TensorFold. NVIDIAInceptionmember

Support the project

Help us test on
more hardware.

TensorFold gets better when we can run it on the machines people actually use.

Your support helps us buy or access hardware, reproduce issues and test changes locally.

More machines. Better evidence.

Local testing needs
real hardware.

Access to different systems lets us check model fit, measure performance and reproduce problems where they happen.

Apple Silicon

Check Metal execution, memory limits and model behaviour across Mac generations.

NVIDIA GPUs

Test CUDA paths on different GPU architectures and memory configurations.

Connected machines

Validate model splits, direct transfers and concurrent runs on physical setups.

The hardware wishlist

The test setup
we want to build.

These are our target configurations, with availability and suitability to be confirmed before a purchase or loan.

10 requested systems6 configurations

Hardware donations, loans, test access and funding introductions can all help us work towards this setup.

Ways to help

Have hardware,
time or a connection?

There are several ways to move the project forward.

Lend hardware or share test access

A machine loan, hardware donation or access to a system can help us investigate a specific model and platform.

Offer hardware or access

Help us find hardware funding

Introduce us to a potential sponsor, grant programme or funding partner who could help us buy or access test hardware.

Make an introduction

Follow the work

See what we're testing.

Explore published changes, qualified hardware and recorded runs.

Release notes Benchmarks

Know someone who could help?

Send them this page.