Warpbuild 4 vCPU x64 vs Blacksmith 4 vCPU x64
Blacksmith does the three builds for 1.2× less in total (less on the Rust and Docker builds and more on the TypeScript build); neither is clearly faster (within 5% or inside each other's run-to-run range). Both 4 vCPU x64.
All three builds 1.2× cheaper on Blacksmith: $0.0741 vs $0.0900Blacksmith 1.3× faster: 8m 27s vs 10m 40s
Rust build 1.2× cheaper on Blacksmith: $0.0155 vs $0.0193within run-to-run spread: 2m 18s · 1m 47s
TypeScript build 1.1× cheaper on Warpbuild: $0.0111 vs $0.0121within run-to-run spread: 1m 13s · 1m 19s
Docker build 1.3× cheaper on Blacksmith: $0.0465 vs $0.0596within run-to-run spread: 7m 09s · 5m 22s
Warpbuild 4 vCPU
jobs ran from WarpBuild (US) and 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 | Warpbuild | Blacksmith | Gap |
|---|---|---|---|
| Cost | |||
| Cost per Rust build | $0.0193 | $0.0155 | 1.2× Blacksmith |
| Cost per TypeScript build | $0.0111 | $0.0121 | 1.1× Warpbuild |
| Cost per Docker build | $0.0596 | $0.0465 | 1.3× Blacksmith |
| Price per minute | $0.0080 | $0.0080 | ≈ even |
| Real builds, wall time | |||
| Rust build (spotify-player) | 2m 18s | 1m 47s | ≈ even |
| TypeScript build (opencode) | 1m 13s | 1m 19s | ≈ even |
| Docker build (PostHog) | 7m 09s | 5m 22s | ≈ even |
| CPU & memory | |||
| PassMark single-thread (points) | 4,255 | 4,332 | ≈ even |
| PassMark CPU Mark, all cores (points) | 11,209 | 14,771 | 1.3× Blacksmith |
| sysbench CPU, 1 thread (events/s) | 6,260 | 6,493 | ≈ even |
| 7-Zip compression (MIPS) | 22,991 | 28,802 | 1.3× Blacksmith |
| Memory read bandwidth | 36.7 GiB/s | 41.7 GiB/s | 1.1× Blacksmith |
| Measurement | Warpbuild | Blacksmith | Gap |
|---|---|---|---|
| Workspace disk (fio) | |||
| Random read, 4k QD32 (IOPS) | 198,918 | 728,035 | 3.7× Blacksmith |
| Random write, 4k QD32 (IOPS) | 75,037 | 218,915 | 2.9× Blacksmith |
| Random read, 4k QD1 (IOPS) | 47,423 | 71,728 | 1.5× Blacksmith |
| QD1 read latency, p99 | 31 µs | 16 µs | 1.9× Blacksmith |
| Sequential read | 4.2 GiB/s | 7.0 GiB/s | 1.7× Blacksmith |
| Sequential write | 2.9 GiB/s | 1.7 GiB/s | 1.7× Warpbuild |
| Sync write, 4k + fdatasync (IOPS) | 2,989 | 17,925 | 6.0× Blacksmith |
| Network, cache & Docker | |||
| Download from GitHub | 100 MiB/s | 76 MiB/s | 1.3× Warpbuild |
| Cache restore, 4 GB (actions/cache) | 21s | 11s | 1.9× Blacksmith |
| Cache save, 4 GB (actions/cache) | 16s | 7s | 2.4× Blacksmith |
| docker pull | 9.7s | 7.9s | 1.2× Blacksmith |
| Database | |||
| pgbench read/write (TPS) | 4,122 | 12,408 | 3.0× Blacksmith |
| pgbench read-only (TPS) | 54,409 | 109,979 | 2.0× Blacksmith |
| Queue | |||
| Queue under burst load | 12s | 10s | 1.2× 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)