Description
Name and Version
- llama-cli --version
ggml_cuda_init: failed to initialize CUDA: OS call failed or operation not supported on this OS
load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
version: 5306 (alt1)
built with x86_64-alt-linux-gcc (GCC) 14.2.1 20241028 (ALT Sisyphus 14.2.1-alt1) for x86_64-alt-linux
Operating systems
Linux
Which llama.cpp modules do you know to be affected?
No response
Command line
Some tests fail:
+ ctest --test-dir x86_64-alt-linux --output-on-failure --force-new-ctest-process --parallel 20 -j1 -E test-eval-callback
Test project /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux
...
04:24:54 Start 23: test-arg-parser
04:24:54 23/28 Test #23: test-arg-parser ...................Subprocess aborted***Exception: 0.13 sec
04:24:54 ggml_cuda_init: failed to initialize CUDA: OS call failed or operation not supported on this OS
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 error while handling argument "-m": expected value for argument
04:24:54
04:24:54 usage:
04:24:54 -m, --model FNAME model path (default: `models/$filename` with filename from `--hf-file`
04:24:54 or `--model-url` if set, otherwise models/7B/ggml-model-f16.gguf)
04:24:54 (env: LLAMA_ARG_MODEL)
04:24:54
04:24:54
04:24:54 to show complete usage, run with -h
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 error while handling argument "-ngl": stoi
04:24:54
04:24:54 usage:
04:24:54 -ngl, --gpu-layers, --n-gpu-layers N number of layers to store in VRAM
04:24:54 (env: LLAMA_ARG_N_GPU_LAYERS)
04:24:54
04:24:54
04:24:54 to show complete usage, run with -h
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 error while handling argument "-sm": invalid value
04:24:54
04:24:54 usage:
04:24:54 -sm, --split-mode {none,layer,row} how to split the model across multiple GPUs, one of:
04:24:54 - none: use one GPU only
04:24:54 - layer (default): split layers and KV across GPUs
04:24:54 - row: split rows across GPUs
04:24:54 (env: LLAMA_ARG_SPLIT_MODE)
04:24:54
04:24:54
04:24:54 to show complete usage, run with -h
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 error: invalid argument: --draft
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 error while handling environment variable "LLAMA_ARG_THREADS": stoi
04:24:54
04:24:54
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 load_backend: loaded CUDA backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cuda.so
04:24:54 load_backend: loaded CPU backend from /usr/src/RPM/BUILD/llama.cpp-5306/x86_64-alt-linux/bin/libggml-cpu-haswell.so
04:24:54 warn: LLAMA_ARG_MODEL environment variable is set, but will be overwritten by command line argument -m
04:24:54 terminate called after throwing an instance of 'std::runtime_error'
04:24:54 what(): error: cannot make GET request: Could not resolve hostname
04:24:54
...
04:24:54 96% tests passed, 1 tests failed out of 28
04:24:54 The following tests FAILED:
04:24:54 23 - test-arg-parser (Subprocess aborted) main
04:24:54 Errors while running CTestProblem description & steps to reproduce
I just built b5306 and run ctest, the same workflow b4855 worked without the error.
First Bad Commit
No response