Alibaba's RISC-V XuanTie C950 runs Qwen 3.8B at 30 tokens/s
DeltaSqueezer · reddit · 2026-08-19
Alibaba's T-Head RISC-V processor XuanTie C950 runs the Qwen 3.8B model at 30 tokens/s, as discussed on Reddit. The poster quips "who needs GPUs?", showing that mid-size LLM inference on RISC-V without a GPU is becoming practical — a notable data point for edge/low-power inference.
Related event: Alibaba's Xuantie C950 RISC-V CPU Runs Qwen 3.8B at 30 Tokens/s(2 posts)→
More from Infra
- Phison CEO calls CXMT memory scare a 'ghost story', warns of materials shortage — BenBajarin · 2026-10-16
- Ben Bajarin: The Future of Compute Is Fungible as Hyperscalers Design for Workload Flexibility — BenBajarin · 2026-10-16
- AirLLM runs frontier-size models layer-by-layer: 671B on 12GB VRAM — Aiden_Tech_Ai · 2026-10-16
- NVIDIA Launches cuPotential in CUDA-X to Bring AI Agents to Semiconductor Materials Simulation — tomaszbednarz · 2026-10-16
- Q4-quantized coder model on 8-core CPU builds interactive solar system demo in one shot — IntrepidMindExplorer · 2026-10-16
- TokenSpeed MLA 2 ships for Vera Rubin: up to 3.44× faster Kimi K3 decode — zhyncs42 · 2026-10-16