# MoE Expert Execution in Disaggregated LLM Serving with a High-Bandwidth ReRAM Near-Memory Architecture

arXiv: https://arxiv.org/abs/2608.13962 · HTML: https://mlsanity.com/paper/2608.13962/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Efficient MoE training and inference systems, one of 12 directions in Mixture-of-experts.
- Builds on 21 of its 37 references that are in the map.
- First posted on arXiv 14 Aug 2026.

## Placed in

- Mixture-of-experts / [Efficient MoE training and inference systems](https://mlsanity.com/moe/efficient-moe-training-and-inference-systems/)
- Efficient LLM inference and serving / [Diffusion LLM and MoE inference](https://mlsanity.com/inference/diffusion-llm-and-moe-inference/)

## Builds on (most cited first, 21 of 21)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Mixture-of-Experts with Expert Choice Routing](https://mlsanity.com/paper/2202.09368/) (2022)
- [DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving](https://mlsanity.com/paper/2401.09670/) (2024)
- [Splitwise: Efficient generative LLM inference using phase splitting](https://mlsanity.com/paper/2311.18677/) (2023)
- [Efficiently Scaling Transformer Inference](https://mlsanity.com/paper/2211.05102/) (2022)
- [Tutel: Adaptive Mixture-of-Experts at Scale](https://mlsanity.com/paper/2206.03382/) (2022)
- [Accelerating Distributed MoE Training and Inference with Lina](https://mlsanity.com/paper/2210.17223/) (2022)
- [Pre-gated MoE: An Algorithm-System Co-Design for Fast and Scalable Mixture-of-Expert Inference](https://mlsanity.com/paper/2308.12066/) (2023)
- [MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs](https://mlsanity.com/paper/2411.11217/) (2024)
- [PUMA: A Programmable Ultra-efficient Memristor-based Accelerator for Machine Learning Inference](https://mlsanity.com/paper/1901.10351/) (2019)
- [NeuPIMs: NPU-PIM Heterogeneous Acceleration for Batched LLM Inferencing](https://mlsanity.com/paper/2403.00579/) (2024)
- [SiDA-MoE: Sparsity-Inspired Data-Aware Serving for Efficient and Scalable Large Mixture-of-Experts Models](https://mlsanity.com/paper/2310.18859/) (2023)
- [Duplex: A Device for Large Language Models with Mixture of Experts, Grouped Query Attention, and Continuous Batching](https://mlsanity.com/paper/2409.01141/) (2024)
- [PIM Is All You Need: A CXL-Enabled GPU-Free System for Large Language Model Inference](https://mlsanity.com/paper/2502.07578/) (2025)
- [PAPI: Exploiting Dynamic Parallelism in Large Language Model Decoding with a Processing-In-Memory-Enabled Computing System](https://mlsanity.com/paper/2502.15470/) (2025)
- [IANUS: Integrated Accelerator based on NPU-PIM Unified Memory System](https://mlsanity.com/paper/2410.15008/) (2024)
- [TIMELY: Pushing Data Movements and Interfaces in PIM Accelerators Towards Local and in Time Domain](https://mlsanity.com/paper/2005.01206/) (2020)
- [MoNDE: Mixture of Near-Data Experts for Large-Scale Sparse Models](https://mlsanity.com/paper/2405.18832/) (2024)
- [Patterns behind Chaos: Forecasting Data Movement for Efficient Large-Scale MoE LLM Inference](https://mlsanity.com/paper/2510.05497/) (2025)
- [CHIME: A Case for Efficient Long-Context Attention-FC Disaggregated Inference with DIMM-PIM](https://mlsanity.com/paper/2504.17584/) (2025)
- [Ouroboros: Wafer-Scale SRAM CIM with Token-Grained Pipelining for Large Language Model Inference](https://mlsanity.com/paper/2603.02737/) (2026)

## Abstract (arXiv)

Attention-FFN disaggregation maps LLM modules to specialized pools, creating an opening to keep Mixture-of-Experts (MoE) weights resident in a high-bandwidth FFN pool. Decode SLOs, however, cap the run-batch while sparse routing expands the activated-expert union, so weight traffic amortizes poorly and routing skew idles cold-expert resources. The FFN pool must therefore deliver weight-read bandwidth density under sparse unions and recover occupancy under skew without a global sharing fabric. We present a ReRAM near-memory architecture that keeps expert weights resident behind high-bandwidth local reads. The design factors actual MFU into ideal MFU and occupancy, recovers occupancy with bounded core-local multicast pooling, coactivation-aware placement, and load-aware fetch, and sizes each communication level from induced demand. A measured + modeled study on Qwen3.5-35B-A3B, Qwen3.5-397B-A17B, and GLM-5.2 shows that side-4 pooling raises occupancy from 0.328 to 0.519 and, at iso-peak compute, lowers per-token FFN-pool latency by 9.5x versus H20 with 20x lower weight-movement energy; an H20-attention + ReRAM-FFN system reduces decode TPOT by 1.25-4.0x, 2.4-10.3x, and 2.5-10.4x versus a homogeneous H20 pool.