My first TR build
Hey everyone, putting together a high-end workstation build and wanted to get some eyes on it before pulling the trigger.
**Use case:**
Primarily local LLM inference (targeting 70B parameter models), running open-source GitHub projects that interact with those LLMs (think local AI assistants, RAG pipelines, agent frameworks), and some virtualization on the side.
**The build:**
| Component | Pick |
|---|---|
| CPU | AMD Threadripper PRO 9975WX |
| Motherboard | ASRock WRX90 WS EVO SATA 6Gb/s DDR5 |
| RAM | Kingston FURY Renegade Pro 128 GB (8×16 GB) DDR5-6400 ECC RDIMM |
| GPU | ASUS ROG Astral RTX 5090 32GB GDDR7 OC Edition |
| Storage (OS) | 2 TB Crucial T710 PCIe Gen5 NVMe |
| Storage (Models) | 4 TB WD Black SN850X PCIe Gen4 NVMe |
| Cooler | Noctua NH-U14S TR5-SP6 |
| PSU | Seasonic Prime TX-1600 |
| Case | Fractal Design Define 7 XL |
| Keyboard | Logitech Ergo K860 |
| Mouse | Logitech MX Anywhere 3S |
| OS | Windows 11 Pro |
A few things I'm intentionally planning for: the RAM is 8×16 GB to keep all 8 memory channels active from day one, with a straight swap to 8×32 GB when I need to go to 256 GB.
Anyone see any compatibility issues I'm missing? Also curious if this setup makes sense for the use case or if there's something I should swap out. Thanks!