[11:21:16] START 총계 사용 여분 공유 버퍼/캐시 가용 메모리: 121 8 5 0 110 113 ds4: Linux cuda backend set oom_score_adj=1000 ds4: CUDA backend initialized on NVIDIA GB10 (sm_121) dev=0 ds4: direct I/O enabled for /mnt/ai/weights/ds4-v4-flash-vision/Basic128-Routed-IQ2_M/DeepSeek-V4-Flash-Vision-Exp-Abliterated-Basic128-Routed-IQ2_M.gguf align=4096 ds4: unsupported tensor type 39 for blk.10.ffn_gate_exps.weight ds4: CUDA (no-copy) host registration skipped: operation not supported ds4: CUDA preparing model tensor mappings ds4: CUDA loading model tensors into device cache ds4: CUDA loading model tensors 2.59 GiB cached ds4: CUDA prepared model tensor mappings 2.87 GiB ds4: CUDA loading model tensors 5.06 GiB cached ds4: CUDA prepared model tensor mappings 5.59 GiB ds4: CUDA loading model tensors 7.46 GiB cached ds4: CUDA prepared model tensor mappings 8.45 GiB ds4: CUDA loading model tensors 9.93 GiB cached ds4: CUDA prepared model tensor mappings 11.31 GiB ds4: CUDA loading model tensors 12.32 GiB cached ds4: CUDA prepared model tensor mappings 14.02 GiB ds4: CUDA loading model tensors 14.81 GiB cached ds4: CUDA loading model tensors 16.02 GiB cached ds4: CUDA prepared model tensor mappings 16.37 GiB ds4: CUDA loading model tensors 18.38 GiB cached ds4: CUDA prepared model tensor mappings 19.01 GiB ds4: CUDA loading model tensors 20.77 GiB cached ds4: CUDA prepared model tensor mappings 21.76 GiB ds4: CUDA loading model tensors 23.24 GiB cached ds4: CUDA prepared model tensor mappings 24.62 GiB ds4: CUDA loading model tensors 25.70 GiB cached ds4: CUDA prepared model tensor mappings 27.26 GiB ds4: CUDA loading model tensors 28.01 GiB cached ds4: CUDA prepared model tensor mappings 30.01 GiB ds4: CUDA loading model tensors 30.51 GiB cached ds4: CUDA loading model tensors 32.01 GiB cached ds4: CUDA prepared model tensor mappings 32.21 GiB ds4: CUDA loading model tensors 34.52 GiB cached ds4: CUDA prepared model tensor mappings 35.07 GiB ds4: CUDA loading model tensors 37.01 GiB cached ds4: CUDA prepared model tensor mappings 37.93 GiB ds4: CUDA loading model tensors 39.52 GiB cached ds4: CUDA prepared model tensor mappings 40.65 GiB ds4: CUDA loading model tensors 42.01 GiB cached ds4: CUDA prepared model tensor mappings 43.51 GiB ds4: CUDA loading model tensors 44.52 GiB cached ds4: CUDA prepared model tensor mappings 46.37 GiB ds4: CUDA loading model tensors 46.88 GiB cached ds4: CUDA loading model tensors 48.02 GiB cached ds4: CUDA prepared model tensor mappings 48.06 GiB ds4: CUDA loading model tensors 50.38 GiB cached ds4: CUDA prepared model tensor mappings 50.77 GiB ds4: CUDA loading model tensors 52.90 GiB cached ds4: CUDA prepared model tensor mappings 53.63 GiB ds4: CUDA loading model tensors 55.26 GiB cached ds4: CUDA prepared model tensor mappings 56.82 GiB ds4: CUDA loading model tensors 57.65 GiB cached ds4: CUDA prepared model tensor mappings 59.68 GiB ds4: CUDA loading model tensors 60.12 GiB cached ds4: CUDA loading model tensors 62.57 GiB cached ds4: CUDA prepared model tensor mappings 62.95 GiB ds4: CUDA loading model tensors 64.01 GiB cached ds4: CUDA prepared model tensor mappings 64.01 GiB ds4: CUDA loading model tensors 66.37 GiB cached ds4: CUDA prepared model tensor mappings 66.76 GiB ds4: CUDA loading model tensors 68.76 GiB cached ds4: CUDA prepared model tensor mappings 69.51 GiB ds4: CUDA loading model tensors 71.26 GiB cached ds4: CUDA prepared model tensor mappings 72.70 GiB ds4: CUDA loading model tensors 73.63 GiB cached ds4: CUDA prepared model tensor mappings 75.88 GiB ds4: CUDA loading model tensors 76.01 GiB cached ds4: CUDA loading model tensors 78.45 GiB cached ds4: CUDA prepared model tensor mappings 79.07 GiB ds4: CUDA loading model tensors 80.01 GiB cached ds4: CUDA prepared model tensor mappings 80.13 GiB ds4: CUDA loading model tensors 82.32 GiB cached ds4: CUDA prepared model tensor mappings 83.32 GiB ds4: CUDA loading model tensors 84.82 GiB cached ds4: CUDA prepared model tensor mappings 86.51 GiB ds4: CUDA loading model tensors 87.26 GiB cached ds4: CUDA loading model tensors 89.31 GiB cached ds4: CUDA prepared model tensor mappings 89.56 GiB ds4: CUDA loading model tensors 91.60 GiB cached ds4: CUDA prepared model tensor mappings 92.51 GiB /tmp/loadgate.sh: 줄 15: 2867471 죽었음 timeout 1800 ./ds4 --cuda -m "$D/DeepSeek-V4-Flash-Vision-Exp-Abliterated-Basic128-Routed-IQ2_M.gguf" --vision "$D/mmproj-DeepSeek-V4-Flash-Vision-Exp-F16.gguf" --ctx 4096 -p "Reply with exactly: LOADGATE OK" [11:27:50] rc=0 총계 사용 여분 공유 버퍼/캐시 가용 메모리: 121 6 115 0 0 115