[GH-ISSUE #2093] [Feature Request]: test code for parallel data transfer with multiple Ascend NPU cards #572

New issue

Closed

opened 2026-10-29 21:16:38 +00:00 by bsakaguchi · 1 comment

bsakaguchi commented

2026-10-29 21:16:38 +00:00

Owner

Originally created by @jinsidong on GitHub (May 13, 2026).
Original GitHub issue: https://github.com/kvcache-ai/Mooncake/issues/2093

Originally assigned to: @ascend-direct-dev on GitHub.

Describe your feature request

Question about multi-Ascend NPU parallel transfer test code

Hi developers,

I'm using mooncake-transfer-engine on Huawei Ascend NPU (e.g., 910B).

I want to confirm:

Does this repo currently have test code for parallel data transfer with multiple Ascend NPU cards (single-node multi-card or multi-node multi-card)?
If yes, which directory/file should I refer to (e.g., under tests/ or examples/)?
If not, is there any plan to add such ADXL-based multi-NPU transfer tests in the near future?

My scenario:

Multiple Ascend NPU cards (single machine / two machines)
Concurrent data transfer via ADXL
Need reference test cases to verify stability and performance

Thanks a lot!

Before submitting a new issue...

Make sure you already searched for relevant issues and read the documentation

Originally created by @jinsidong on GitHub (May 13, 2026). Original GitHub issue: https://github.com/kvcache-ai/Mooncake/issues/2093 Originally assigned to: @ascend-direct-dev on GitHub. ### Describe your feature request ## Question about multi-Ascend NPU parallel transfer test code Hi developers, I'm using `mooncake-transfer-engine` on **Huawei Ascend NPU (e.g., 910B)**. I want to confirm: 1. Does this repo currently have **test code for parallel data transfer with multiple Ascend NPU cards** (single-node multi-card or multi-node multi-card)? 2. If yes, which directory/file should I refer to (e.g., under `tests/` or `examples/`)? 3. If not, is there any plan to add such ADXL-based multi-NPU transfer tests in the near future? My scenario: - Multiple Ascend NPU cards (single machine / two machines) - Concurrent data transfer via ADXL - Need reference test cases to verify stability and performance Thanks a lot! ### Before submitting a new issue... - [x] Make sure you already searched for relevant issues and read the [documentation](https://kvcache-ai.github.io/Mooncake/)

bsakaguchi closed this issue

2026-10-29 21:16:38 +00:00

bsakaguchi commented

2026-10-29 21:16:39 +00:00

Author

Owner

@ascend-direct-dev commented on GitHub (May 18, 2026):

dumplicate with 2096

@ascend-direct-dev commented on GitHub (May 18, 2026): dumplicate with 2096

bsakaguchi referenced this issue

2026-10-29 21:29:18 +00:00

[PR #572] [MERGED] [TransferEngine] Add support to force MNNVL transport by MC_FORCE_MNNVL #1662