Efficiency & FP64: Preparing for Intel Panther Lake (Xe3) iGPU crunching


Message boards : Number crunching : Efficiency & FP64: Preparing for Intel Panther Lake (Xe3) iGPU crunching

Message board moderation

To post messages, you must log in.
AuthorMessage
kasdashdfjsah

Send message
Joined: 3 Feb 24
Posts: 44
Credit: 110,841
RAC: 4
Message 9376 - Posted: 16 Jan 2026, 17:17:52 UTC
I’m planning a "Green Crunching" upgrade to replace my current Ryzen 3700/RTX 3060 desktop with a high-TDP (80W) Intel Panther Lake Mini PC once they ship in early February.

I'm particularly interested in the Asteroids@Home performance on the new Xe3 iGPU (Arc B390). Given that this architecture finally re-introduces hardware-native FP64 (1:16 ratio) and is built on the 18A node, the science-per-watt should be a significant leap over previous Intel generations.

I’ll be running a 64GB DDR5 setup to ensure the iGPU has plenty of shared memory for large work units. Has anyone else looked into the OpenCL 3.0 performance for this architecture yet? I’m happy to post logs and TPF (Time Per Frame) data once the unit arrives to help optimize the app for these new chips.
ID: 9376 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : Efficiency & FP64: Preparing for Intel Panther Lake (Xe3) iGPU crunching