Gourieff
commented
11 months ago Hello These messages are not errors, but they shouldn't go after "ReActor preheating... Device: CUDA"... I don't use Stability Matrix, so cannot tell you exactly what the reason is What "Console Log Level" option do you use?
Sorry, don't know how to get this
Just copy a full log of Stability Matrix console while you have it running and errors appear
First, confirm
What happened?
I get a ton of error messages, which started recently, which seems to correlate with slower processing. I'm afraid I can't associate this with any changes I made or with any updates. I'm running in Stability Matrix and I posted this on there also because I don't know if it is Reactor related or not. Reactor still works but perhaps slower.
ReActor preheating... Device: CUDA Applied providers: ['CUDAExecutionProvider', 'CPUExecutionProvider'], with options: {'CUDAExecutionProvider': {'cudnn_conv_algo_search': 'EXHAUSTIVE', 'device_id': '0', 'cudnn_conv1d_pad_to_nc1d': '0', 'has_user_compute_stream': '0', 'gpu_external_alloc': '0', 'enable_cuda_graph': '0', 'gpu_mem_limit': '18446744073709551615', 'gpu_external_free': '0', 'gpu_external_empty_cache': '0', 'arena_extend_strategy': 'kNextPowerOfTwo', 'do_copy_in_default_stream': '1', 'cudnn_conv_use_max_workspace': '1', 'tunable_op_enable': '0', 'tunable_op_tuning_enable': '0', 'tunable_op_max_tuning_duration_ms': '0', 'enable_skip_layer_norm_strict_mode': '0'}, 'CPUExecutionProvider': {}} model ignore: C:\Users\D S Tong/.insightface\models\buffalo_l\1k3d68.onnx landmark_3d_68 Applied providers: ['CUDAExecutionProvider', 'CPUExecutionProvider'], with options: {'CUDAExecutionProvider': {'cudnn_conv_algo_search': 'EXHAUSTIVE', 'device_id': '0', 'cudnn_conv1d_pad_to_nc1d': '0', 'has_user_compute_stream': '0', 'gpu_external_alloc': '0', 'enable_cuda_graph': '0', 'gpu_mem_limit': '18446744073709551615', 'gpu_external_free': '0', 'gpu_external_empty_cache': '0', 'arena_extend_strategy': 'kNextPowerOfTwo', 'do_copy_in_default_stream': '1', 'cudnn_conv_use_max_workspace': '1', 'tunable_op_enable': '0', 'tunable_op_tuning_enable': '0', 'tunable_op_max_tuning_duration_ms': '0', 'enable_skip_layer_norm_strict_mode': '0'}, 'CPUExecutionProvider': {}} Applied providers: ['CUDAExecutionProvider', 'CPUExecutionProvider'], with options: {'CUDAExecutionProvider': {'cudnn_conv_algo_search': 'EXHAUSTIVE', 'device_id': '0', 'cudnn_conv1d_pad_to_nc1d': '0', 'has_user_compute_stream': '0', 'gpu_external_alloc': '0', 'enable_cuda_graph': '0', 'gpu_mem_limit': '18446744073709551615', 'gpu_external_free': '0', 'gpu_external_empty_cache': '0', 'arena_extend_strategy': 'kNextPowerOfTwo', 'do_copy_in_default_stream': '1', 'cudnn_conv_use_max_workspace': '1', 'tunable_op_enable': '0', 'tunable_op_tuning_enable': '0', 'tunable_op_max_tuning_duration_ms': '0', 'enable_skip_layer_norm_strict_mode': '0'}, 'CPUExecutionProvider': {}} find model: C:\Users\D S Tong/.insightface\models\buffalo_l\det_10g.onnx detection [1, 3, '?', '?'] 127.5 128.0 Applied providers: ['CUDAExecutionProvider', 'CPUExecutionProvider'], with options: {'CUDAExecutionProvider': {'cudnn_conv_algo_search': 'EXHAUSTIVE', 'device_id': '0', 'cudnn_conv1d_pad_to_nc1d': '0', 'has_user_compute_stream': '0', 'gpu_external_alloc': '0', 'enable_cuda_graph': '0', 'gpu_mem_limit': '18446744073709551615', 'gpu_external_free': '0', 'gpu_external_empty_cache': '0', 'arena_extend_strategy': 'kNextPowerOfTwo', 'do_copy_in_default_stream': '1', 'cudnn_conv_use_max_workspace': '1', 'tunable_op_enable': '0', 'tunable_op_tuning_enable': '0', 'tunable_op_max_tuning_duration_ms': '0', 'enable_skip_layer_norm_strict_mode': '0'}, 'CPUExecutionProvider': {}} find model: C:\Users\D S Tong/.insightface\models\buffalo_l\genderage.onnx genderage ['None', 3, 96, 96] 0.0 1.0 Applied providers: ['CUDAExecutionProvider', 'CPUExecutionProvider'], with options: {'CUDAExecutionProvider': {'cudnn_conv_algo_search': 'EXHAUSTIVE', 'device_id': '0', 'cudnn_conv1d_pad_to_nc1d': '0', 'has_user_compute_stream': '0', 'gpu_external_alloc': '0', 'enable_cuda_graph': '0', 'gpu_mem_limit': '18446744073709551615', 'gpu_external_free': '0', 'gpu_external_empty_cache': '0', 'arena_extend_strategy': 'kNextPowerOfTwo', 'do_copy_in_default_stream': '1', 'cudnn_conv_use_max_workspace': '1', 'tunable_op_enable': '0', 'tunable_op_tuning_enable': '0', 'tunable_op_max_tuning_duration_ms': '0', 'enable_skip_layer_norm_strict_mode': '0'}, 'CPUExecutionProvider': {}} model ignore: C:\Users\D S Tong/.insightface\models\buffalo_l\w600k_r50.onnx recognition
There's a lot more of this, and I think this is partly why A1111 takes a lot longer to fully load up in Stab Matrix.
Any help will be greatly appreciated.
Thanks !
Steps to reproduce the problem
Not relevant
Sysinfo
Windows 10 64 bit, nVidia RTX-3090 Things still work, but slower.
Relevant console log
Additional information
No response