OpenAgentsInc / workerbee

GPUtopia worker process for client machine
GNU Affero General Public License v3.0
18 stars 8 forks source link

--main_gpu does not select specified gpu - Intel Arc GPUs #9

Open Chadwicksracing opened 1 year ago

Chadwicksracing commented 1 year ago

C:\Users\rtm\Documents>gputopia-worker-opencl-win-64.exe --ln_url silentmeadow15054@getalby.com --force_layers 33 --test_model TheBloke/CodeLlama-7B-Instruct-GGUF:Q4_K_M --debug --main_gpu 0 --tensor_split 0,1,0,1,0,1,0,1,0,1 2023-10-07 10:52:12,482 - ai_worker.main - DEBUG - no nvidia: NVML Shared Library Not Found {'auth_key': '', 'cl_driver_version': 'OpenCL 3.0 ', 'cl_gpus': [{'clock': 2400, 'clock_unit': 'mhz', 'memory': 8319.0, 'name': 'Intel(R) Arc(TM) A750 Graphics', 'uuid': None}, {'clock': 2400, 'clock_unit': 'mhz', 'memory': 8319.0, 'name': 'Intel(R) Arc(TM) A750 Graphics', 'uuid': None}, {'clock': 2400, 'clock_unit': 'mhz', 'memory': 8319.0, 'name': 'Intel(R) Arc(TM) A750 Graphics', 'uuid': None}, {'clock': 2400, 'clock_unit': 'mhz', 'memory': 8319.0, 'name': 'Intel(R) Arc(TM) A750 Graphics', 'uuid': None}, {'clock': 2400, 'clock_unit': 'mhz', 'memory': 8319.0, 'name': 'Intel(R) Arc(TM) A750 Graphics', 'uuid': None}], 'cpu_count': 16, 'disk_space': 428351, 'ln_address': 'silentmeadow15054@getalby.com', 'ln_url': 'silentmeadow15054@getalby.com', 'nv_driver_version': None, 'nv_gpu_count': None, 'nv_gpus': [], 'vram': 5996646400, 'web_gpus': [], 'worker_id': '', 'worker_version': '0.1.9'} 2023-10-07 10:52:13,051 - ai_worker.main - DEBUG - loading model: TheBloke/CodeLlama-7B-Instruct-GGUF:Q4_K_M ggml_opencl: selecting platform: 'Intel(R) OpenCL Graphics' ggml_opencl: selecting device: 'Intel(R) Arc(TM) A750 Graphics' ggml_opencl: device FP16 support: true Traceback (most recent call last): File "ai_worker__main.py", line 4, in File "ai_worker\main.py", line 385, in main File "asyncio\runners.py", line 190, in run File "asyncio\runners.py", line 118, in run File "asyncio\base_events.py", line 653, in run_until_complete File "ai_worker\main.py", line 144, in run File "ai_worker\main.py", line 114, in test_model File "ai_worker\main.py", line 201, in load_model File "llama_cpp\server\app.py", line 342, in create_app File "llama_cpp\llama.py", line 312, in init IndexError: invalid index [12344] Failed to execute script 'main__' due to unhandled exception!

image

--tensor_split not working either

earonesty commented 12 months ago

agreed, needs to be looked into.