# BuildPulse Runner Benchmarks

CoreMark, one execution thread, 2000-byte dataset. Higher iterations/s is better.

Linux/Windows require hard affinity to one logical CPU. macOS is explicitly **unpinned**: its one thread may migrate between CPUs. These are runner + compiler measurements; VM host scheduling, other workloads, turbo, and image updates can affect them.

## Benchmark parameters

Compare providers using the same parameter fingerprint. Runner selection and provider identity are excluded; measurement protocol is included.

| Profile | Protocol | CPU/disk samples | CPU seconds | Disk MiB | Cache MiB | Internet | Transfer MiB |
| --- | --- | ---: | ---: | ---: | ---: | --- | ---: |
| `e021cd7ddc71` | 6 | 5 | 30 | 1024 | 512 | Enabled | 1024 |

## Pinned results

| Runner | Binary | Median iterations/s | Min–max | Variation | CPU / group | Samples |
| --- | --- | ---: | ---: | ---: | --- | ---: |
| buildpulse-linux-x64 | x64 (64-bit) | 43146.27 | 42918.46–43242.82 | 0.28% | 0 / 0 | 5 |
| buildpulse-linux-arm64 | arm64 (64-bit) | 33115.74 | 33075.97–33129.44 | 0.06% | 0 / 0 | 5 |

Variation is the population coefficient of variation across accepted samples.

## Disk I/O

One worker, queue depth one, 1 MiB sequential I/O, 4 KiB random I/O. OS file-cache bypass is required; write timings include the final flush. Protocol 2 and later repeat each operation for at least 5 seconds against the same file; total transferred bytes can exceed the file size. Medians across fully verified samples. Every requested sample is attempted, including after a 60-second sample timeout. Incomplete results remain visible and are excluded from complete-sample comparisons.

| Runner | Affinity | File MiB | Completed / attempted / requested | Status | Read MiB/s | Write MiB/s | Random read IOPS | Random write IOPS |
| --- | --- | ---: | --- | --- | ---: | ---: | ---: | ---: |
| buildpulse-linux-arm64 | Pinned | 1024 | 1 / 5 / 5 | incomplete | 73.82 | 142.38 | 2146.79 | 1165.69 |
| buildpulse-linux-x64 | Pinned | 1024 | 5 / 5 / 5 | ok | 125.90 | 133.56 | 2132.28 | 1161.92 |

### Disk observations including interrupted samples

These medians include completed phases and acknowledged work from interrupted phases. `≥` marks a conservative lower bound: the last progress count divided by elapsed time through termination, including stalls. An interrupted write may be unflushed, and timed-out samples lack full-file write integrity verification. Operations never reached have no measurement; they are not assigned zero speed. Use the verified table above for complete-sample comparisons.

| Runner | Read MiB/s (observations) | Write MiB/s (observations) | Random read IOPS (observations) | Random write IOPS (observations) |
| --- | --- | --- | --- | --- |
| buildpulse-linux-arm64 | ≥19.75 (5/5) | 142.38 (5/5) | ≥2145.77 (3/5) | ≥1064.46 (2/5) |

## Cache backends

Protocol 2 and later measure three unique-key save/restore pairs per backend. Each trial restores in a separate runner job after the save job completes; reused save OS instances are rejected. Backend order reverses on trial two. Archive services use zstd `--single-thread`; attached-volume transfers use one thread without compression. Linux/Windows action processes inherit one-CPU affinity; macOS is unpinned. Identical random payload bytes across trials and backends on each runner; an empty local target before restore, exact cache hit, SHA-256 and backend routing checked. Each time includes the action and adjacent step transitions; MiB/s uses logical payload bytes, including compression/extraction for archive services.

Profile `e021cd7ddc71` waits 120 seconds after the save job finishes, before allocating its first restore runner. All providers use this publication gap; it is excluded from transfer speeds. It gives cache publication time to settle but does not guarantee cache availability. Misses remain failures; there are no retries or payload fallbacks.


### archive-service

| Runner | Cache backend | Payload MiB | Verified pairs | Save s (median; range) | Save MiB/s | Save variation | Restore s (median; range) | Restore MiB/s | Restore variation |
| --- | --- | ---: | --- | --- | ---: | --- | --- | ---: | --- |
| buildpulse-linux-arm64 | BuildPulse artifact cache | 512.0 | 3/3 | 15.89; 12.58–16.87 | 32.21 | 12.17% | 17.94; 16.55–19.40 | 28.55 | 6.49% |
| buildpulse-linux-arm64 | GitHub Actions cache | 512.0 | 3/3 | 8.80; 7.65–25.36 | 58.17 | 58.05% | 10.92; 8.63–12.26 | 46.91 | 14.14% |
| buildpulse-linux-x64 | BuildPulse artifact cache | 512.0 | 3/3 | 15.20; 11.98–18.26 | 33.69 | 16.95% | 17.49; 16.36–25.63 | 29.27 | 20.83% |
| buildpulse-linux-x64 | GitHub Actions cache | 512.0 | 3/3 | 6.98; 6.57–23.25 | 73.40 | 63.33% | 11.18; 9.31–11.63 | 45.81 | 9.38% |

## Internet

One HTTP/1.1 stream at a time, three downloads and three uploads. Each sample transfers the configured payload through sequential requests of at most 32 MiB on one persistent connection. Mbps uses decimal megabits. Zero-byte HTTPS request time includes connection/TLS/DNS overhead and is not ICMP RTT. Compare only results using the same endpoint and payload size. Unavailable tests have no speed score.

- **LibreSpeed (Sharktech, Amsterdam)**: https://amsspeed.sharktech.net/backend/

| Runner | Service | Status | Transfer MiB | Download Mbps | Upload Mbps | HTTPS request ms |
| --- | --- | --- | ---: | ---: | ---: | ---: |
| buildpulse-linux-arm64 | LibreSpeed (Sharktech, Amsterdam) | unavailable | 1024 | — | — | — |
| buildpulse-linux-x64 | LibreSpeed (Sharktech, Amsterdam) | unavailable | 1024 | — | — | — |

## Optional stages and notices

- **buildpulse-linux-arm64 / disk**: Incomplete: 1/5 samples completed; 5 attempted; 4 exceeded 60s. Progress retained.
- **buildpulse-linux-arm64 / internet**: HTTPS upload sample 1 exceeded 60s; curl reported 805306368 bytes; the test did not complete, so no speed score is reported
- **buildpulse-linux-arm64 / buildpulse / trial 1 / cleanup**: The provider-managed cache entry is left to provider retention. Automatic native deletion requires a separately authenticated provider cleanup adapter; no deletion of unrelated entries was requested.
- **buildpulse-linux-arm64 / buildpulse / trial 2 / cleanup**: The provider-managed cache entry is left to provider retention. Automatic native deletion requires a separately authenticated provider cleanup adapter; no deletion of unrelated entries was requested.
- **buildpulse-linux-arm64 / buildpulse / trial 3 / cleanup**: The provider-managed cache entry is left to provider retention. Automatic native deletion requires a separately authenticated provider cleanup adapter; no deletion of unrelated entries was requested.
- **buildpulse-linux-x64 / internet**: HTTPS upload sample 1 exceeded 60s; curl reported 704643072 bytes; the test did not complete, so no speed score is reported
- **buildpulse-linux-x64 / buildpulse / trial 1 / cleanup**: The provider-managed cache entry is left to provider retention. Automatic native deletion requires a separately authenticated provider cleanup adapter; no deletion of unrelated entries was requested.
- **buildpulse-linux-x64 / buildpulse / trial 2 / cleanup**: The provider-managed cache entry is left to provider retention. Automatic native deletion requires a separately authenticated provider cleanup adapter; no deletion of unrelated entries was requested.
- **buildpulse-linux-x64 / buildpulse / trial 3 / cleanup**: The provider-managed cache entry is left to provider retention. Automatic native deletion requires a separately authenticated provider cleanup adapter; no deletion of unrelated entries was requested.

## Machine and build details

| Runner | CPU model | OS image | Compiler | Flags |
| --- | --- | --- | --- | --- |
| buildpulse-linux-arm64 | aarch64 | Linux-6.8.0-1066-aws-aarch64-with-glibc2.39 | GNU 13.3.0 | -O2 -DNDEBUG; C99; MULTITHREAD=1; TOTAL_DATA_SIZE=2000; ITERATIONS=0 |
| buildpulse-linux-x64 | AMD EPYC 9R45 | Linux-6.8.0-1066-aws-x86_64-with-glibc2.39 | GNU 13.3.0 | -O2 -DNDEBUG; C99; MULTITHREAD=1; TOTAL_DATA_SIZE=2000; ITERATIONS=0 |