GPU usage going from python 3.10 > 3.11 stuck 100% cpu usage

[lobsterchan27]

I'm using rvc in 3.11 python environment for compatibility reasons

2024-05-15 01:54:17 | INFO | fairseq.tasks.hubert_pretraining | current directory is C:\Users\Lobby\Desktop\New folder (4)\WhisperX-Server 2024-05-15 01:54:17 | INFO | fairseq.tasks.hubert_pretraining | HubertPretrainingTask Config {'_name': 'hubert_pretraining', 'data': 'metadata', 'fine_tuning': False, 'labels': ['km'], 'label_dir': 'label', 'label_rate': 50.0, 'sample_rate': 16000, 'normalize': False, 'enable_padding': False, 'max_keep_size': None, 'max_sample_size': 250000, 'min_sample_size': 32000, 'single_target': False, 'random_crop': True, 'pad_audio': False} 2024-05-15 01:54:17 | INFO | fairseq.models.hubert.hubert | HubertModel Config: {'_name': 'hubert', 'label_rate': 50.0, 'extractor_mode': default, 'encoder_layers': 12, 'encoder_embed_dim': 768, 'encoder_ffn_embed_dim': 3072, 'encoder_attention_heads': 12, 'activation_fn': gelu, 'layer_type': transformer, 'dropout': 0.1, 'attention_dropout': 0.1, 'activation_dropout': 0.0, 'encoder_layerdrop': 0.05, 'dropout_input': 0.1, 'dropout_features': 0.1, 'final_dim': 256, 'untie_final_proj': True, 'layer_norm_first': False, 'conv_feature_layers': '[(512,10,5)] + [(512,3,2)] * 4 + [(512,2,2)] * 2', 'conv_bias': False, 'logit_temp': 0.1, 'target_glu': False, 'feature_grad_mult': 0.1, 'mask_length': 10, 'mask_prob': 0.8, 'mask_selection': static, 'mask_other': 0.0, 'no_mask_overlap': False, 'mask_min_space': 1, 'mask_channel_length': 10, 'mask_channel_prob': 0.0, 'mask_channel_selection': static, 'mask_channel_other': 0.0, 'no_mask_channel_overlap': False, 'mask_channel_min_space': 1, 'conv_pos': 128, 'conv_pos_groups': 16, 'latent_temp': [2.0, 0.5, 0.999995], 'skip_masked': False, 'skip_nomask': False, 'checkpoint_activations': False, 'required_seq_len_multiple': 2, 'depthwise_conv_kernel_size': 31, 'attn_type': '', 'pos_enc_type': 'abs', 'fp16': False} 2024-05-15 01:54:20 | INFO | rvc.infer.modules.vc.pipeline | Loading rmvpe model,rvc/assets/rmvpe/rmvpe.pt

it gets stuck here with 100% cpu usage and seems to be stuck here forever.

i still have vram left over so it doesnt seem to be a vram issue.

when i pass shorter audio 2 seconds vs 20 clip it seems to work.

[ic1149]

if python 3.10 is faster and working then use 3.10

[lobsterchan27]

thank you.

[ZeaMays14142]

Same problem. Any way to solve this?

[ZeaMays14142]

Wouldnt python 3.10 still have the https://github.com/facebookresearch/fairseq/issues/5012 problem?

if python 3.10 is faster and working then use 3.10

[lobsterchan27]

yes i solved this issue. i found the issue it was related to faiss-cpu version 1.7.3 to 1.8 caused issues for some reason try using the 1.7.3 version and see if it helps.

[ZeaMays14142]

yes i solved this issue. i found the issue it was related to faiss-cpu version 1.7.3 to 1.8 caused issues for some reason try using the 1.7.3 version and see if it helps.

i tried uninstalling faiss-cpu and pip installing the 1.7.3 version as you suggested, but i still got the same behavior, instantly a list of "f0ing xxx now"

then

Loading rmvpe model Loading rmvpe model Loading rmvpe model Loading rmvpe model Loading rmvpe model Loading rmvpe model Loading rmvpe model Loading rmvpe model Loading rmvpe model Loading rmvpe model

and it is stuck

[lobsterchan27]

sorry i have no idea. mine was also stuck there.

[meisa233]

Same here. After I upgrade my gradio to 3.50.2, it worked.