Edge Device Compute Sizing Calculator

Work out edge device compute sizing instantly with clear inputs, formula shown and shareable results.

About Edge Device Compute Sizing Calculator

Required compute is operations per inference times inference rate, divided by achievable utilisation — and that last term is the one people forget. Real accelerators sustain only 20-50% of peak on typical models because of memory bandwidth and layer shapes, so a 0.036 TOPS raw demand needs a 0.1 TOPS part.

Formula

Edge compute

raw TOPS = GFLOP per inference x inferences per second / 1000; required = raw / utilisation

Worked example

With the default values below, Edge Device Compute Sizing Calculator returns Compute required of 0.10. Change any input above and every figure here is recalculated in your browser.

Inputs used

Operations per inference: 2.4 GFLOP
Inferences per second: 15 inf/s
Achievable utilisation: 35 %
Efficiency: 4 TOPS/W

Results

Compute required: 0.10 TOPS
Raw compute demand: 0.0360 TOPS
Estimated power draw: 0.03 W
Sizing note: Achievable on an embedded NPU or high-end MCU

Step by step

1Values used
2Edge compute
3Compute required= 0.10 TOPS
4Raw compute demand= 0.0360 TOPS
5Estimated power draw= 0.03 W
6Sizing note= Achievable on an embedded NPU or high-end MCU

Assumptions and limits

The result is derived only from the 4 inputs listed above. Anything not among them is not part of the calculation.
In particular, this calculator does not take inflation, tax, fees or charges or compounding frequency into account.
Figures are rounded for display to 4 decimal places, so totals built from several rounded parts can differ slightly from the unrounded arithmetic.

How these calculators are built and tested is described on the Methodology page. If a figure looks wrong, please report it.

Frequently Asked Questions

Why is utilisation so low?

Peak TOPS assumes perfectly shaped dense matrix operations. Depthwise convolutions, small batches and activation memory traffic all leave the multipliers idle.

Does quantisation help?

Substantially. Int8 inference typically runs 2-4 times faster than float and uses a quarter of the memory bandwidth, often with negligible accuracy loss.