OpenCL not working

Software darktable
1

DT 5.4.1. Upgraded a couple of days ago to the nvidia 590 driver. Now, it takes forever (27sec) to export a 1000px images, it used to be less than 2sec.

It passes the darktable-cltest but I get the
I have 'NVIDIA CUDA NVIDIA GeForce RTX 4060 Laptop GPU' id=0: NOT utilized

$ darktable-cltest 
darktable 5.4.1
Copyright (C) 2012-2026 Johannes Hanika and other contributors.

Compile options:
  Bit depth              -> 64 bit
  Exiv2                  -> 0.28.7
  Lensfun                -> 0.3.4
  Debug                  -> DISABLED
  SSE2 optimizations     -> ENABLED
  OpenMP                 -> ENABLED
  OpenCL                 -> ENABLED
  Lua                    -> ENABLED  - API version 9.6.0
  Colord                 -> ENABLED
  gPhoto2                -> ENABLED
  OSMGpsMap              -> ENABLED  - map view is available
  GMIC                   -> ENABLED  - Compressed LUTs are supported
  GraphicsMagick         -> ENABLED
  ImageMagick            -> DISABLED
  libavif                -> ENABLED
  libheif                -> ENABLED
  libjxl                 -> ENABLED
  LibRaw                 -> ENABLED  - Version 0.22.0-Release
  OpenJPEG               -> ENABLED
  OpenEXR                -> ENABLED
  WebP                   -> ENABLED

See https://www.darktable.org/resources/ for detailed documentation.
See https://github.com/darktable-org/darktable/issues/new/choose to report bugs.

     0.3641 [opencl_init] opencl disabled via darktable preferences
     0.3642 [opencl_init] opencl library 'libOpenCL' found on your system and loaded, preference 'default path'
     0.4982 [opencl_init] found 1 platform
[opencl_init] found 1 device

[dt_opencl_device_init]
   DEVICE:                   0: 'NVIDIA GeForce RTX 4060 Laptop GPU'
   CONF KEY:                 cldevice_v5_nvidiacudanvidiageforcertx4060laptopgpu
   PLATFORM, VENDOR & ID:    NVIDIA CUDA, NVIDIA Corporation, ID=4318
   CANONICAL NAME:           nvidiacudanvidiageforcertx4060laptopgpu
   DRIVER VERSION:           590.48.01
   DEVICE VERSION:           OpenCL 3.0 CUDA, SM_20 SUPPORT
   DEVICE_TYPE:              GPU, dedicated mem
   GLOBAL MEM SIZE:          7808 MB
   MAX MEM ALLOC:            1952 MB
   MAX IMAGE SIZE:           32768 x 32768
   MAX CONSTANT BUFFER:      64 KB
   ADDRESS ALIGN:            512
   COMPUTE UNITS:            24
   MAX WORK GROUP SIZE:      1024
   MAX WORK ITEM DIMENSIONS: 3
   MAX WORK ITEM SIZES:      [ 1024 1024 64 ]
   ASYNC PIXELPIPE:          NO
   PINNED MEMORY TRANSFER:   NO
   AVOID ATOMICS:            NO
   MICRO NAP:                250
   ROUNDUP WIDTH & HEIGHT    16x16
   CHECK EVENT HANDLES:      128
   TILING ADVANTAGE:         0.000
   DEFAULT DEVICE:           NO
   KERNEL BUILD DIRECTORY:   /usr/share/darktable/kernels
   KERNEL DIRECTORY:         /home/froggy/.cache/darktable/cached_v5_kernels_for_NVIDIACUDANVIDIAGeForceRTX4060LaptopGPU_5904801
   CL COMPILER OPTION:       -cl-fast-relaxed-math
   CL COMPILER COMMAND:      -w -cl-fast-relaxed-math -DNVIDIA_SM_20=1 -DNVIDIA=1 -I"/usr/share/darktable/kernels"
   CL EXCEPTION:             DT_OPENCL_ONLY_CUDA
   KERNEL LOADING TIME:       0.0129 sec
[opencl_init] OpenCL successfully initialized. internal numbers and names of available devices:
[opencl_init]           0       'NVIDIA CUDA NVIDIA GeForce RTX 4060 Laptop GPU'
     0.5832 [opencl_init] FINALLY: opencl PREFERENCE=OFF is AVAILABLE and NOT ENABLED.
[opencl_update_priorities] these are your device priorities:
[opencl_update_priorities]              image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[opencl_update_priorities]              image   preview export  thumbs  preview2
[opencl_update_priorities]              0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
[opencl_update_priorities] these are your device priorities:
[opencl_update_priorities]              image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[opencl_update_priorities]              image   preview export  thumbs  preview2
[opencl_update_priorities]              0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200

What am I missing or is the problem with the 590 drivers?

Here’s the debug:

>$ darktable -d opencl -d perf
darktable 5.4.1
Copyright (C) 2012-2026 Johannes Hanika and other contributors.

Compile options:
  Bit depth              -> 64 bit
  Exiv2                  -> 0.28.7
  Lensfun                -> 0.3.4
  Debug                  -> DISABLED
  SSE2 optimizations     -> ENABLED
  OpenMP                 -> ENABLED
  OpenCL                 -> ENABLED
  Lua                    -> ENABLED  - API version 9.6.0
  Colord                 -> ENABLED
  gPhoto2                -> ENABLED
  OSMGpsMap              -> ENABLED  - map view is available
  GMIC                   -> ENABLED  - Compressed LUTs are supported
  GraphicsMagick         -> ENABLED
  ImageMagick            -> DISABLED
  libavif                -> ENABLED
  libheif                -> ENABLED
  libjxl                 -> ENABLED
  LibRaw                 -> ENABLED  - Version 0.22.0-Release
  OpenJPEG               -> ENABLED
  OpenEXR                -> ENABLED
  WebP                   -> ENABLED

See https://www.darktable.org/resources/ for detailed documentation.
See https://github.com/darktable-org/darktable/issues/new/choose to report bugs.

     0.0000 [dt starting]
 darktable -d opencl -d perf
     0.2970 [opencl_init] opencl disabled via darktable preferences
     0.2974 [opencl_init] opencl library 'libOpenCL' found on your system and loaded, preference 'default path'
     0.4294 [opencl_init] found 1 platform
[opencl_init] found 1 device

[dt_opencl_device_init]
   DEVICE:                   0: 'NVIDIA GeForce RTX 4060 Laptop GPU'
   CONF KEY:                 cldevice_v5_nvidiacudanvidiageforcertx4060laptopgpu
   PLATFORM, VENDOR & ID:    NVIDIA CUDA, NVIDIA Corporation, ID=4318
   CANONICAL NAME:           nvidiacudanvidiageforcertx4060laptopgpu
   DRIVER VERSION:           590.48.01
   DEVICE VERSION:           OpenCL 3.0 CUDA, SM_20 SUPPORT
   DEVICE_TYPE:              GPU, dedicated mem
   GLOBAL MEM SIZE:          7808 MB
   MAX MEM ALLOC:            1952 MB
   MAX IMAGE SIZE:           32768 x 32768
   MAX CONSTANT BUFFER:      64 KB
   ADDRESS ALIGN:            512
   COMPUTE UNITS:            24
   MAX WORK GROUP SIZE:      1024
   MAX WORK ITEM DIMENSIONS: 3
   MAX WORK ITEM SIZES:      [ 1024 1024 64 ]
   ASYNC PIXELPIPE:          NO
   PINNED MEMORY TRANSFER:   NO
   AVOID ATOMICS:            NO
   MICRO NAP:                250
   ROUNDUP WIDTH & HEIGHT    16x16
   CHECK EVENT HANDLES:      128
   TILING ADVANTAGE:         0.000
   DEFAULT DEVICE:           NO
   KERNEL BUILD DIRECTORY:   /usr/share/darktable/kernels
   KERNEL DIRECTORY:         /home/froggy/.cache/darktable/cached_v5_kernels_for_NVIDIACUDANVIDIAGeForceRTX4060LaptopGPU_5904801
   CL COMPILER OPTION:       -cl-fast-relaxed-math
   CL COMPILER COMMAND:      -w -cl-fast-relaxed-math -DNVIDIA_SM_20=1 -DNVIDIA=1 -I"/usr/share/darktable/kernels"
   CL EXCEPTION:             DT_OPENCL_ONLY_CUDA
   KERNEL LOADING TIME:       0.0132 sec
[opencl_init] OpenCL successfully initialized. internal numbers and names of available devices:
[opencl_init]           0       'NVIDIA CUDA NVIDIA GeForce RTX 4060 Laptop GPU'
     0.5155 [opencl_init] FINALLY: opencl PREFERENCE=OFF is AVAILABLE and NOT ENABLED.
[opencl_update_priorities] these are your device priorities:
[opencl_update_priorities]              image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[opencl_update_priorities]              image   preview export  thumbs  preview2
[opencl_update_priorities]              0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
[opencl_update_priorities] these are your device priorities:
[opencl_update_priorities]              image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[opencl_update_priorities]              image   preview export  thumbs  preview2
[opencl_update_priorities]              0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
     1.2963 [lib_load_module] failed to open `midi': libportmidi.so.2: cannot open shared object file: No such file or directory
     4.9977 [dt_dev_load_raw] loading the image. took 0.080 secs (0.084 CPU)
     5.0390 [export] creating pixelpipe took 0.038 secs (0.066 CPU)
     5.0395 [dev_pixelpipe] took 0.000 secs (0.000 CPU) initing base buffer [export]
     5.0436 [dev_pixelpipe] took 0.004 secs (0.046 CPU) [export] processed `rawprepare' on CPU, blended on CPU
     5.0494 [dev_pixelpipe] took 0.006 secs (0.042 CPU) [export] processed `temperature' on CPU, blended on CPU
     5.0606 [dev_pixelpipe] took 0.011 secs (0.115 CPU) [export] processed `highlights' on CPU, blended on CPU
     5.1729 [dev_pixelpipe] took 0.112 secs (1.300 CPU) [export] processed `cacorrect' on CPU, blended on CPU
     5.2506 [dev_pixelpipe] took 0.078 secs (0.864 CPU) [export] processed `demosaic' on CPU, blended on CPU
     5.4662 [dev_pixelpipe] took 0.216 secs (3.051 CPU) [export] processed `lens' on CPU, blended on CPU
     5.5424 [dev_pixelpipe] took 0.076 secs (1.061 CPU) [export] processed `ashift' on CPU, blended on CPU
     5.5602 [dev_pixelpipe] took 0.018 secs (0.262 CPU) [export] processed `flip' on CPU, blended on CPU
     5.5730 [dev_pixelpipe] took 0.013 secs (0.151 CPU) [export] processed `exposure' on CPU, blended on CPU
     5.6344 [dev_pixelpipe] took 0.061 secs (0.713 CPU) [export] processed `toneequal' on CPU, blended on CPU
     5.6435 [dev_pixelpipe] took 0.009 secs (0.038 CPU) [export] processed `crop' on CPU, blended on CPU
     5.6672 [dev_pixelpipe] took 0.024 secs (0.170 CPU) [export] processed `colorin' on CPU, blended on CPU
     5.6798 [dt_ioppr_transform_image_colorspace] IOP_CS_LAB-->IOP_CS_RGB took 0.013 secs (0.114 CPU) [channelmixerrgb]
     5.7723 [dev_pixelpipe] took 0.105 secs (1.324 CPU) [export] processed `channelmixerrgb' on CPU, blended on CPU
    18.0094 [dev_pixelpipe] took 12.237 secs (192.682 CPU) [export] processed `diffuse' on CPU, blended on CPU
    18.7467 [dev_pixelpipe] took 0.737 secs (10.794 CPU) [export] processed `diffuse.1' on CPU, blended on CPU
    31.3278 [dev_pixelpipe] took 12.581 secs (203.196 CPU) [export] processed `diffuse.2' on CPU, blended on CPU
    31.8447 [dev_pixelpipe] took 0.517 secs (8.526 CPU) [export] processed `colorbalancergb' on CPU, blended on CPU
    32.1881 [dev_pixelpipe] took 0.343 secs (4.465 CPU) [export] processed `agx' on CPU, blended on CPU
    32.2281 [resample_plain] plan 0.001 secs (0.000 CPU) resample 0.039 secs (0.637 CPU)
    32.2281 [dev_pixelpipe] took 0.040 secs (0.637 CPU) [export] processed `finalscale' on CPU, blended on CPU
    32.2289 [dt_ioppr_transform_image_colorspace] IOP_CS_RGB-->IOP_CS_LAB took 0.001 secs (0.005 CPU) [colorout]
    32.2307 [dev_pixelpipe] took 0.003 secs (0.019 CPU) [export] processed `colorout' on CPU, blended on CPU
    32.2307 [dev_process_export] pixel pipeline processing took 27.192 secs (429.462 CPU)
    32.2928 [export_job] exported to `/evo/temp/darktable/darktable_exported/dogs-vancouver-20140101-0337_02.jpg'
 [opencl_summary_statistics] device 'NVIDIA CUDA NVIDIA GeForce RTX 4060 Laptop GPU' id=0: NOT utilized
2

Did you try to turn it on?

3

Just check darktablerc and you are right, it was:

opencl=FALSE

I just changed it to TRUE and now it works. Thanks, but how did it get changed?

Because it was working last week. The only change since last week was upgrading the NVIDIA driver from the 580 serie to the 590 serie, because it solves a few WAYLAND problems.

4

I’m going to assume user error.

You can also enable/disable via preferences.

5

I recently had a weird crash after which some settings were changed (e.g. Dt switched itself to sigmoid workflow). Maybe it was something similar in your case.

6

I don’t remember changing the preferences, but…

In anyway, thanks for taking the time to solve my problem.