Home › GPUs & Macs › NVIDIA RTX 4080 vs NVIDIA RTX 4080 SUPER
NVIDIA RTX 4080 vs NVIDIA RTX 4080 SUPER for local LLMs
Across 4 tests measured on both, the NVIDIA RTX 4080 is faster in 2 and generates a median 1.10× the tokens/s of the NVIDIA RTX 4080 SUPER.
On Llama 2 7B (Q4_0, llama-bench) the NVIDIA RTX 4080 generates 143 tokens/s and the NVIDIA RTX 4080 SUPER 150 tokens/s (measured).
Same model, same quant, same test
| Model | Quant / test | RTX 4080 tok/s | RTX 4080 SUPER tok/s | Faster | Sources and run details |
|---|---|---|---|---|---|
| Llama 2 7B | Q4_0 llama-bench | 143 CUDA | 150 Vulkan | RTX 4080 SUPER +5% | llama.cpp #15013 RTX 4080 · llama.cpp #10879 RTX 4080 SUPER |
| Llama 3.1 8B Instruct | Q4_K_M LocalScore | 88.3 llamafile | 66.6 llamafile | RTX 4080 +33% | LocalScore RTX 4080 · LocalScore RTX 4080 SUPER |
| Llama 3.2 1B Instruct | Q4_K_M LocalScore | 241 llamafile | 194 llamafile | RTX 4080 +24% | LocalScore RTX 4080 · LocalScore RTX 4080 SUPER |
| Qwen2.5 14B Instruct | Q4_K_M LocalScore | 35.2 llamafile | 44.9 llamafile | RTX 4080 SUPER +27% | LocalScore RTX 4080 · LocalScore RTX 4080 SUPER |
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
| NVIDIA RTX 4080 | NVIDIA RTX 4080 SUPER | |
|---|---|---|
| Memory | 16 GB | 16 GB |
| Memory bandwidth | 717 GB/s | 736 GB/s |
All results for the NVIDIA RTX 4080 · All results for the NVIDIA RTX 4080 SUPER · More comparisons