NumberFields@home has done an incredible job supporting NVIDIA, AMD, and even the newer Intel Arc GPUs.

However, there is a massive "Green" elephant in the room: Apple Silicon.

For years, the exclusion of the Mac iGPU was blamed on a lack of native FP64 or 64-bit integer efficiency.

But as of 2026, the success of Einstein@Home and PrimeGrid has completely debunked this.

If PrimeGrid can run complex number theory searches on the M-series iGPU with 100% stability, there is no technical reason NumberFields cannot do the same.

On my M4 Mac Mini, I am currently running 10 concurrent tasks on the iGPU for other projects.

The unified memory architecture provides a massive bandwidth advantage that traditionally bottlenecked GPUs struggle with.

By ignoring Apple Silicon, NumberFields is missing out on thousands of high-efficiency nodes that could significantly speed up our search for decic fields.

We have the hardware. We have the proof of concept from other projects.

It’s time to move past the "Metal vs. OpenCL" legacy excuses and bring NumberFields into the modern Apple Silicon era.

When can we expect an experimental arm64-apple-darwin GPU plan?

When can we expect an experimental arm64-apple-darwin GPU plan?

When I find more time?

For anyone who wants to take a stab at it, here is the source code on github:
https://github.com/drivere/get-decics-numberfields
It just needs to be ported...