Home › GPUs & Macs › NVIDIA RTX 4090 vs NVIDIA RTX 4090 D
NVIDIA RTX 4090 vs NVIDIA RTX 4090 D for local LLMs
Across 3 tests measured on both, the NVIDIA RTX 4090 D is faster in 3 and generates a median 1.23× the tokens/s of the NVIDIA RTX 4090.
On Llama 2 7B (Q4_0, llama-bench) the NVIDIA RTX 4090 generates 190 tokens/s and the NVIDIA RTX 4090 D 192 tokens/s (measured).
Same model, same quant, same test
| Model | Quant / test | RTX 4090 tok/s | RTX 4090 D tok/s | Faster | Sources and run details |
|---|---|---|---|---|---|
| Llama 2 7B | Q4_0 llama-bench | 190 Vulkan | 192 CUDA | even | llama.cpp #10879 RTX 4090 · llama.cpp #15013 RTX 4090 D |
| Llama 3.2 1B Instruct | Q4_K_M LocalScore | 292 llamafile | 359 llamafile | RTX 4090 D +23% | LocalScore RTX 4090 · LocalScore RTX 4090 D |
| Qwen2.5 14B Instruct | Q4_K_M LocalScore | 49.1 llamafile | 70.5 llamafile | RTX 4090 D +44% | LocalScore RTX 4090 · LocalScore RTX 4090 D |
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 4090 | NVIDIA RTX 4090 D | |
|---|---|---|
| Memory | 24 GB | 24 GB |
| Memory bandwidth | 1008 GB/s | 1008 GB/s |
All results for the NVIDIA RTX 4090 · All results for the NVIDIA RTX 4090 D · More comparisons