Can I Run Phi 3.5 Family on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)?
作者： Jakub Rusinowski · 最后更新： 2024年8月20日
Yes — comfortably
Yes, comfortably — Phi 3.5 Mini at Q8_0 needs about 5.6 GB of the 32 GB usable on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM), leaving ~26.4 GB spare and running at ~180.2 tok/s (estimated), with room for about 131,072 tokens of context.
Confidence: high · Recommended quantization: Q8_0 · Estimated speed: ~180.2 tok/s
See what else this hardware can run →
或在 Vast.ai 比较，低至 $0.35/小时 (typical low · varies)
作为亚马逊联盟成员，我们从符合条件的购买中获得收入。云 GPU 链接为推荐链接——我们可能获得佣金，您无需额外付费。
RTX 5090 Desktop (32 GB VRAM, 64 GB RAM) — what it gives a model
| Usable memory for models | 32 GB |
| Memory bandwidth | 1792 GB/s |
| Form factor | Desktop |
| Operating system | Windows or Linux |
| Memory upgradeable | Yes |
Phi 3.5 Family on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM): memory by quantization
| Quant | Memory needed | Fits 32 GB? | Max context | Est. speed | Download |
|---|---|---|---|---|---|
| F16 | 9.2 GB | ✓ Yes | 128K | ~121.9 tok/s | 7.6 GB |
| Q8_0 | 5.6 GB | ✓ Yes | 128K | ~180.2 tok/s | 4 GB |
| Q6_K | 4.7 GB | ✓ Yes | 128K | ~205.6 tok/s | 3.1 GB |
| Q5_K_M | 4.3 GB | ✓ Yes | 128K | ~219.8 tok/s | 2.7 GB |
| Q4_K_M | 3.9 GB | ✓ Yes | 128K | ~235.1 tok/s | 2.3 GB |
| Q3_K_M | 3.2 GB | ✓ Yes | 128K | ~266.6 tok/s | 1.6 GB |
| Q2_K | 2.9 GB | ✓ Yes | 128K | ~287.8 tok/s | 1.2 GB |
RTX 5090 desktop limitations
- 575 W board power — budget for a 1000 W+ PSU and the heat it puts into the room.
- Models larger than 32 GB must offload to system RAM, which costs roughly an order of magnitude in speed.
Recommended setup
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
How these numbers are calculated
- Memory need = quantized weights + KV cache at 8,192 tokens (f16) + 0.8 GB runtime overhead.
- 32 GB of VRAM on the NVIDIA GeForce RTX 5090 at 1792 GB/s.
- 64 GB of system RAM available for CPU offload when a model exceeds VRAM.
- Throughput is a memory-bandwidth roofline estimate, not a measurement. It is labelled "estimated" everywhere it appears.
- KV cache is computed from this model's published attention configuration.
FAQ
Can I run Phi 3.5 Family on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)?
Yes, comfortably — Phi 3.5 Mini at Q8_0 needs about 5.6 GB of the 32 GB usable on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM), leaving ~26.4 GB spare and running at ~180.2 tok/s (estimated), with room for about 131,072 tokens of context.
Which quantization of Phi 3.5 Family should I use on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)?
Q8_0 — it needs about 5.6 GB of the 32 GB available, downloads as roughly 4 GB, and runs at an estimated 180.2 tokens/sec with up to 128K of context.
What limits Phi 3.5 Family on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)?
Nothing binding — the model fits with headroom and generates at a usable speed on this hardware.
Which runtime should I use?
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
Other Computers
Other Models on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)
- Phi-4 Family on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)
- Phi-4 Mini on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)
- Poolside Laguna XS 2.1 on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)
- Qwen 2.5 Family on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)
- Qwen 2.5 VL on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)
Phi 3.5 Family on GPUs
- Phi 3.5 Family on NVIDIA GeForce RTX 5060 Ti 8GB
- Phi 3.5 Family on NVIDIA GeForce RTX 5060
- Phi 3.5 Family on NVIDIA GeForce RTX 4060
What This Model Is Good At
Model & Tools
← Can I Run It? | Phi 3.5 Family model page | Check your hardware