Builds / 2x RTX 3090
2x RTX 3090 (llm-speed signed run, Oct 2026): Llama 3.3 70B Q4_K_M at 19.1 tok/s
Source: llm-speed.com ↗📏 How measured: reported by the original source; RigBase re-checked the numbers at that source
Legacy models only. Every model measured on this rig now has a newer successor: Llama 3.3 70B → Muse Glimmer 30B. The hardware numbers are still a useful reference.
- GPU(s)
- 2x RTX 3090
Amazon · eBay (used) · Newegg - VRAM / unified memory (GB)
- 48
- CPU
- AMD EPYC 7H12 (64 cores)
Amazon · eBay (used) · Newegg - Motherboard
- not stated
- RAM
- not stated
- PSU
- not stated
- Other
- Server host; llm-speed suite-v1 chat-short workload (128-token prompt, 256 tokens out, batch 1)
- Reported total cost
- not stated
- Power
- not stated
Measured speed (generation tokens/sec)
| Model | Quant | Engine | tok/s | Details |
|---|---|---|---|---|
| Llama 3.3 70B legacy | Q4_K_M | llama.cpp 0.7.0-dev | 19.1 |
Signed community run on llm-speed.com, 17 Oct 2026. chat-long (4K prompt) decode: 18.7 tok/s.
Source: https://llm-speed.com/r/r_n93ukk0mkmm
Blank fields = not stated in the source. We never estimate build data; GPU street prices above are separately sourced and dated.
Rent before you buy: try this model on a cloud GPU for a few dollars first: Runpod (referral link; sign up with Google to get the new-user bonus)