"GeForce RTX 3090 doesnt response, terminating it."

[GuusDeKroon]

Extraction on the 30x build either is very slow, or doesn't work at all. The slowness also occurs with merging, but the "doesn't respond" error is exclusive to extraction.

[alexandrulita91]

I have the same issue with a GeForce GTX 1050 Ti on extraction using DeepFaceLab_NVIDIA_build_12_22_2020, here is the log:

Choose one or several GPU idxs (separated by comma).

[CPU] : CPU
  [0] : GeForce GTX 1050 Ti

[0] Which GPU indexes to choose? :
0

[wf] Face type ( f/wf/head ?:help ) :
wf
[512] Image size ( 256-2048 ?:help ) :
512
[90] Jpeg quality ( 1-100 ?:help ) :
90
Extracting faces...
Running on GeForce GTX 1050 Ti
 55%|###########################################6                                    | 480/879 [06:45<05:36,  1.18it/s]
GeForce GTX 1050 Ti doesnt response, terminating it.
 55%|###########################################6                                    | 480/879 [08:45<07:16,  1.09s/it]
-------------------------
Images found:        879
Faces detected:      480
-------------------------
Done.

[seaniedan]

Same issue on RTX 5000: Extracting faces... Running on Quadro RTX 5000 0%| | 0/10867 [00:00<?, ?it/s]Quadro RTX 5000 doesnt response, terminating it. 0%| | 0/10867 [02:00<?, ?it/s]

Images found: 10867 Faces detected: 0

One workaround is to extract as CPU.

[KernelA]

I have the same issue with RTX 2080. NVIDIA CUDA 11.0.3 cuDNN 8.

[aksy1999]

I have same issue with Nvidia A40. For the time, I used CPU. Anyone tried debugging it? Would appreciate any inputs.

[RadionovM]

Just increase this option self.no_response_time_sec in core/joblib/SubprocessorBase.py

[VincentVanNF]

Dose not work @RadionovM