CLUSTERATLAS
A field guide to supercomputing

Tuolumne

DOE/NNSA/LLNL

CPU Accelerator Interconnect
i
Reading this map

Tuolumne node architecture

Tuolumne is operated by DOE/NNSA/LLNL in United States. This profile covers Ranked system / partition. Specifications are listed below; anything the operator has not published is itemised in the documented-fields list.

Published benchmark (TOP500 · June 2026)

HPL rank
#14
HPL Rmax
208.1 PFlop/s
Theoretical peak Rpeak
288.88 PFlop/s
Ranked core count
1,161,216
Measured power
3387 kW
Energy efficiency
61.44 GFlop/s per watt

Cluster · MI300A compute

Operator / site
DOE/NNSA/LLNL
Country
United States
Profile scope
Ranked system / partition
Data checked
14 September 2026
Manufacturer
HPE
Ranked core count
1161216
Theoretical peak (PFlop/s)
288.88
Measured power (kW)
3387
Energy efficiency (GFlop/s per watt)
61.44080307056392
Operating system
TOSS
Network topology
Dragonfly

MI300A compute

System / node model
HPE Cray EX255a
Compute nodes of this type
1144
CPU model
AMD Zen 4 · 24 cores per MI300A
Accelerator model
AMD MI300A
Physical accelerators per node
4
Host / unified memory
512 GB unified HBM3 per node
Accelerator memory
128 GB shared CPU/GPU memory per APU
Accelerator ↔ accelerator link
Infinity Fabric
On-node fabric
AMD Infinity Fabric — each APU fully connected to its peers by two links at 256 GB/s
CPU ↔ accelerator link
Infinity Fabric
Network interfaces per node
4
Interface model
HPE Slingshot 11
Interface speed
200 Gb/s (25 GB/s) per interface · 100 GB/s node injection
Form factor
9 compute cabinets (HPE Cray EX4000 with EX255a blades)

Scope and uncertainty

1,144 compute nodes (1,100 batch + 44 debug); the published 1,152 total includes eight login nodes. CPU cores and GPU share memory within each MI300A; unified memory is not an additional separate RAM allocation.

Documented fields: 21 / 26

Every field this catalog checks. Fields listed above have a value; the remainder are shown here. A field marked absent is not publicly available in the sources we could verify, which does not mean the hardware is missing. Fields that cannot apply, such as benchmark results for an unranked system, are excluded from the total.

Cluster — 7 / 7

  • Manufacturer
  • Ranked core count
  • Theoretical peak (PFlop/s)
  • Measured power (kW)
  • Energy efficiency (GFlop/s per watt)
  • Operating system
  • Network topology

MI300A compute — 14 / 19

  • System / node model
  • Compute nodes of this type
  • CPU model
  • CPU sockets per node
  • Cores per socket
  • Physical CPU cores
  • CPU note
  • Accelerator model
  • Physical accelerators per node
  • Host / unified memory
  • Accelerator memory
  • Accelerator ↔ accelerator link
  • On-node fabric
  • CPU ↔ accelerator link
  • Network interfaces per node
  • Interface model
  • Interface speed
  • Interface-to-accelerator topology
  • Aggregate scale-out per node
  • GPUDirect RDMA
  • Management / storage networking
  • Local storage
  • NVMe data disks
  • Max remote data disks
  • Uncached disk IOPS
  • Uncached disk throughput (MBps)
  • Form factor

Sources

Browse all cluster architecture pages →

Data & sources