# old Xeon X99 PC with 256GB RAM + RTX 3090 Ti worth it for Strata LLM inference? — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/6d48d08b435245d1

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-10-29 17:44 ingested 2026-10-30 02:40

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1xiqujs/old_xeon_x99_pc_with_256gb_ram_rtx_3090_ti_worth/)

## retained text



```
&#x200B;

I found this used PC in my local market for around $420:

\- CPU: Xeon E5-2640 v3 (8C/16T)

\- Motherboard: X99 (exact model unknown)

\- RAM: 16GB DDR4 (planning to upgrade to 256GB)

\- Storage: 128GB SSD

\- Includes a 1080p 144Hz monitor and accessories.

I'm planning to add my RTX 3090 Ti 24GB and upgrade to 256GB DDR4, mainly to run large LLMs using Strata Engine with RAM/VRAM offloading.

This is one of the few affordable platforms available locally that might support more than 128GB RAM.

My question: Would this old Xeon be usable for Strata inference, or would its CPU and memory bandwidth make it painfully slow? Is it worth buying, or should I avoid it?
```



document fetched 2026-10-30 00:44 · canonical https://reddit.com/r/LocalLLaMA/comments/1xiqujs/old_xeon_x99_pc_with_256gb_ram_rtx_3090_ti_worth
