I'm a developer looking to upgrade my Dell Precision 3440 SFF workstation to run local LLMs and other AI applications, but I'm facing some challenges. The machine has a tiny case and a proprietary 200W power supply, which only supports 50W for the graphics card. I'm considering installing a new graphics card with 16GB or 20GB of VRAM, like the RTX 2000 or 4000 Ada, and possibly upgrading the power supply to either 260W or 500W. My main concern is whether these changes will be stable or if I'll just create an overheating problem. Given the strict space constraints, I'm also worried about thermal management and noise levels while running heavy AI models. I need some advice on the feasibility of these upgrades and any suggestions to maximize VRAM without compromising on stability.
1 Answer
If you can swing it, I'd recommend harvesting the core components (CPU, RAM, SSD) and getting a new case, motherboard, and power supply. This would give you a lot more flexibility with upgrades and greatly improve your thermal situation.
Related Questions
Lenovo Thinkpad Stuck In Update Loop Install FilterDriverU2_Reload