GPUs & Macs › NVIDIA Quadro P2000
NVIDIA Quadro P2000: local LLM speed
Llama 2 7B (Q4_0) generates 23.8 tokens/s on the NVIDIA Quadro P2000 (Vulkan), and processes prompts at 182 tokens/s.
Measured results
| Model | Generate tok/s | Prompt tok/s | Quant | Backend | Test | Source |
|---|---|---|---|---|---|---|
| Llama 2 7B | 23.8 | 182 | Q4_0 | Vulkan FA on | tg128 | llama.cpp #10879 details |
| Llama 2 7B | 23.8 | 312 | Q4_0 | CUDA FA on | tg128 | llama.cpp #15013 details |
| Llama 2 7B | 23.6 | 309 | Q4_0 | CUDA FA off | tg128 | llama.cpp #15013 details |
| Llama 2 7B | 23.1 | 170 | Q4_0 | Vulkan FA off | tg128 | llama.cpp #10879 details |
| Llama 3.2 1B Instruct | 58.3 | 1,353 | Q4_K_M | llamafile | LocalScore avg | LocalScore details |
What fits on the NVIDIA Quadro P2000
Estimates for this chip are scaled from a measured run: Llama 2 7B Q4_0 at 23.8 tok/s (Vulkan, source). Fit assumes the whole model in GPU memory with ~8K context.
| Model | Best quant that fits | File | Expected tok/s |
|---|---|---|---|
| Qwen3.5-9B | UD-IQ2_XXS | 3.2 GB | 28.5 estimate |
| Qwen3.5-4B | Q6_K | 3.5 GB | 25.8 estimate |
| Qwen3-4B | Q6_K | 3.3 GB | 27.5 estimate |
| gemma-3-1b-it | Q8_0 | 1.1 GB | 75.8 estimate |
| gemma-4-E2B-it | Q4_0 | 2.8 GB | 32.0 estimate |
| DeepSeek-R1-0528-Qwen3-8B | UD-IQ3_XXS | 3.4 GB | 26.7 estimate |
| Llama-3.2-3B-Instruct | Q8_0 | 3.4 GB | 26.6 estimate |
| Phi-4-mini-instruct | Q6_K | 3.2 GB | 28.8 estimate |
| gemma-3-4b-it | Q6_K | 3.2 GB | 28.5 estimate |
| gemma-3-12b-it | UD-IQ1_M | 3.2 GB | 28.0 estimate |
| Qwen3.5-0.8B | Q8_0 | 0.8 GB | 75.8 estimate |
| Llama-3.2-3B-Instruct-uncensored | Q6_K | 3.0 GB | 30.6 estimate |
| Qwen3-0.6B | Q8_0 | 0.8 GB | 75.8 estimate |
| Qwen2.5-Coder-7B-Instruct-abliterated | Q3_K_S | 3.5 GB | 26.0 estimate |
Similar hardware
NVIDIA T1000 · NVIDIA Quadro P1000 · NVIDIA RTX A2000 · NVIDIA Quadro RTX 4000 · NVIDIA RTX A2000 12GB · NVIDIA RTX A4000 · NVIDIA RTX 2000 Ada · NVIDIA RTX A4500