RunsOn r8a.xlarge · tmpfs vs Blacksmith 4 vCPU x64
RunsOn does the Rust and TypeScript builds for 2.5–2.7× less; the two take about the same time on the Rust and TypeScript builds. Both 4 vCPU x64.
Price per minute 3.8× cheaper on RunsOn: $0.0021 vs $0.0080the builds are not all compared on cost yet
Rust build 2.7× cheaper on RunsOn: $0.0065 vs $0.0175about the same time: 2m 06s · 1m 59s
TypeScript build 2.5× cheaper on RunsOn: $0.0036 vs $0.0089about the same time: 48s · 54s
Docker build 5m 19s Blacksmith only · RunsOn not run
RunsOn r8a.xlarge · tmpfs
jobs ran from Amazon.com (US)
Blacksmith 4 vCPU
jobs ran from Secured Servers (US)
Side by side
Every measurement both runners share, cost first. The dot marks the better value; gap is the larger over the smaller.
| Measurement | RunsOn | Blacksmith | Gap |
|---|---|---|---|
| Cost | |||
| Cost per Rust build | $0.0065 | $0.0175 | 2.7× RunsOn |
| Cost per TypeScript build | $0.0036 | $0.0089 | 2.5× RunsOn |
| Cost per Docker build | n/a: Not run: needs more than 32 GB of RAM with tmpfs, where Docker's data root lives in RAM; PostHog's layers plus the build's own memory exceeded it and the runner was shut down mid-build. The 64 GB tmpfs runner builds it. | $0.0464 | — |
| Price per minute | $0.0021 | $0.0080 | 3.8× RunsOn |
| Real builds, wall time | |||
| Rust build (spotify-player) | 2m 06s | 1m 59s | ≈ even |
| TypeScript build (opencode) | 48s | 54s | ≈ even |
| Docker build (PostHog) | n/a: Not run: needs more than 32 GB of RAM with tmpfs, where Docker's data root lives in RAM; PostHog's layers plus the build's own memory exceeded it and the runner was shut down mid-build. The 64 GB tmpfs runner builds it. | 5m 19s | — |
| CPU & memory | |||
| PassMark single-thread (points) | 3,799 | 4,236 | 1.1× Blacksmith |
| PassMark CPU Mark, all cores (points) | 13,460 | 15,036 | 1.1× Blacksmith |
| sysbench CPU, 1 thread (events/s) | 5,534 | 6,419 | 1.2× Blacksmith |
| 7-Zip compression (MIPS) | 28,331 | 27,863 | ≈ even |
| Memory read bandwidth | 36.5 GiB/s | 31.5 GiB/s | 1.2× RunsOn |
| Measurement | RunsOn | Blacksmith | Gap |
|---|---|---|---|
| Workspace disk (fio) | |||
| Random read, 4k QD32 (IOPS) | 4,014,510 | 873,112 | 4.6× RunsOn |
| Random write, 4k QD32 (IOPS) | 3,665,859 | 251,762 | 14.6× RunsOn |
| Random read, 4k QD1 (IOPS) | 1,054,041 | 84,854 | 12.4× RunsOn |
| QD1 read latency, p99 | <1 µs | 14 µs | 71.5× RunsOn |
| Sequential read | 18.2 GiB/s | 6.9 GiB/s | 2.6× RunsOn |
| Sequential write | 9.9 GiB/s | 1.6 GiB/s | 6.1× RunsOn |
| Sync write, 4k + fdatasync (IOPS) | 912,188 | 18,024 | 50.6× RunsOn |
| Network, cache & Docker | |||
| Download from GitHub | 330 MiB/s | 107 MiB/s | 3.1× RunsOn |
| Cache restore, 4 GB (actions/cache) | 8s | 14s | 1.8× RunsOn |
| Cache save, 4 GB (actions/cache) | 13s | 8s | 1.6× Blacksmith |
| docker pull | 5.1s | 6.5s | 1.3× RunsOn |
| Database | |||
| pgbench read/write (TPS) | 20,139 | 11,775 | 1.7× RunsOn |
| pgbench read-only (TPS) | 147,043 | 89,914 | 1.6× RunsOn |
| Queue | |||
| Queue under burst load | 17s | 11s | 1.5× Blacksmith |
- ≈ even: under 5% apart, or a build inside both runners' run-to-run spread
- Queue: identical jobs queued at the same instant (burst suite)
- 2 missing on one side or both
- RunsOn at EC2 spot, its default (7-day average of the cheapest us-east-1 zone's spot price, fetched 5 Oct 2026); the jobs ran on spot
- Cost: a managed SaaS runner against RunsOn self-hosted on your AWS account; the effort of running it was not measured