ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 2 CUDA devices:
  Device 0: NVIDIA GeForce RTX 3090, compute capability 8.6, VMM: yes
  Device 1: NVIDIA GeForce RTX 3090, compute capability 8.6, VMM: yes
| model                          |       size |     params | backend    | ngl |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | --------------: | -------------------: |
| qwen3vl 8B MXFP4 MoE           |   7.28 GiB |     8.19 B | CUDA       |  35 |             pp8 |        269.24 ± 4.22 |
| qwen3vl 8B MXFP4 MoE           |   7.28 GiB |     8.19 B | CUDA       |  35 |           tg128 |         43.42 ± 0.27 |

build: 92bb442ad (7040)