Smart glasses promise hours of runtime. We measure the gap.
Time2Throttle (T2T) is an independent lab measuring where the heat and battery actually go in smart glasses and wearables - under real workloads, with published methodology, starting from devices bought at retail.
Build-in-public devlog + first benchmark results. No spam, no vendor money behind the scores.
The gap nobody measures
What spec sheets promise.
What devices actually deliver in battery life under real load-often achieved by secretly throttling performance just minutes in.
Independent labs publishing standardized thermal & battery benchmarks for this category. That's the gap.
Every claim about runtime and thermal comfort is currently graded by the vendor that makes it. Phones have independent reviewers. Cameras have DxOMark. TVs have RTINGS. Smart glasses - the most thermally constrained computers people wear on their faces - have marketing copy. We're building the instrument and the methodology to change that.
The T2T score
One number, honestly produced: how long a device sustains a real workload before it limits itself and why.
Time-to-Throttle (T2T)
Minutes from workload start to the first performance limit with the cause recorded: thermal or power. Heat and tired batteries both end the party; we log which one turned off the music.
Surface temperature
Measured at skin-contact points with external instruments. The 39 °C comfort ceiling is biology, not marketing.
Real runtime
100% to 10% under a standardized capture-encode-stream workload. Not idle-on-a-shelf numbers.
Why 10%? To prevent deep-discharge damage to retail units so they can be accurately re-tested.
Claim delta
Vendor-claimed vs measured. The headline number, every device, every test.
Full methodology — workloads, instrumentation, pass/fail thresholds — will be published openly and versioned before the first result. Reproducibility is the product.
Firewall rules
A benchmark is worth exactly as much as its independence. Ours is structural:
- 01Scores are never for sale — no vendor can pay to alter, publish, or remove a result.
- 02Test devices are bought at retail; vendor-supplied units are disclosed inline.
- 03Affiliate links, where used, are added after the verdict and never influence it.
- 04Raw telemetry is published with every result.
- 05Corrections are append-only — results are timestamped and hash-anchored so history can't be quietly edited.
Building the instrument in public
The profiler, the methodology, and the lab itself are being built in the open — starting with a 2015 Huawei phone from a drawer feeding the first telemetry pipeline, because the methodology has to be right before the hardware is expensive.
First results are coming
Get the devlog and the first benchmark tables as they're published. One email when it matters, silence when it doesn't.