x64arm64 at 2 vCPU4 vCPU8 vCPU
docker pullsecondslower is betterarm64 · 2 vCPU
RunsOn tmpfs pulls an image fastest.
RunsOn tmpfs to GitHub: 4.2× apart.
- RunsOntmpfs r9g.large · 16 GB 5.0s
- NamespaceApple silicon 2x8 · 8 GB 6.3s
- RunsOn median of 3 machines 6.4s
- Avrea 8 GB 7.1s
- Warpbuild 8 GB 10.1s
- Blacksmith 6 GB 13.7s
- Namespace 2x8 · 8 GB 17.0s
- Ubicloud standard · 8 GB 17.6s
- GitHub 8 GB 21.0s
- AWS CodeBuild no arm64 runner
- boat no arm64 runner
- WarpbuildApple silicon no 2 vCPU arm64 runner
The same image pulled on every runner, registry to disk: the download, then the layers unpacked, so the disk and the CPU count too.