Skip to benchmark
spacetime / benchmarks

Spacetime / Benchmarks

Measured.
Reproducible.
Inspectable.

Measure the environment we deliver, the queries it supports, and what happens when evidence fails.

Inspect the Ojai benchmark

OJAI-COMP-001 / Retained terrain

Ojai → CA-33

One place. One frozen input. A result you can reproduce.

Measuredspacetime.benchmark.v1

Real-data local benchmark; injected faults tested separately

Retained USGS terrain across Ojai and the CA-33 study area. The rendered surface shows the exported elevation grid used by this benchmark.
USGS 3D Elevation Program · 512 × 512 exported cells15.447 m export spacing · EPSG:26911

Compiled terrain pack

1.0008MiB
1,049,378 bytes: explicit metadata and 262,144 Float32 heights.

Local lookup p95

1.356µs
10,000 warm in-memory bilinear lookups; per-call timer included. Not end-to-end SDK, network, browser or device latency.

Finite source cells

100%
262,144 of 262,144 retained cells finite. This is grid completeness, not mission or sensor coverage.

Serialization difference

0m
Maximum absolute difference across all retained Float32 cells after pack decode. This is exact serialization, not ground accuracy.

A real-data software benchmark. Independent survey accuracy and field navigation remain unmeasured.

Download reproduction bundle Inspect result JSON ↗Query a retained terrain sample ↗Read the protocol ↓

Representation / Trade-off

Fidelity versus size.

Coarser grids use fewer height samples. How much terrain detail does that remove?

A 128 × 128 grid retains 6.25% of the source height samples, with 3.94 m RMS deviation after reconstruction.

Compared with the retained 512 × 512 raster, not surveyed ground truth. These coarser grids are presentation representations.

Inspect fidelity results ↗
Deviation after bilinear reconstruction to the retained grid
GridFewer samplesRMS
m		Absolute p95
m		Absolute max
m
256 × 25675%1.633.6616.24
128 × 12893.75%3.948.9233.67
64 × 6498.4375%8.7819.8654.10
32 × 3299.6094%18.0940.94105.44

Method / Evidence

Follow the result
back to its inputs.

Keep the task, data, runtime and timing boundary attached to every measurement.

Measurements and their scope
MeasurementResultWhat it measures
Compiled terrain pack1.0008 MiB1,049,378 bytes: explicit metadata and 262,144 Float32 heights.
Local lookup p951.356 µs10,000 warm in-memory bilinear lookups; per-call timer included. Not end-to-end SDK, network, browser or device latency.
Finite source cells100 %262,144 of 262,144 retained cells finite. This is grid completeness, not mission or sensor coverage.
Serialization difference0 mMaximum absolute difference across all retained Float32 cells after pack decode. This is exact serialization, not ground accuracy.
Data and runtime
Coordinate reference
EPSG:26911
Vertical reference
NAVD88
Grid
512 × 512 · 15.447 m export spacing
runtime
v24.19.0
platform
linux x64
os
Linux 6.18.55
cpu
INTEL(R) XEON(R) PLATINUM 8573C
logical cpu count
9
process
single Node.js process; warm in-memory numeric lookup; no browser/GPU/network

Export spacing describes this retained grid. It is not a claim about native source resolution or spatial accuracy.

Reproduce this run

Extract the reproduction bundle, then run from its root:

node scripts/benchmarks/run.mjs --output /tmp/spacetime-ojai-repro

The command uses retained inputs. Review the source manifest and result definitions before comparing a new run.

source commit
Not supplied
source commit meaning
Exact source file SHA-256 digests are recorded below; no fabricated commit identity.
script hashes
{"runtime.mjs":"0a4a43fbe6576aeda0edf1065e8deea8991536d494ee41e14ddb9881dbeded3c","run.mjs":"7f9466e13bc5c23f7c44791b08b64373336f4a7e559994dba7e021390fa106eb"}
seed
LCG seed20261020; state=(1664525*state+1013904223) mod2^32; fractional pixel-center positions in retained grid
sample count
10,000
warmup count
1,000
encode measured runs
10
analytical oracle
2x2 planar grid: midpoint15m and endpoint0/30m recovered exactly; outside samples unknown
receipt policy
{"frame":"EPSG:26911","verticalDatum":"NAVD88","evaluationTime":"2026-10-20T19:27:37.644758+00:00","maxReceiptAgeDays":30,"requireCaptureDate":false}
pack sha256
34ca012965ce7688eaa59eb8f3278e8f61b6f57bb0faec05a207e2990e11c100
network calls during replay
0
offline replay samples
100
offline method
Same runtime evaluated in isolated JavaScript realm without fetch, XMLHttpRequest, WebSocket, require or imports; 100 answers equal ordinary execution. Not an OS-level network-disabled device test.
Failure and policy checks

Injected faults and stricter policies exercise specified software behavior. The unknown-acquisition-date case uses the actual source limitation. These checks do not estimate field reliability.

  • Corrupted bytesblocked

    Injected bit flip fails expected SHA-256 before sampling.

  • Wrong horizontal frameblocked

    Injected incompatible frame is refused; runtime does not guess a projection.

  • Wrong vertical datumblocked

    Injected unresolved height reference is refused.

  • Stale receiptblocked

    Injected stale retrieval receipt fails the declared 30-day test policy; this is not a source-acquisition freshness claim.

  • Unknown acquisition dateblocked

    Actual mixed-mosaic acquisition date is unknown; strict known-capture-date policy refuses it.

  • Incorrect payload identityblocked

    Injected lineage mismatch is refused even with a recomputed outer digest.

  • Outside coverageblocked

    Out-of-extent query returns unknown; no extrapolation.

  • Invalid coordinateblocked

    Nonfinite coordinate returns unknown.

  • Missing terrain cellblocked

    Injected missing cell returns unknown; validity is preserved.

Scope and remaining evidence
  • No independent surveyed control or field navigation validation. Absolute terrain accuracy and positioning integrity are not established.
  • The export samples a mixed-resolution USGS mosaic at 15.447 m spacing; spacing is not native source resolution or accuracy.
  • Acquisition date is not established per pixel. Retrieval time is not capture time; terrain is not current-conditions evidence.
  • EPQS comparisons use the same USGS source family and compare sampling behavior, not independent accuracy.
  • Injected faults test declared software behavior; they do not establish real-world detection rates or an integrity guarantee.
  • Timing measures warm local numeric lookups and in-memory encoding on the recorded host; it excludes download, reprojection, source decoding and real-world data admission.
  • Offline evidence is capability-isolated local replay, not a disconnected physical-device or navigation test.
Retained source and artifacts

The benchmark program

Test the conditions
that matter.

01 / Compile

What does the environment cost to produce?

Measure retained bytes, compilation work, data fidelity and provenance.

Local terrain baseline published above
02 / Query

How does the artifact answer?

State the exact lookup, runtime and warm-up. Separate local computation from network and application latency.

Local lookup baseline published above
03 / Degrade

What happens when evidence fails?

Inject known faults and record whether the software returns a result, blocks or recovers.

Specified injected cases published above
04 / Operate

Does it improve the mission?

Evaluate localization, uncertainty and decisions against representative operating conditions and independent truth.

Field validation not yet measured

Build with Spacetime

Define the operating envelope.
Agree on the proof.

Start with the environment, platform and decision. Set the acceptance criteria before measuring.

Discuss a benchmark ↗