CXL memory pooling
CXL memory pooling allocates device-attached capacity among hosts while assigning each pooled memory region to one host at a time.
Pooling redistributes capacity; sharing permits concurrent access to the same region. In the CXL 2.0 pooling model, one device physical location maps to only one host's host-managed device memory (HDM). A multi-logical device can serve several hosts without giving them the same bytes. [1] [2]
Address interleaving is independent of pooling. Shared link bandwidth is also distinct from shared memory ownership: the fabric manager configures bandwidth allocation for logical devices. Neither a larger pool nor interleaving alone establishes a latency improvement. [1]
Relations
- related
- CXL memory expansion, CXL memory sharing, CXL Type-3 device, Host-managed device memory decoder, Pond, Memory stranding
Citations
- CXL Consortium: CXL 2.0 memory pooling Q&A, part 2 (21 May 2021); physical-location ownership and independent interleaving.
- CXL Consortium: Flash Memory Summit tutorial (August 2023), slides 27-28; pooled/shared FAM and two coherence models.
Sources & caveats
- Source synthesis of Consortium explainers, not a normative specification review, interoperability test or native CXL experiment.
- Status is unknown because these sources establish mechanisms, not current market availability. Synchronization and durability boundaries are engineering distinctions, not measured results.