golgotha ~/build>~/local/llama/bin/llama-bench --hf-repo unsloth/gemma-4-E4B-it-GGUF
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 23983 MiB):
  Device 0: NVIDIA GeForce RTX 5090 Laptop GPU, compute capability 12.0, VMM: yes, VRAM: 23983 MiB
| model                          |       size |     params | backend    | ngl |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | --------------: | -------------------: |
| gemma4 E4B Q4_K - Medium       |   4.62 GiB |     7.52 B | CUDA       |  -1 |           pp512 |     9555.89 ± 101.42 |
| gemma4 E4B Q4_K - Medium       |   4.62 GiB |     7.52 B | CUDA       |  -1 |           tg128 |        165.64 ± 0.19 |

build: e3f542d8 (4236)