Dual 3060 MoE loading issue.

I had one RTX3060 and was able to use all models like Qwen3.6 27B, Qwen3.6 35B and Qwen3.5 122B with CPU offload. I've added second 3060 and now only dense model can be loaded as usual, 35B loaded only once and 122B refuses to load at all. What setting should I change or where to look for potential problem solving solutions? Update: Seems 12 layers in VRAM is the limit for dual 3060 with qwen3.5-122b-a10b. Reduced to 12 and model loaded. 35b-a3b loaded as well after PC restart and when I changed GPU priorit

添加评论
点赞收藏
点踩分享查看原文
评论
?
参与讨论