 -1)
📝 mooncake-transfer-engine/src/multi_transport.cpp (+3 -4)
📝 mooncake-transfer-engine/src/transport/CMakeLists.txt (+1 -1)
📝 mooncake-transfer-engine/src/transport/nvmeof_transport/cufile_context.cpp (+1 -1)
📝 mooncake-transfer-engine/tests/rdma_transport_test.cpp (+5 -3)
📝 mooncake-transfer-engine/tests/rdma_transport_test2.cpp (+1 -1)
📝 mooncake-transfer-engine/tests/tcp_transport_test.cpp (+3 -1)

📄 Description

The current nvmeof transport only support CUDA 12, we can keep it disabled for older CUDA versions, as a workaround now.
Also, rename the gpu: prefix to cuda:, to align with doc & topology.cpp.

_{🔄 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/106 **Author:** [@doujiang24](https://github.com/doujiang24) **Created:** 2/14/2025 **Status:** ✅ Merged **Merged:** 2/20/2025 **Merged by:** [@stmatengss](https://github.com/stmatengss) **Base:** `main` ← **Head:** `cufile` --- ### 📝 Commits (1) - [`97888e4`](https://github.com/kvcache-ai/Mooncake/commit/97888e48cba55325829011e417d7c26dbe936593) [TransferEngine] feature: introduce USE_NVMEOF to enable NVMe-oF separately. ### 📊 Changes **9 files changed** (+33 additions, -15 deletions) <details> <summary>View changed files</summary> 📝 `CMakeLists.txt` (+10 -0) 📝 `mooncake-transfer-engine/example/transfer_engine_bench.cpp` (+5 -3) 📝 `mooncake-transfer-engine/src/CMakeLists.txt` (+4 -1) 📝 `mooncake-transfer-engine/src/multi_transport.cpp` (+3 -4) 📝 `mooncake-transfer-engine/src/transport/CMakeLists.txt` (+1 -1) 📝 `mooncake-transfer-engine/src/transport/nvmeof_transport/cufile_context.cpp` (+1 -1) 📝 `mooncake-transfer-engine/tests/rdma_transport_test.cpp` (+5 -3) 📝 `mooncake-transfer-engine/tests/rdma_transport_test2.cpp` (+1 -1) 📝 `mooncake-transfer-engine/tests/tcp_transport_test.cpp` (+3 -1) </details> ### 📄 Description The current nvmeof transport only support CUDA 12, we can keep it disabled for older CUDA versions, as a workaround now. Also, rename the `gpu:` prefix to `cuda:`, to align with doc & topology.cpp. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-29 21:26:26 +00:00