Reddit users test whether unlocked CMP 170HX cards can power local AI rigs
Tritheone69 · reddit · 2026-07-28
A Reddit user says they bought a CMP 170HX 8GB card to take advantage of a recent unlock that exposes the full 64GB of VRAM for local AI workloads. They ask whether multiple cards can be run in one rig, how much x1 PCIe bandwidth hurts model loading, prompt processing, and token throughput, and whether prices will keep rising.
The post is a practical local-inference and GPU economics question centered on the CMP 170HX’s newly attractive AI use case.
More from Infra
- Kimi K3 runs on AMD MI350X with SGLang and hits 327 tok/s across four requests — burny_tech · 2026-07-28
- Kimi K3 reportedly keeps block attention residuals in a 93-layer design — burny_tech · 2026-07-28
- llama.cpp benchmarks show ROCm and Vulkan trading wins on AMD Radeon AI PRO R9700 — Gesha24 · 2026-07-28
- NASA’s new administrator says SpaceX-backed orbital data centers will happen — elonmusk · 2026-07-28
- Open-weight models are pitched as a security win for large companies — jessi_cata · 2026-07-28
- Dolphin is being trained on Trinity Large Thinking 398B across 72 RTX 4090s — QuixiAI · 2026-07-28