Skip to the content

TechnologyCompute & Inference

More intelligence from every watt

Every watt and every cycle should count. Our optimization and inference work asks one question: how can intelligence do more with less?

A cut crystal on a black granite base; inside it, a classical torso with the heart lit in blue.

Research

Related results

Technical note

One CPU core against a high-end GPU

METRON’s compiled laws on one CPU core serve about 27 times more requests per second than a neural network on an RTX 5090, at 100 % accuracy.

Read the technical note

Technical note

One million requests. Zero errors.

METRON answered 1,000,000 requests with 0 errors, at a median of 17.6 µs per request on one CPU core.

Read the technical note

Technical note

The same laws, from 224,000 times less data

On the same laws, METRON needs about 224,000 times less data and 13,000 times less time than a neural network of 871k parameters, and learns them exactly.

Read the technical note

Read more

Where to go next

Energy

Less energy, measured

The same task, on the same machine, against a neural network.

See the comparison

Product

METRON

AIDON’s exact co-processor.

Explore METRON

Research

Technical notes

One technical note for each result.

All 14 notes

Contact

Talk to the founders

Get in touch. We would be glad to share our ideas and walk you through the results, the proof and the measurements.