# Huawei Cloud Model-as-a-Service on the CloudMatrix384 SuperPod

arXiv: https://arxiv.org/abs/2508.02520 · HTML: https://mlsanity.com/paper/2508.02520/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 4 Aug 2025.

## Built on by (most cited first, 3 of 3)

- [Revealing the Challenges of Attention-FFN Disaggregation for Modern MoE Models and Hardware Systems](https://mlsanity.com/paper/2602.09721/) (2026)
- [Efficient Mixture-of-Experts LLM Inference with Apple Silicon NPUs](https://mlsanity.com/paper/2604.18788/) (2026)
- [FastEEP: Fast Intra-Instance Elastic Scaling for Expert-Parallel MoE Serving](https://mlsanity.com/paper/2610.18451/) (2026)

## Abstract (arXiv)

Scaled-out MoE LLMs and scaled-up SuperPods create new systems challenges for production Model-as-a-Service (MaaS), requiring disaggregation, low-latency communication, and decentralized serving. This report presents xDeepServe, the production serving system behind Huawei Cloud's MaaS offering on CloudMatrix384, a 48-server SuperPod with 384 Ascend 910C chips connected by a high-bandwidth UB fabric and global shared memory. It serves models including DeepSeek, Kimi, GLM, Qwen, and MiniMax, among others. xDeepServe is built around Transformerless, a disaggregated execution architecture that decomposes transformer inference into modular units -- attention, feedforward, and MoE -- and supports disaggregated Prefill-Decode and MoE-Attention deployments. To enable disaggregation, we develop XCCL, a memory-semantic communication layer providing microsecond-level point-to-point and scalable all-to-all primitives, and we extend FlowServe with decentralized DP groups and techniques to mitigate stragglers and synchronization variance. In a peak decoding configuration, xDeepServe reaches 2400 tokens/s per Ascend 910C chip at ~50ms time-per-output-token (TPOT).