Home › GPUs & Macs › AMD Radeon RX 7900 XTX vs NVIDIA RTX 3090
AMD Radeon RX 7900 XTX vs NVIDIA RTX 3090 for local LLMs
Across 6 tests measured on both, the NVIDIA RTX 3090 is faster in 4 and generates a median 1.52× the tokens/s of the AMD Radeon RX 7900 XTX.
On Qwen3.8 27B (Q4_K_M, llama-server) the AMD Radeon RX 7900 XTX generates 30.7 tokens/s and the NVIDIA RTX 3090 36.6 tokens/s (measured).
Same model, same quant, same test
| Model | Quant / test | Radeon RX 7900 XTX tok/s | RTX 3090 tok/s | Faster | Sources and run details |
|---|---|---|---|---|---|
| Qwen3.8 27B | Q4_K_M llama-server | 30.7 | 36.6 CUDA | RTX 3090 +19% | qwen38-mtp Radeon RX 7900 XTX · qwen38-mtp RTX 3090 |
| Qwen3.8 27B | UD-Q4_K_XL llama-server | 41.0 Vulkan | 34.0 CUDA | Radeon RX 7900 XTX +21% | qwen38-mtp Radeon RX 7900 XTX · qwen38-mtp RTX 3090 |
| Llama 2 7B | Q4_0 llama-bench | 191 Vulkan | 172 Vulkan | Radeon RX 7900 XTX +11% | llama.cpp #10879 Radeon RX 7900 XTX · llama.cpp #10879 RTX 3090 |
| Llama 3.1 8B Instruct | Q4_K_M LocalScore | 42.3 llamafile | 96.3 llamafile | RTX 3090 +128% | LocalScore Radeon RX 7900 XTX · LocalScore RTX 3090 |
| Llama 3.2 1B Instruct | Q4_K_M LocalScore | 78.4 llamafile | 270 llamafile | RTX 3090 +244% | LocalScore Radeon RX 7900 XTX · LocalScore RTX 3090 |
| Qwen2.5 14B Instruct | Q4_K_M LocalScore | 26.8 llamafile | 56.2 llamafile | RTX 3090 +110% | LocalScore Radeon RX 7900 XTX · LocalScore RTX 3090 |
Generation speed in tokens/s, best measured backend for each chip (shown under the number). Backends can differ between the two chips; check them before drawing conclusions. Measured results only, nothing estimated.
Specs
| AMD Radeon RX 7900 XTX | NVIDIA RTX 3090 | |
|---|---|---|
| Memory | 24 GB | 24 GB |
| Memory bandwidth | 960 GB/s | 936 GB/s |
All results for the AMD Radeon RX 7900 XTX · All results for the NVIDIA RTX 3090 · More comparisons