Hi everyone,

I’m preparing to move my "Green Crunching" setup to an 80W Intel Panther Lake Mini PC in early February. I’m specifically targeting MilkyWay@home to test the new Xe3 iGPU (Arc B390) on the 18A node.

As many here know, Intel iGPUs have been largely sidelined for this project due to a lack of native double-precision support. However, the Xe3 architecture restores hardware-native FP64 with a 1:16 ratio. This should allow these high-TDP iGPUs to finally contribute to N-Body simulations with the stability required for scientific validation.

I’ll be running 64GB of DDR5-9600 to provide a massive VRAM buffer. Once the unit arrives in Denmark, I will provide clinfo logs and TPF benchmarks. Is the project team open to whitelisting these new 18A-based PCI IDs for the OpenCL N-Body applications?

In case you didn't notice: Milkyway@home does not have applications for any GPU since years, i.e. since Separation ended. N-Body is CPU only as can be seen on the applications page.

---

At least can still crunch with tje cpu