What does HPL measure?
Why D is the answer
Sustained floating-point performance on a dense linear system
High-Performance Linpack solves a dense system of linear equations.
In detail
- HPL measures sustained floating-point performance on a dense system of equations.
- It solves a dense linear system.
Use case
An illustrative scenario, not a real incident.
A team runs HPL to measure sustained floating-point performance across a cluster.
See it in different ways
flowchart TD A["Single-node stress"] --> B["Single-node NCCL"] B --> C["Single-node HPL"] C --> D["Multi-node NCCL"] D --> E["Multi-node HPL and burn-in"]
mindmap
root((Verification))
Compute
HPL
ClusterKit
Fabric
NCCL
Signal quality
Versions
Switches
BlueField
Transceivers
Storage
fioflowchart LR S0["Set up"] --> S1["Solve dense system"] --> S2["Compare with expected"]
Key point: HPL measures sustained compute.
Why the other options are wrong
- A. The latency of a single RDMA write between two hosts
- That is ib_write_lat.
- B. The sequential read speed of one storage volume
- That is a storage test.
- C. The tokens per second of a language model
- That is an LLM measure.