ScalePool: Hybrid XLink-CXL Fabric for Composable Resource Disaggregation in Unified Scale-up Domains
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912652289114112 |
|---|---|
| author | Woo, Hyein Kwon, Miryeong Kim, Jiseon Na, Eunjee Choi, Hanjin Jang, Seonghyeon Jung, Myoungsoo |
| author_facet | Woo, Hyein Kwon, Miryeong Kim, Jiseon Na, Eunjee Choi, Hanjin Jang, Seonghyeon Jung, Myoungsoo |
| contents | This paper proposes ScalePool, a novel cluster architecture designed to interconnect numerous accelerators using unified hardware interconnects rather than traditional long-distance networking. ScalePool integrates Accelerator-Centric Links (XLink) and Compute Express Link (CXL) into a unified XLink-CXL hybrid fabric. Specifically, ScalePool employs XLink for intra-cluster, low-latency accelerator communication, while using hierarchical CXL-based switching fabrics for scalable and coherent inter-cluster memory sharing. By abstracting interfaces through CXL, ScalePool structurally resolves interoperability constraints, enabling heterogeneous cluster operation and composable resource disaggregation. In addition, ScalePool introduces explicit memory tiering: the latency-critical tier-1 combines accelerator-local memory with coherence-centric CXL and XLink, whereas the highcapacity tier-2 employs dedicated memory nodes interconnected by a CXL-based fabric, achieving scalable and efficient memory pooling. Evaluation results show that ScalePool accelerates LLM training by 1.22x on average and up to 1.84x compared to conventional RDMA-based environments. Furthermore, the proposed tier-2 memory disaggregation strategy reduces latency by up to 4.5x for memory-intensive workloads. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_14580 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ScalePool: Hybrid XLink-CXL Fabric for Composable Resource Disaggregation in Unified Scale-up Domains Woo, Hyein Kwon, Miryeong Kim, Jiseon Na, Eunjee Choi, Hanjin Jang, Seonghyeon Jung, Myoungsoo Distributed, Parallel, and Cluster Computing This paper proposes ScalePool, a novel cluster architecture designed to interconnect numerous accelerators using unified hardware interconnects rather than traditional long-distance networking. ScalePool integrates Accelerator-Centric Links (XLink) and Compute Express Link (CXL) into a unified XLink-CXL hybrid fabric. Specifically, ScalePool employs XLink for intra-cluster, low-latency accelerator communication, while using hierarchical CXL-based switching fabrics for scalable and coherent inter-cluster memory sharing. By abstracting interfaces through CXL, ScalePool structurally resolves interoperability constraints, enabling heterogeneous cluster operation and composable resource disaggregation. In addition, ScalePool introduces explicit memory tiering: the latency-critical tier-1 combines accelerator-local memory with coherence-centric CXL and XLink, whereas the highcapacity tier-2 employs dedicated memory nodes interconnected by a CXL-based fabric, achieving scalable and efficient memory pooling. Evaluation results show that ScalePool accelerates LLM training by 1.22x on average and up to 1.84x compared to conventional RDMA-based environments. Furthermore, the proposed tier-2 memory disaggregation strategy reduces latency by up to 4.5x for memory-intensive workloads. |
| title | ScalePool: Hybrid XLink-CXL Fabric for Composable Resource Disaggregation in Unified Scale-up Domains |
| topic | Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2510.14580 |