| --- |
| license: cc-by-4.0 |
| pretty_name: RTX 5080 Local LLM Throughput (Measured, Ollama) |
| language: |
| - en |
| tags: |
| - benchmark |
| - gpu |
| - llm-inference |
| - nvidia |
| - rtx-5080 |
| - ollama |
| size_categories: |
| - n<1K |
| --- |
| |
| # RTX 5080 Local LLM Throughput — Measured, Not Estimated |
|
|
| Measured decode and prefill tokens/second, load times, and VRAM residency for |
| local LLMs on a single retail NVIDIA GeForce RTX 5080 (16 GB): gpt-oss:20b |
| (MXFP4), Qwen 2.5 14B and 7B (Q4_K_M), and Llama 3.2 3B (Q4_K_M), at 4k and |
| 16k context with ~500 and ~1,650-token prompts. |
|
|
| ## Method |
|
|
| - Every figure is the **median of 3 runs** using ollama's native |
| `eval_count`/`eval_duration` counters — never wall-clock division. |
| - Every repetition prepends a unique nonce so each run pays a **full |
| prefill**; without this, warm runs hit ollama's prompt cache and report |
| ~60,000 tok/s cache-hit rates that measure nothing. |
| - Generation pinned: temperature 0, seed 42, 256-token target output. |
| - Model identity pinned by digest per row (a tag like `qwen2.5:14b` can |
| silently move upstream). |
| - Bench environment recorded per row: RTX 5080, driver 610.88, ollama 0.32.1, |
| Windows 11, Ryzen 7 7800X3D. |
|
|
| ## Headline results |
|
|
| | Model | Quant | Decode tok/s | Resident VRAM | |
| |---|---|---:|---:| |
| | Llama 3.2 3B | Q4_K_M | 298–313 | 5.5 GiB (alone) | |
| | gpt-oss 20B | MXFP4 | 187–190 | 14.0–14.4 GiB | |
| | Qwen 2.5 7B | Q4_K_M | 177–180 | 6.2–7.0 GiB | |
| | Qwen 2.5 14B | Q4_K_M | 94–97 | 10.8–13.2 GiB | |
|
|
| The 20.9B MXFP4 model decodes 2x faster than the dense 14B at Q4_K_M and |
| matches the dense 7B — architecture and quantization beat parameter count. |
| Decode speed holds flat from 4k to 16k context at these prompt sizes. |
|
|
| ## Caveats |
|
|
| Single retail unit, single driver, single runtime — these are *this card's* |
| numbers. The VRAM column is whole-GPU `nvidia-smi` at measurement time and can |
| include a previously-benched model still resident (rows measured alone are the |
| lower readings). |
|
|
| ## Canonical source |
|
|
| Method detail, change log, and living updates: |
| https://techfuelhq.com/data/rtx-5080-llm-throughput/ |
|
|
| ## License |
|
|
| CC BY 4.0 — attribute as "TechFuelHQ RTX 5080 LLM Throughput Dataset" linking |
| to the canonical page. |
|
|
