Skip to content
HIRAK

Benchmark

Docker Build Performance Across Different CPUs

How much does the underlying CPU actually matter when building container images?

1 min read
  • Docker
  • Performance

The question

Building images feels CPU-bound until you add caching, at which point it stops being CPU-bound at all. I wanted to know where the boundary sits.

The benchmark varies three things and measures wall-clock build time:

  • Compute configuration (cores and clock speed)
  • Cache state (cold, warm layer cache, warm BuildKit cache)
  • Project size and dependency graph

What I expect to measure

Variables under test.

text
Compute
 ├── vCPU count
 └── clock speed
Cache
 ├── cold
 ├── layer cache
 └── BuildKit cache
Project
 ├── dependency count
 └── source size

The useful output of an experiment like this is not a single winner. It is knowing which lever to pull for a given project: more cores, better caching, or a smaller dependency graph.

This write-up is being expanded with the raw results table and methodology.

Related