Three drivers were ordered to pay a total of £1,471’ Court for speeding and failing to identify a driver, issuing ...

Compiling the model physics step... Module mujoco_warp._src.smooth 7c923ff load on device 'cuda:0' took 2115.64 ms (compiled) Module mujoco_warp._src.smooth._tile_cholesky 6aea4a9 load on device 'cuda ...

it happens with both Qwen3-235B-A22B-UD-Q3_K_XL and Qwen3-235B-A22B-UD-Q4_K_XL. I am using 2 3090 gpus and 2 A4000, built with 1 copy of cache parameter. I think the multiple gpus can be the issue..