#3632 [Engine] DirectFile: cannot open - with ervery model i Downloaded

Descrição

Hey there,

great work with Strata.

I have a PC with 32gb RAM and a dedicated RTX 5060Ti with 16gb RAM. With every model I downloaded using Strata (with the new installer or the old script) I always ran in ti the same issue:

cat /home/michael/.local/share/strata/install/strata-swift-iq2_xs.log
ggml_cuda_init: failed to initialize CUDA: no CUDA-capable device is detected
load_hparams: Qwen-VL models require at minimum 1024 image tokens to function correctly on grounding tasks
load_hparams: if you encounter problems with accuracy, try adding --image-min-tokens 1024
load_hparams: more info: https://github.com/ggml-org/llama.cpp/issues/16842

strata-vision: model files loaded in 1.8 s
strata-vision: on the CPU, 8 threads, no warm-up
[strata] 2026-10-21 17:50:34 engine started: /home/michael/.local/share/strata/components/engine/0.1.47-linux-x86_64-cuda-cuda13/strata --serve --pack /home/michael/AI/Strata/packs/swift-iq2_xs --native /home/michael/AI/Strata/models/swift-IQ2_XS/Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ2_XS-00001-of-00002.gguf --ple-gguf /home/michael/AI/Strata/models/swift-IQ2_XS/Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ2_XS-00001-of-00002.gguf --expert-profile /tmp/.mount_StrataAAodBP/usr/lib/Strata/control/data/expert-profile.bin --expert-cache auto --prefill auto --spec 4 --spec-min-p 0.5 --mtp /home/michael/AI/Strata/mtp/rt --max-context 131072 --kv int8 --mmap-experts --vision --pool-workers 6 --vram-reserve-mib 2048
strata generate: memory guard: limit 30.7 GiB (source: meminfo), current 4.8 GiB, reclaimable cache credited 0.0 GiB; available 25.9 GiB (MemAvailable 25.9 GiB), floor 2560 MiB
strata generate: PCIe probe: 6.7 GB/s host->device (best of 6.7 6.7 6.7 6.7) -> pcie_frac 0.18 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: /home/michael/AI/Strata/packs/swift-iq2_xs experts (largest blob 1.51 MB), token embedding IQ4_XS in mapped host memory (322 MiB, 0.6 s)
strata generate: 1466 MiB of weights loaded from /home/michael/AI/Strata/packs/swift-iq2_xs in 1.9 s (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1474.85 MiB of weights, in 2.3 s
strata generate: DirectFile: cannot open /home/michael/AI/Strata/models/swift-IQ2_XS/Swift-Qwen3.8-Flash-Next-GSQ-RCO-IQ2_XS-00001-of-00002.gguf

at /home/michael/.local/share/strata/install/strata-gyro-gyro-s.log
ggml_cuda_init: failed to initialize CUDA: no CUDA-capable device is detected
load_hparams: Qwen-VL models require at minimum 1024 image tokens to function correctly on grounding tasks
load_hparams: if you encounter problems with accuracy, try adding --image-min-tokens 1024
load_hparams: more info: https://github.com/ggml-org/llama.cpp/issues/16842

strata-vision: model files loaded in 1.9 s
strata-vision: on the CPU, 8 threads, no warm-up
[strata] 2026-10-21 20:01:40 engine started: /home/michael/.local/share/strata/components/engine/0.1.47-linux-x86_64-cuda-cuda13/strata --serve --pack /home/michael/AI/Strata/packs/gyro-gyro-s --native /home/michael/AI/Strata/models/gyro-Gyro-S/Qwen3.8-Flash-Next-Gyro-S-TQ1_0.gguf --ple-gguf /home/michael/AI/Strata/models/gyro-Gyro-S/Qwen3.8-Flash-Next-Gyro-S-TQ1_0.gguf --expert-profile /tmp/.mount_StrataAAodBP/usr/lib/Strata/control/data/expert-profile.bin --expert-cache auto --prefill auto --spec 4 --spec-min-p 0.5 --mtp /home/michael/AI/Strata/mtp/rt --max-context 131072 --kv int8 --mmap-experts --kv-resident 32768 --vision --pool-workers 6 --vram-reserve-mib 2048
strata generate: memory guard: limit 30.7 GiB (source: meminfo), current 4.2 GiB, reclaimable cache credited 0.0 GiB; available 26.5 GiB (MemAvailable 26.5 GiB), floor 2560 MiB
strata generate: APR pack: Hadamard-folded experts (block 128)
strata generate: PCIe probe: 6.7 GB/s host->device (best of 6.7 6.7 6.7 6.7) -> pcie_frac 0.18 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: /home/michael/AI/Strata/packs/gyro-gyro-s experts (largest blob 1.05 MB), token embedding Q6_K in mapped host memory (497 MiB, 0.9 s)
strata generate: 1466 MiB of weights loaded from /home/michael/AI/Strata/packs/gyro-gyro-s in 1.9 s (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 1935.94 MiB of weights, in 5.2 s
strata generate: DirectFile: cannot open /home/michael/AI/Strata/models/gyro-Gyro-S/Qwen3.8-Flash-Next-Gyro-S-TQ1_0.gguf

Is there any problem with my RTX oder with my RAM? Any help would be great.

Thx in advcnce, mike