Free tool · Edge / GPU / Container
Edge AI, plant GPU or container DC — size the tier before you buy silicon.
Enter cameras, resolution, frame rate and model class. Get an indicative compute tier, ingest and storage sketch — runs in your browser, no upload.
Inputs
Your cameras, FPS, model class and tier recommendation are attached to the assessment form automatically.
FAQ
Compute sizer questions.
How accurate is the edge / GPU / container recommendation?
It is an order-of-magnitude load index from camera count, resolution, FPS and model class. Real BOMs also consider model size, batching, preprocessing, spare policy and power/cooling.
What does the load index mean?
It is a weighted gigapixels-per-second style score used only for tiering (edge vs plant GPU vs container). It is not a vendor FLOPS rating.
Why does SAS often recommend hybrid architectures?
Millisecond rejects belong at the edge. Multi-line or heavy models usually share plant GPU or container capacity. Hybrid keeps OT reliable without overbuying silicon on day one.
Does on-prem / air-gap change the answer?
Sovereignty does not remove the need for local compute — it usually strengthens the case for edge and plant/container hosts instead of public cloud inference.
Next step
Need a BOM, not a guess?
Share camera count and KPIs. We return a tiered architecture with edge vs plant vs container recommendations for your site.