Can I Run StarCoder 2 on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)?

Written by Jakub Rusinowski · Last updated February 28, 2024

Yes — comfortably

Yes, comfortably — StarCoder 2 15B at Q8_0 needs about 17.9 GB of the 32 GB usable on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM), leaving ~14.1 GB spare and running at ~67.8 tok/s (estimated), with room for about 16,384 tokens of context.

Confidence: high · Recommended quantization: Q8_0 · Estimated speed: ~67.8 tok/s

RTX 5090 Desktop (32 GB VRAM, 64 GB RAM) — what it gives a model

Usable memory for models32 GB
Memory bandwidth1792 GB/s
Form factorDesktop
Operating systemWindows or Linux
Memory upgradeableYes

StarCoder 2 on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM): memory by quantization

QuantMemory neededFits 32 GB?Max contextEst. speedDownload
F1632.5 GB✗ No31 GB
Q8_017.9 GB✓ Yes16K~67.8 tok/s16.5 GB
Q6_K14.2 GB✓ Yes16K~83.7 tok/s12.7 GB
Q5_K_M12.5 GB✓ Yes16K~93.7 tok/s11 GB
Q4_K_M10.8 GB✓ Yes16K~105.7 tok/s9.4 GB
Q3_K_M8.1 GB✓ Yes16K~134.9 tok/s6.6 GB
Q2_K6.6 GB✓ Yes16K~159.1 tok/s5.1 GB

Which StarCoder 2 sizes fit

VariantNeeds at Q4_K_MFits?Est. speed
StarCoder 2 15B10.8 GB✓ Fits~105.7 tok/s
StarCoder 2 7B5.7 GB✓ Fits~176.1 tok/s
StarCoder 2 3B2.9 GB✓ Fits~271.2 tok/s

RTX 5090 desktop limitations

Recommended setup

Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving

How these numbers are calculated

FAQ

Can I run StarCoder 2 on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)?

Yes, comfortably — StarCoder 2 15B at Q8_0 needs about 17.9 GB of the 32 GB usable on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM), leaving ~14.1 GB spare and running at ~67.8 tok/s (estimated), with room for about 16,384 tokens of context.

Which quantization of StarCoder 2 should I use on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)?

Q8_0 — it needs about 17.9 GB of the 32 GB available, downloads as roughly 16.5 GB, and runs at an estimated 67.8 tokens/sec with up to 16K of context.

What limits StarCoder 2 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)

StarCoder 2 on GPUs

What This Model Is Good At

Model & Tools

← Can I Run It? | StarCoder 2 model page | Check your hardware