windows11 cuda12.0 cdnn 8.9.5

[gushenjie]

trafficstars

我在windows11上安装了cuda12.0 cdnn 8.9.5版本,也通过命令验证了是否安装成功

我一开始使用docker compose -f docker-compose.gpu.yaml up -d启动程序 2024-09-19 10:34:57 voice='1031.pt' 2024-09-19 10:34:57 当前使用音色 seed_path='/app/speaker/1031.pt' 2024-09-19 10:34:57 prompt='[break_6]' 2024-09-19 10:34:57 te=['如果我们需要您提供某些信息以帮助您在使用产品时遇到的疑问，以及向您推荐第三方服务供应商，我们会严格遵守隐私政策来处理这些信息。如果您已经开始使用米林空间，则表示您已经接受、并同意我们按照此隐私政策中所描述的方式来处理和使用您的个人信息'] 2024-09-19 10:35:16 code: 0%| | 0/2048(max) [00:00, ?it/s] code: 0%| | 1/2048(max) [00:00, 6.08it/s] code: 0%| | 7/2048(max) [00:00, 29.60it/s] code: 1%| | 13/2048(max) [00:00, 40.77it/s] code: 1%| | 18/2048(max) [00:00, 42.57it/s] code: 1%| | 24/2048(max) [00:00, 45.97it/s] code: 1%|▏ | 29/2048(max) [00:00, 45.43it/s] code: 2%|▏ | 35/2048(max) [00:00, 48.61it/s] code: 2%|▏ | 41/2048(max) [00:00, 50.29it/s] code: 2%|▏ | 47/2048(max) [00:01, 52.38it/s] code: 3%|▎ | 53/2048(max) [00:01, 53.86it/s] code: 3%|▎ | 60/2048(max) [00:01, 56.05it/s] code: 3%|▎ | 66/2048(max) [00:01, 56.92it/s] code: 4%|▎ | 72/2048(max) [00:01, 56.83it/s] code: 4%|▍ | 78/2048(max) [00:01, 57.54it/s] code: 4%|▍ | 84/2048(max) [00:01, 56.57it/s] code: 4%|▍ | 90/2048(max) [00:01, 56.23it/s] code: 5%|▍ | 96/2048(max) [00:01, 56.70it/s] code: 5%|▍ | 102/2048(max) [00:02, 56.34it/s] code: 5%|▌ | 108/2048(max) [00:02, 55.56it/s] code: 6%|▌ | 114/2048(max) [00:02, 55.80it/s] code: 6%|▌ | 121/2048(max) [00:02, 57.57it/s] code: 6%|▌ | 127/2048(max) [00:02, 56.40it/s] code: 6%|▋ | 133/2048(max) [00:02, 56.91it/s] code: 7%|▋ | 140/2048(max) [00:02, 58.28it/s] code: 7%|▋ | 147/2048(max) [00:02, 59.40it/s] code: 7%|▋ | 153/2048(max) [00:02, 59.32it/s] code: 8%|▊ | 160/2048(max) [00:02, 60.18it/s] code: 8%|▊ | 167/2048(max) [00:03, 60.41it/s] code: 8%|▊ | 174/2048(max) [00:03, 56.03it/s] code: 9%|▉ | 181/2048(max) [00:03, 57.31it/s] code: 9%|▉ | 187/2048(max) [00:03, 56.26it/s] code: 9%|▉ | 193/2048(max) [00:03, 52.90it/s] code: 10%|▉ | 199/2048(max) [00:03, 53.94it/s] code: 10%|█ | 205/2048(max) [00:03, 53.86it/s] code: 10%|█ | 211/2048(max) [00:03, 55.02it/s] code: 11%|█ | 217/2048(max) [00:04, 55.52it/s] code: 11%|█ | 223/2048(max) [00:04, 54.78it/s] code: 11%|█ | 229/2048(max) [00:04, 56.21it/s] code: 12%|█▏ | 236/2048(max) [00:04, 58.13it/s] code: 12%|█▏ | 242/2048(max) [00:04, 57.25it/s] code: 12%|█▏ | 249/2048(max) [00:04, 58.97it/s] code: 12%|█▎ | 256/2048(max) [00:04, 60.23it/s] code: 13%|█▎ | 263/2048(max) [00:04, 59.41it/s] code: 13%|█▎ | 270/2048(max) [00:04, 60.76it/s] code: 14%|█▎ | 277/2048(max) [00:05, 61.26it/s] code: 14%|█▍ | 284/2048(max) [00:05, 58.80it/s] code: 14%|█▍ | 290/2048(max) [00:05, 58.67it/s] code: 14%|█▍ | 296/2048(max) [00:05, 57.72it/s] code: 15%|█▍ | 302/2048(max) [00:05, 56.15it/s] code: 15%|█▌ | 308/2048(max) [00:05, 56.43it/s] code: 15%|█▌ | 314/2048(max) [00:05, 57.15it/s] code: 16%|█▌ | 320/2048(max) [00:05, 57.75it/s] code: 16%|█▌ | 327/2048(max) [00:05, 58.89it/s] code: 16%|█▋ | 333/2048(max) [00:06, 53.50it/s] code: 17%|█▋ | 339/2048(max) [00:06, 54.78it/s] code: 17%|█▋ | 346/2048(max) [00:06, 57.77it/s] code: 17%|█▋ | 352/2048(max) [00:06, 57.61it/s] code: 17%|█▋ | 358/2048(max) [00:06, 56.51it/s] code: 18%|█▊ | 365/2048(max) [00:06, 58.04it/s] code: 18%|█▊ | 371/2048(max) [00:06, 53.99it/s] code: 18%|█▊ | 377/2048(max) [00:06, 55.22it/s] code: 19%|█▊ | 383/2048(max) [00:06, 56.36it/s] code: 19%|█▉ | 389/2048(max) [00:07, 57.10it/s] code: 19%|█▉ | 395/2048(max) [00:07, 57.60it/s] code: 20%|█▉ | 401/2048(max) [00:07, 58.16it/s] code: 20%|█▉ | 408/2048(max) [00:07, 59.43it/s] code: 20%|██ | 414/2048(max) [00:07, 58.85it/s] code: 21%|██ | 420/2048(max) [00:07, 58.46it/s] code: 21%|██ | 427/2048(max) [00:07, 59.30it/s] code: 21%|██ | 434/2048(max) [00:07, 60.24it/s] code: 22%|██▏ | 441/2048(max) [00:07, 61.35it/s] code: 22%|██▏ | 448/2048(max) [00:07, 61.88it/s] code: 22%|██▏ | 455/2048(max) [00:08, 61.71it/s] code: 23%|██▎ | 462/2048(max) [00:08, 56.65it/s] code: 23%|██▎ | 468/2048(max) [00:08, 57.37it/s] code: 23%|██▎ | 474/2048(max) [00:08, 54.22it/s] code: 23%|██▎ | 480/2048(max) [00:08, 55.15it/s] code: 24%|██▎ | 486/2048(max) [00:08, 53.66it/s] code: 24%|██▍ | 492/2048(max) [00:08, 55.10it/s] code: 24%|██▍ | 499/2048(max) [00:08, 56.84it/s] code: 25%|██▍ | 506/2048(max) [00:09, 58.34it/s] code: 25%|██▌ | 513/2048(max) [00:09, 59.50it/s] code: 25%|██▌ | 520/2048(max) [00:09, 60.27it/s] code: 26%|██▌ | 527/2048(max) [00:09, 61.20it/s] code: 26%|██▌ | 534/2048(max) [00:09, 59.24it/s] code: 26%|██▋ | 541/2048(max) [00:09, 60.30it/s] code: 27%|██▋ | 548/2048(max) [00:09, 61.45it/s] code: 27%|██▋ | 555/2048(max) [00:09, 61.34it/s] code: 27%|██▋ | 562/2048(max) [00:09, 61.00it/s] code: 28%|██▊ | 569/2048(max) [00:10, 53.05it/s] code: 28%|██▊ | 575/2048(max) [00:10, 53.66it/s] code: 28%|██▊ | 581/2048(max) [00:10, 53.50it/s] code: 29%|██▊ | 587/2048(max) [00:10, 51.99it/s] code: 29%|██▉ | 593/2048(max) [00:10, 52.87it/s] code: 29%|██▉ | 599/2048(max) [00:10, 54.71it/s] code: 30%|██▉ | 605/2048(max) [00:10, 55.60it/s] code: 30%|██▉ | 611/2048(max) [00:10, 56.55it/s] code: 30%|███ | 617/2048(max) [00:10, 57.49it/s] code: 30%|███ | 623/2048(max) [00:11, 56.99it/s] code: 31%|███ | 629/2048(max) [00:11, 57.52it/s] code: 31%|███ | 635/2048(max) [00:11, 54.98it/s] code: 31%|███▏ | 641/2048(max) [00:11, 52.56it/s] code: 32%|███▏ | 647/2048(max) [00:11, 49.02it/s] code: 32%|███▏ | 652/2048(max) [00:11, 48.85it/s] code: 32%|███▏ | 658/2048(max) [00:11, 50.08it/s] code: 32%|███▏ | 664/2048(max) [00:11, 50.67it/s] code: 33%|███▎ | 670/2048(max) [00:12, 50.05it/s] code: 33%|███▎ | 676/2048(max) [00:12, 51.17it/s] code: 33%|███▎ | 682/2048(max) [00:12, 51.59it/s] code: 34%|███▎ | 688/2048(max) [00:12, 51.35it/s] code: 34%|███▍ | 694/2048(max) [00:12, 50.39it/s] code: 34%|███▍ | 700/2048(max) [00:12, 51.81it/s] code: 34%|███▍ | 706/2048(max) [00:12, 52.50it/s] code: 35%|███▍ | 712/2048(max) [00:12, 45.68it/s] code: 35%|███▌ | 717/2048(max) [00:13, 45.05it/s] code: 35%|███▌ | 722/2048(max) [00:13, 46.09it/s] code: 36%|███▌ | 728/2048(max) [00:13, 48.41it/s] code: 36%|███▌ | 734/2048(max) [00:13, 49.94it/s] code: 36%|███▌ | 740/2048(max) [00:13, 49.29it/s] code: 36%|███▋ | 746/2048(max) [00:13, 49.88it/s] code: 37%|███▋ | 752/2048(max) [00:13, 52.18it/s] code: 37%|███▋ | 758/2048(max) [00:13, 53.51it/s] code: 37%|███▋ | 765/2048(max) [00:13, 55.75it/s] code: 38%|███▊ | 772/2048(max) [00:14, 57.23it/s] code: 38%|███▊ | 778/2048(max) [00:14, 57.73it/s] code: 38%|███▊ | 784/2048(max) [00:14, 53.40it/s] code: 39%|███▊ | 790/2048(max) [00:14, 50.77it/s] code: 39%|███▉ | 796/2048(max) [00:14, 48.76it/s] code: 39%|███▉ | 802/2048(max) [00:14, 49.85it/s] code: 39%|███▉ | 808/2048(max) [00:14, 46.45it/s] code: 40%|███▉ | 814/2048(max) [00:14, 49.48it/s] code: 40%|████ | 820/2048(max) [00:14, 51.93it/s] code: 40%|████ | 826/2048(max) [00:15, 54.04it/s] code: 41%|████ | 832/2048(max) [00:15, 55.30it/s] code: 41%|████ | 838/2048(max) [00:15, 56.35it/s] code: 41%|████ | 844/2048(max) [00:15, 49.11it/s] code: 42%|████▏ | 850/2048(max) [00:15, 49.93it/s] code: 42%|████▏ | 856/2048(max) [00:15, 51.19it/s] code: 42%|████▏ | 862/2048(max) [00:15, 52.28it/s] code: 42%|████▏ | 868/2048(max) [00:15, 53.93it/s] code: 43%|████▎ | 874/2048(max) [00:15, 53.83it/s] code: 43%|████▎ | 880/2048(max) [00:16, 50.00it/s] code: 43%|████▎ | 886/2048(max) [00:16, 51.49it/s] code: 44%|████▎ | 892/2048(max) [00:16, 52.64it/s] code: 44%|████▍ | 898/2048(max) [00:16, 50.67it/s] code: 44%|████▍ | 904/2048(max) [00:16, 47.56it/s] code: 44%|████▍ | 910/2048(max) [00:16, 49.86it/s] code: 45%|████▍ | 916/2048(max) [00:16, 52.33it/s] code: 45%|████▌ | 922/2048(max) [00:16, 50.59it/s] code: 45%|████▌ | 928/2048(max) [00:17, 51.60it/s] code: 46%|████▌ | 934/2048(max) [00:17, 53.22it/s] code: 46%|████▌ | 940/2048(max) [00:17, 49.52it/s] code: 46%|████▌ | 946/2048(max) [00:17, 50.42it/s] code: 46%|████▋ | 952/2048(max) [00:17, 49.32it/s] code: 47%|████▋ | 957/2048(max) [00:17, 49.03it/s] code: 47%|████▋ | 962/2048(max) [00:17, 43.23it/s] code: 47%|████▋ | 967/2048(max) [00:17, 44.90it/s] code: 48%|████▊ | 974/2048(max) [00:18, 49.44it/s] code: 48%|████▊ | 981/2048(max) [00:18, 52.75it/s] code: 48%|████▊ | 988/2048(max) [00:18, 55.07it/s] code: 49%|████▊ | 994/2048(max) [00:18, 55.95it/s] code: 49%|████▉ | 1000/2048(max) [00:18, 52.68it/s] code: 49%|████▉ | 1006/2048(max) [00:18, 54.11it/s] code: 49%|████▉ | 1012/2048(max) [00:18, 55.44it/s] code: 50%|████▉ | 1018/2048(max) [00:18, 56.47it/s] code: 50%|████▉ | 1020/2048(max) [00:18, 54.06it/s] 2024-09-19 10:35:16 /home/venv/lib/python3.9/site-packages/torch/nn/modules/conv.py:306: UserWarning: Plan failed with a cudnnException: CUDNN_BACKEND_EXECUTION_PLAN_DESCRIPTOR: cudnnFinalize Descriptor Failed cudnn_status: CUDNN_STATUS_NOT_SUPPORTED (Triggered internally at ../aten/src/ATen/native/cudnn/Conv_v8.cpp:919.) 2024-09-19 10:35:16 return F.conv1d(input, weight, bias, self.stride, 2024-09-19 10:35:16 推理时长: 18.91 秒 � � 大概信息如上,好像没用到cuda加速

接着我又使用源代码部署方式,也执行了pip install torch==2.2.0 torchaudio==2.2.0 --index-url https://download.pytorch.org/whl/cu118,运行同样的操作和文本内容,以下是我的详细信息,还没有之前docker运行的速度快？

(venv) D:\jackgu\ai\chatTTS-ui>python app.py Starting... {'sha256_asset_Decoder_pt': '9964e36e840f0e3a748c5f716fe6de6490d2135a5f5155f4a642d51860e2ec38', 'sha256_asset_DVAE_full_pt': '553eb75763511e23f3e5f86303e2163c5ca775489d637fb635d979c8ae58bbe5', 'sha256_asset_GPT_pt': 'd7d4ee6461ea097a2be23eb40d73fb94ad3b3d39cb64fbb50cb3357fd466cadb', 'sha256_asset_spk_stat_pt': '3228d8a4cbbf349d107a1b76d2f47820865bd3c9928c4bdfe1cefd5c7071105f', 'sha256_asset_tokenizer_pt': 'e911ae7c6a7c27953433f35c44227a67838fe229a1f428503bdb6cd3d1bcc69c', 'sha256_asset_Vocos_pt': '09a670eda1c08b740013679c7a90ebb7f1a97646ea7673069a6838e6b51d6c58'} Start:0.0.0.0:9966 voice='1031.pt' 当前使用音色 seed_path='D:/jackgu/ai/chatTTS-ui/speaker/1031.pt' prompt='[break_6]' te=['如果我们需要您提供某些信息以帮助您在使用产品时遇到的疑问，以及向您推荐第三方服务供应商，我们会严格遵守隐私政策来处理这些信息。如果您已经开始使用米林空间，则表示您已经接受、并同意我们按照此隐私政策中所描述的方式来处理和使用您的个人信息'] code: 0%| | 0/2048(max) [00:00, ?it/s]We detected that you are passing past_key_values as a tuple and this is deprecated and will be removed in v4.43. Please use an appropriate Cache class (https://huggingface.co/docs/transformers/v4.41.3/en/internal/generation_utils#transformers.Cache) D:\jackgu\ai\chatTTS-ui\venv\Lib\site-packages\transformers\models\llama\modeling_llama.py:660: UserWarning: 1Torch was not compiled with flash attention. (Triggered internally at ..\aten\src\ATen\native\transformers\cuda\sdp_utils.cpp:263.) attn_output = torch.nn.functional.scaled_dot_product_attention( code: 50%|████████████████████████████████████████████████████████████████████████████████▋ | 1020/2048(max) [00:24, 41.10it/s] 推理时长: 26.26 秒�