WORLD GPU CENTERCOMPUTE · CONTEXT · CLARITY
Home / DECISION LAB

DECISION LAB

Capacity & Energy

Translate scheduled devices and operating hours into GPU-hours and an explicit facility-energy estimate.

Your assumptions

Illustrative defaults. Currency inputs use US dollars.

Accounting share of allocated time, not hardware utilisation.
Average operating draw; not the nameplate TDP.
YOUR SCENARIO

Keep the boundary visible.

Ready to calculate

JavaScript is required for local calculations.

No information is sent.

Formula and units

GPU-h = devices × days × scheduled hours/day. Productive-equivalent GPU-h = GPU-h × productive share. Facility kWh = GPU-h × (average GPU kW + host kW/GPU) × PUE. Energy-only cost = facility kWh × US$/kWh.

How to read the result

The model treats the entered average draw as applying during all scheduled hours. Productive share does not reduce that assumed draw. Off-schedule idle energy, embodied impacts, storage/network power beyond the host allocation and peak design are excluded. The result is an allocation, not a metered bill or performance forecast.

Show the method

GPU-h = devices × days × scheduled hours/day. Productive-equivalent GPU-h = GPU-h × productive share. Facility kWh = GPU-h × (average GPU kW + host kW/GPU) × PUE. Energy-only cost = facility kWh × US$/kWh.

Recalculate after changing inputs. The download records inputs and results together. There is no live data connection and no automatic persistence. Reloading may discard your inputs.

Read the supporting analysis

Turn the GPU question
into a clearer brief.

Open the Decision Lab