Can't use GPU inference after the new update: CachingAllocator option max_split_size_mb too small, must be > 20

[frnksmdlkedjnfr]

UVR 5.5.0 was working fine for me. But today I updated it to 5.5.1, and I wasn't able to perform GPU inference since then. All I get is this error message. And if I ignore the error and proceed to process, the whole application would freeze and force quit.

If I leave the 'GPU Conversion' unchecked, there would be no error.

I have experimented with all three models and all got the same error.

Last Error Received:

Process: VR Architecture

If this error persists, please contact the developers with the error details.

Raw Error Details:

RuntimeError: "CachingAllocator option max_split_size_mb too small, must be > 20"
Traceback Error: "
  File "UVR.py", line 4716, in process_start
  File "separate.py", line 668, in seperate
  File "torch\nn\modules\module.py", line 989, in to
  File "torch\nn\modules\module.py", line 641, in _apply
  File "torch\nn\modules\module.py", line 641, in _apply
  File "torch\nn\modules\module.py", line 641, in _apply
  [Previous line repeated 2 more times]
  File "torch\nn\modules\module.py", line 664, in _apply
  File "torch\nn\modules\module.py", line 987, in convert
  File "torch\cuda\__init__.py", line 229, in _lazy_init
"

Error Time Stamp [2023-04-10 15:10:09]

Full Application Settings:

vr_model: 5_HP-Karaoke-UVR
aggression_setting: 10
window_size: 512
batch_size: Default
crop_size: 256
is_tta: False
is_output_image: False
is_post_process: False
is_high_end_process: False
post_process_threshold: 0.2
vr_voc_inst_secondary_model: No Model Selected
vr_other_secondary_model: No Model Selected
vr_bass_secondary_model: No Model Selected
vr_drums_secondary_model: No Model Selected
vr_is_secondary_model_activate: False
vr_voc_inst_secondary_model_scale: 0.9
vr_other_secondary_model_scale: 0.7
vr_bass_secondary_model_scale: 0.5
vr_drums_secondary_model_scale: 0.5
demucs_model: Choose Model
segment: Default
overlap: 0.25
shifts: 2
chunks_demucs: Auto
margin_demucs: 44100
is_chunk_demucs: False
is_chunk_mdxnet: False
is_primary_stem_only_Demucs: False
is_secondary_stem_only_Demucs: False
is_split_mode: True
is_demucs_combine_stems: True
demucs_voc_inst_secondary_model: No Model Selected
demucs_other_secondary_model: No Model Selected
demucs_bass_secondary_model: No Model Selected
demucs_drums_secondary_model: No Model Selected
demucs_is_secondary_model_activate: False
demucs_voc_inst_secondary_model_scale: 0.9
demucs_other_secondary_model_scale: 0.7
demucs_bass_secondary_model_scale: 0.5
demucs_drums_secondary_model_scale: 0.5
demucs_pre_proc_model: No Model Selected
is_demucs_pre_proc_model_activate: False
is_demucs_pre_proc_model_inst_mix: False
mdx_net_model: UVR-MDX-NET Karaoke
chunks: Auto
margin: 44100
compensate: Auto
is_denoise: False
is_invert_spec: False
is_mixer_mode: False
mdx_batch_size: Default
mdx_voc_inst_secondary_model: No Model Selected
mdx_other_secondary_model: No Model Selected
mdx_bass_secondary_model: No Model Selected
mdx_drums_secondary_model: No Model Selected
mdx_is_secondary_model_activate: False
mdx_voc_inst_secondary_model_scale: 0.9
mdx_other_secondary_model_scale: 0.7
mdx_bass_secondary_model_scale: 0.5
mdx_drums_secondary_model_scale: 0.5
is_save_all_outputs_ensemble: True
is_append_ensemble_name: False
chosen_audio_tool: Manual Ensemble
choose_algorithm: Min Spec
time_stretch_rate: 2.0
pitch_rate: 2.0
is_gpu_conversion: True
is_primary_stem_only: False
is_secondary_stem_only: False
is_testing_audio: False
is_add_model_name: False
is_accept_any_input: False
is_task_complete: False
is_normalization: False
is_create_model_folder: False
mp3_bit_set: 320k
save_format: WAV
wav_type_set: PCM_16
help_hints_var: False
model_sample_mode: False
model_sample_mode_duration: 30
demucs_stems: All Stems

System Info:

NVIDIA System Information report created on: 04/10/2023 15:15:22
System name: ü-BLADE-17

[Display]
Operating System:   Windows 10 Pro for Workstations, 64-bit
DirectX version:    12.0 
GPU processor:      NVIDIA GeForce RTX 3080 Ti Laptop GPU
Driver version:     531.41
Driver Type:        DCH
Direct3D feature level: 12_1
CUDA Cores:     7424 
Max-Q Technologies  Yes
Dynamic Boost       Yes
WhisperMode     Yes
Advanced Optimus    No
Maximum Graphics Power  175 W
Core clock:     1395 MHz 
Memory data rate:   16.00 Gbps
Memory interface:   256-bit 
Memory bandwidth:   512.06 GB/s
Total available graphics memory:    48993 MB
Dedicated video memory: 16384 MB GDDR6
System video memory:    0 MB
Shared system memory:   32609 MB
Video BIOS version: 94.03.19.40.36
IRQ:            Not used
Bus:            PCI Express x8 Gen4
Device ID:      10DE 2420 30051A58
Part Number:        4735 0310

[Components]

nvui.dll        8.17.15.3141        NVIDIA User Experience Driver Component
nvxdplcy.dll        8.17.15.3141        NVIDIA User Experience Driver Component
nvxdbat.dll     8.17.15.3141        NVIDIA User Experience Driver Component
nvxdapix.dll        8.17.15.3141        NVIDIA User Experience Driver Component
NVCPL.DLL       8.17.15.3141        NVIDIA User Experience Driver Component
nvCplUIR.dll        8.1.940.0       NVIDIA Control Panel
nvCplUI.exe     8.1.940.0       NVIDIA Control Panel
nvWSSR.dll      31.0.15.3141        NVIDIA Workstation Server
nvWSS.dll       31.0.15.3141        NVIDIA Workstation Server
nvViTvSR.dll        31.0.15.3141        NVIDIA Video Server
nvViTvS.dll     31.0.15.3141        NVIDIA Video Server
nvLicensingS.dll        6.14.15.3141        NVIDIA Licensing Server
nvDevToolSR.dll     31.0.15.3141        NVIDIA Licensing Server
nvDevToolS.dll      31.0.15.3141        NVIDIA 3D Settings Server
nvDispSR.dll        31.0.15.3141        NVIDIA Display Server
nvDispS.dll     31.0.15.3141        NVIDIA Display Server
PhysX       09.21.0713      NVIDIA PhysX
NVCUDA64.DLL        31.0.15.3141        NVIDIA CUDA 12.1.98 driver
nvGameSR.dll        31.0.15.3141        NVIDIA 3D Settings Server
nvGameS.dll     31.0.15.3141        NVIDIA 3D Settings Server

[Anjok07]

This could be an issue with the filename. Can you try renaming the file using English characters? I just want to rule it out.

[frnksmdlkedjnfr]

No. The app just froze and force closed after a while.

I should mention that I copied the models downloaded from 5.5.0 and pasted them into the 5.5.1 dir. I wonder if that may be problematic.

Also, even with GPU Conversion unchecked, the ensemble mode isn't working properly. It would just process the audio file with the first model and then freeze, the way shown in the picture.

[Anjok07]

This is very unusual. I just tested it on my end using your exact settings and it seems to be running fine for me. I have a few questions:

• Do you have UVR installed on your main C drive?
• Did you update from the patch or the full install?

[frnksmdlkedjnfr]

1. Yes. It IS installed on C and WAS installed on C. It was working fine.
2. I tried both. Clean installed 5.5.1, re-downloaded the models, and got the same error. Dirty update yielded the same error.

I'm not sure if this has something to do with my WSL2, since I've been doing some machine learning experiment in there. Or can it be the graphics driver related?

[frnksmdlkedjnfr]

Okay I found the bane. I set the max_split_size_mb to be 16 in environment variable. Deleted that system variable and the app runs normal.