[PLACEHOLDER — This case study is ready for content. Fill each section below with real, measured results from your published work. Do not invent numbers.]
The Question
[What did the client or project need to know? State it as a single, specific question. Example: "Can YOLOv8-nano run at 30fps on Jetson Orin NX 8GB at INT8 precision while staying under 8W sustained power draw?"]
The Constraint
[Describe the hardware platform, power budget, thermal envelope, latency target, accuracy floor, and connectivity requirement. Be specific — this is where technical buyers recognize their own situation.]
What We Measured and How
[Describe the measurement protocol: clock settings, warm-up procedure, number of runs, metrics collected (p50/p95/p99/max), thermal guard conditions, energy measurement method. A technical buyer should be able to reproduce this.]
The Number
[MEASURED RESULT — Fill from published work. State the key finding as a single, specific, reproducible number. Example: "p99 latency: 34.7ms. Mean power: 7.2W sustained. Thermal throttle onset: 68°C after 4 minutes of sustained load."]
What It Changed
[What decision did this evidence enable? Did it confirm viability, rule out a hardware choice, select a quantization strategy, or change the architecture? State the outcome plainly.]
What We Would Still Want to Test
[Honest limitations: what the measurement does not cover, what would change the result, what the next experiment would be. This section is what separates a measurement from a claim.]