El Capitan node architecture
El Capitan is operated by Lawrence Livermore National Laboratory in United States. This profile covers the documented compute inventory.
Cluster interconnect: HPE Slingshot 11. Topology: Dragonfly.
MI300A compute
- Node count
- 11520
- CPU
- AMD Zen 4 · 24 cores per APU
- CPUs per node
- 4
- Accelerator
- AMD MI300A
- Physical accelerators per node
- 4
- Host / unified memory
- 512 GiB unified HBM3
- Accelerator memory
- Shared with CPU
- Local accelerator links
- Infinity Fabric
- CPU-to-accelerator link
- On-package Infinity Fabric
- Storage
- Not verified
Scope and uncertainty
Four APUs per node: each combines CPU and GPU with 128 GiB shared memory. Installed node count differs from the benchmark configuration.
Not verified means a detail has not been confirmed in this catalog. Diagrams are schematic; node indices are illustrative, not physical hostnames. Data checked 13 September 2026.
Sources
Browse all cluster architecture pages →
CLUSTER