RESULTSWEBGL2 / SDF
Benchmark results
Interactive first-response artifacts, stored performance measurements, validation findings, generation metadata, and source. Select any two runs for a side-by-side comparison.
- Host input
HTMLCanvasElement- Submission
- First response · no repair
- Prompt version
- v1.0.0
RUNS—published records
LIMIT32 KiBsource module
VIEWPORT960 × 540reference size
PROMPTcd21d5a8…SHA-256
RUN CATALOG
Published runs
Raw first responses are retained. Performance values are comparable only when environment keys match; visual scores and requirement findings are reported separately.
No runs match this filter.
COMPARISON
Compare two runs
Orbit, zoom, and inspect both artifacts. Stored deltas appear only when their environments match.
SAME-DEVICE TEST
Comparison metrics
Run the serial local test to produce a fair same-device comparison. The previews pause while each artifact is measured alone.
MetricABDelta
Equivalent throughput higher is better———
Median GPU / frame lower is better———
p95 GPU / frame lower is better———
Source size lower is leaner———
Generation harness——metadata
Generation tokens provider-reported total——metadata
Generation cost actual API charge or API-equivalent estimate——metadata
Requirements unmet named findings, not an overall score——reported separately
METHOD
Benchmark methodology
The host supplies a blank canvas and no scene helpers. The submitted module owns geometry, shaders, camera state, rendering, and interaction. The canonical prompt explicitly states that frame rate will be measured.
- 01Generate
Fresh conversation, byte-exact prompt, first response only, no repair pass.
- 02Validate
Compile and render checks, twelve camera views, close zoom, and required input paths.
- 03Measure
GPU timer queries where available; synchronized wall-clock timing otherwise.
Loading the canonical prompt…