Question 1

Can I run StarCoder2 15B on a AMD Radeon RX 7900 XTX?

Accepted Answer

Yes — AMD Radeon RX 7900 XTX runs StarCoder2 15B fully on GPU at the Q8_0 quantization. Full GPU inference — 24 GB VRAM meets the 20 GB recommendation.

Question 2

What quantization of StarCoder2 15B should I use on a AMD Radeon RX 7900 XTX?

Accepted Answer

For 24 GB VRAM on the AMD Radeon RX 7900 XTX, the Q8_0 variant is the best fit. Estimated ~57.1 tokens/sec on the Q8_0 quantization.

Question 3

How fast does StarCoder2 15B run on AMD Radeon RX 7900 XTX?

Accepted Answer

Roughly 57.1 tokens/sec for Q8_0. Real speed depends on context length, backend (Ollama, llama.cpp, LM Studio), and KV cache size.

Question 4

What if AMD Radeon RX 7900 XTX is not enough for StarCoder2 15B?

Accepted Answer

Try a smaller quantization variant or run on CPU + RAM with the trade-off of slower generation.

Quantization	File Size	Min VRAM	Rec VRAM	Context	Verdict	Estimated tok/s
Q4_K_M	9 GB	10.5 GB	12 GB	4K / 16K	Full GPU	~85.3
Q8_0Best fit	16 GB	17.5 GB	20 GB	4K / 16K	Full GPU	~57.1

Can I Run StarCoder2 15B on AMD Radeon RX 7900 XTX?

Share this matchup

Every StarCoder2 15B quantization on AMD Radeon RX 7900 XTX