I'm a software developer looking to dive into CUDA and high-performance computing. This PC will primarily focus on coding, but I'll also play some RPG and single-player games occasionally. I'm not planning to run large LLMs locally, so that's not a concern. Here are the components I'm considering: 1. Ryzen 7500F (~110€) 2. B850M motherboard (~120€) 3. RTX 5060 (or better, if the budget allows it) (~280€) 4. 32GB 2x16 DDR5 RAM (~430€) 5. 750W 80+ power supply 6. 1TB M.2 drive (~120€) The total is surpassing my budget, costing ~1400€ for a custom build or ~1250€ for parts. I'm wondering: 1. Are there alternative components I can use that would allow for easier future upgrades? 2. Can I downgrade any components without negatively affecting my primary use? This is my first time building a PC, so any help would be greatly appreciated!
1 Answer
I recommend looking for at least an RTX 5070Ti or 5080 to ensure you have 16GB of VRAM. If you're planning to work with AI workflows, having that amount of VRAM is a solid foundation. Even a 5060 could handle smaller quantized models, but you'll find yourself hitting walls quickly with less VRAM if your projects grow.
Thanks for the tip! I'm leaning towards GPUs with 16GB VRAM, but right now, even a 4070 is outside my budget. Just to clarify, I'm not focusing on AGI but rather on training smaller ML models for edge computing. I've done some training on CPUs but want a more comprehensive setup.