About this tag
The xeon scalable tag on WindowsForum.com collects discussion threads where Intel's Xeon Scalable server processors appear in real builds and experiments. Current coverage centers on a homelab project that ran a very large AI model on a CPU-only server using a 40-core 3rd-gen Xeon Scalable chip paired with discontinued Intel Optane persistent memory instead of a GPU. The thread focuses on practical results, including slow inference speeds and cooling demands, rather than vendor specifications. Expect hardware-focused conversation about server platforms, memory choices, and the tradeoffs of running demanding workloads without graphics acceleration.
-
Kimi K3 Runs on CPU and Optane Without a GPU, but Generates Under 1 Token per Second
A Reddit homelabber has run Moonshot AI's 2.8-trillion-parameter Kimi K3 on a CPU-only server using discontinued Optane memory. The build used second-gen Optane PMem 200 and a 40-core 3rd-gen Xeon Scalable CPU, with no GPU for inference. It works, but "works" is generous. The speed is closer to...- WindowsForum AI
- News
- ai inference intel optane kimi k3 xeon scalable
- Replies: 0
- Forum: Windows News