search

darktable-org / darktable

darktable is an open source photography workflow application and raw developer
https://www.darktable.org
GNU General Public License v3.0
9.49k stars 1.12k forks source link

FYI rocm bugreport "darktable crashing on rocm 5.7.1 __pthread_kill_implementation" #16641

Closed Germano0 closed 2 months ago

Germano0 commented 4 months ago

Describe the bug

I am posting this in case if you would like to keep track of this rocm bug triggered by darktable https://github.com/ROCm/ROCm/issues/3039

Steps to reproduce

N/A

Expected behavior

No response

Logfile | Screenshot | Screencast

No response

Commit

No response

Where did you obtain darktable from?

distro packaging

darktable version

4.6.1

What OS are you using?

Linux

What is the version of your OS?

Fedora 39

Describe your system?

No response

Are you using OpenCL GPU in darktable?

None

If yes, what is the GPU card and driver?

No response

Please provide additional context if applicable. You can attach files too, but might need to rename to .txt or .zip

No response

jenshannoschwalm commented 4 months ago

I you want assistance here we need the log from '-d opencl -d pipe'

Germano0 commented 4 months ago 
$ darktable -d opencl -d pipe
darktable 4.6.1
Copyright (C) 2012-2024 Johannes Hanika and other contributors.

Compile options:
  Bit depth              -> 64 bit
  Debug                  -> DISABLED
  SSE2 optimizations     -> ENABLED
  OpenMP                 -> ENABLED
  OpenCL                 -> ENABLED
  Lua                    -> ENABLED  - API version 9.2.0
  Colord                 -> ENABLED
  gPhoto2                -> ENABLED
  GMIC                   -> ENABLED  - Compressed LUTs are supported
  GraphicsMagick         -> ENABLED
  ImageMagick            -> DISABLED
  libavif                -> ENABLED
  libheif                -> ENABLED
  libjxl                 -> ENABLED
  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.

     0.0381 [dt_get_sysresource_level] switched to 1 as `default'
     0.0381   total mem:       30845MB
     0.0381   mipmap cache:    3855MB
     0.0381   available mem:   15422MB
     0.0381   singlebuff:      240MB
     0.0521 [opencl_init] opencl library 'libOpenCL' found on your system and loaded, preference 'default path'
     0.1082 [opencl_init] found 1 platform
[opencl_init] found 1 device

[dt_opencl_device_init]
   DEVICE:                   0: 'gfx1035'
   PLATFORM, VENDOR & ID:    AMD Accelerated Parallel Processing, Advanced Micro Devices, Inc., ID=4098
   CANONICAL NAME:           amdacceleratedparallelprocessinggfx1035
   DRIVER VERSION:           3590.0 (HSA1.1,LC)
   DEVICE VERSION:           OpenCL 2.0 
   DEVICE_TYPE:              GPU, dedicated mem
   GLOBAL MEM SIZE:          1024 MB
   MAX MEM ALLOC:            870 MB
   MAX IMAGE SIZE:           16384 x 16384
   MAX WORK GROUP SIZE:      256
   MAX WORK ITEM DIMENSIONS: 3
   MAX WORK ITEM SIZES:      [ 1024 1024 1024 ]
   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/user/.cache/darktable/cached_v3_kernels_for_AMDAcceleratedParallelProcessinggfx1035_35900HSA11LC
   CL COMPILER OPTION:       -cl-fast-relaxed-math
   CL COMPILER COMMAND:      -w -cl-fast-relaxed-math  -DAMD=1 -I"/usr/share/darktable/kernels"
   KERNEL LOADING TIME:       0.0252 sec
[opencl_init] OpenCL successfully initialized. internal numbers and names of available devices:
[opencl_init]           0       'AMD Accelerated Parallel Processing gfx1035'
     0.3863 [opencl_init] FINALLY: opencl is AVAILABLE and ENABLED.
[opencl_init] opencl_scheduling_profile: 'default'
[opencl_init] opencl_device_priority: '*/!0,*/*/*/!0,*'
[opencl_init] opencl_mandatory_timeout: 400
[dt_opencl_update_priorities] these are your device priorities:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[dt_opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
[dt_opencl_update_priorities] these are your device priorities:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[dt_opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
    21.1897 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0429s, history 0.0108s
    21.1898 pixelpipe starting CL      [export]                                (   0/   0) 6000x4000 scale=1.0000 --> (   0/   0) 6000x4000 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    21.1898 [dt_opencl_check_tuning] use 289MB (headroom=OFF, pinning=OFF) on device `AMD Accelerated Parallel Processing gfx1035' id=0
    21.1898 pixelpipe data: full       [export]                                (   0/   0) 6000x4000 scale=1.0000 --> (   0/   0) 6000x4000 scale=1.0000 
    21.8466 pixelpipe process CPU      [export]         gamma                  (   0/   0) 6000x4000 scale=1.0000 --> (   0/   0) 6000x4000 scale=1.0000 IOP_CS_RGB
    21.9147 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 732MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    21.9148 pixelpipe finished         [export]                                (   0/   0) 6000x4000 scale=1.0000 --> (   0/   0) 6000x4000 scale=1.0000 

    22.3664 [export_job] exported to `/home/user/img/cm/darktable_exported/IMG_7633.jpg'
    22.5397 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0326s, history 0.0070s
    22.5397 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    22.5398 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    22.5398 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    22.5478 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    22.5538 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    22.5574 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    22.5660 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: 0.0000, blue: 0.0000 for hash=ebe51e57ef99313b
    23.8387 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    23.8812 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    23.8812 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    24.2362 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4656.jpg'
    24.4025 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0345s, history 0.0077s
    24.4025 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    24.4026 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    24.4026 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    24.4105 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    24.4161 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    24.4240 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    24.4319 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: 0.0000, blue: 0.0000 for hash=ebe51e57ef99313b
    25.6452 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    25.6831 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    25.6831 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    26.0003 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4657.jpg'
    26.1627 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0328s, history 0.0072s
    26.1628 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    26.1628 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    26.1628 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    26.1705 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    26.1762 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    26.1850 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    26.1936 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: 0.0000, blue: 0.0000 for hash=ebe51e57ef99313b
    27.4670 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    27.5050 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    27.5050 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    27.8388 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4658.jpg'
    28.0135 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0407s, history 0.0080s
    28.0136 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    28.0136 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    28.0136 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    28.0226 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    28.0282 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    28.0373 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    28.0606 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: 0.0000, blue: 0.0000 for hash=ebe51e57ef99313b
    29.6930 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    29.7307 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    29.7307 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    30.0548 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4659.jpg'
    30.2227 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0342s, history 0.0074s
    30.2227 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    30.2227 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    30.2227 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    30.2313 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    30.2374 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    30.2459 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    30.2527 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: 0.0000, blue: 0.0000 for hash=ebe51e57ef99313b
    31.7447 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    31.7813 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    31.7814 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    32.1236 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4660.jpg'
    32.2899 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0328s, history 0.0075s
    32.2900 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    32.2900 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    32.2900 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    32.2977 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    32.3034 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    32.3124 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    32.3363 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: 0.0000, blue: 0.0000 for hash=ebe51e57ef99313b
    33.6306 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    33.6740 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    33.6740 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    34.0070 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4661.jpg'
    34.1734 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0352s, history 0.0078s
    34.1734 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    34.1734 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    34.1734 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    34.1824 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    34.1882 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    34.1969 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    34.2043 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: -0.0115, blue: 0.0000 for hash=ebe51e57ef99313b
    35.7676 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    35.8035 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    35.8035 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    36.1922 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4662.jpg'
    36.3507 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0338s, history 0.0072s
    36.3507 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    36.3508 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    36.3508 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    36.3594 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    36.3658 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    36.3737 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    36.3875 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: -0.0063, blue: 0.0000 for hash=ebe51e57ef99313b
    37.7508 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    37.7934 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    37.7934 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

[dt_opencl_update_priorities] these are your device priorities:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[dt_opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
[dt_opencl_update_priorities] these are your device priorities:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[dt_opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
[dt_opencl_update_priorities] these are your device priorities:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[dt_opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
[dt_opencl_update_priorities] these are your device priorities:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[dt_opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
[dt_opencl_update_priorities] these are your device priorities:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       -1      0       0       -1
[dt_opencl_update_priorities] show if opencl use is mandatory for a given pixelpipe:
[dt_opencl_update_priorities]           image   preview export  thumbs  preview2
[dt_opencl_update_priorities]           0       0       0       0       0
[opencl_synchronization_timeout] synchronization timeout set to 200
    38.3367 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4663.jpg'
    38.5357 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0411s, history 0.0093s
    38.5357 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    38.5358 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    38.5358 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    38.5445 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    38.5502 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    38.5591 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    38.5996 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: -0.0068, blue: 0.0000 for hash=ebe51e57ef99313b
    39.9624 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    39.9985 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    39.9985 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    40.3756 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4664.jpg'
    40.5537 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0383s, history 0.0088s
    40.5537 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    40.5537 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    40.5538 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    40.5617 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    40.5675 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    40.5754 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    40.6060 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: -0.0136, blue: 0.0000 for hash=ebe51e57ef99313b
    42.1582 pixelpipe process CPU      [export]         gamma                  (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RGB
    42.1970 cache report               [export]                                2 lines (important=0, used=0, invalid=0). Using 377MB, limit=0MB. Hits/run=0.00. Hits/test=0.000
    42.1971 pixelpipe finished         [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 

    42.6114 [export_job] exported to `/home/user/img/cm/darktable_exported/DSCF4665.jpg'
    42.8172 dt_dev_pixelpipe_synch_all [export]                                defaults 0.0424s, history 0.0096s
    42.8172 pixelpipe starting CL      [export]                                (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 device=0 (amdacceleratedparallelprocessinggfx1035)
    42.8173 modify roi IN              [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 
    42.8173 pixelpipe data: full       [export]                                (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4096x3012 scale=1.0000 
    42.8270 pixelpipe process CL       [export]         rawprepare             (   0/   0) 4096x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    42.8326 pixelpipe process CL       [export]         temperature            (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    42.8362 pixelpipe process CL       [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 IOP_CS_RAW
    42.8444 opposed chroma CL          [export]         highlights             (   0/   0) 4032x3012 scale=1.0000 --> (   0/   0) 4032x3012 scale=1.0000 red: 0.0000, green: -0.0145, blue: 0.0000 for hash=ebe51e57ef99313b
Memory access fault by GPU node-1 (Agent handle: 0x7fe1040f1a40) on address 0x7fe10a615000. Reason: Page not present or supervisor privilege.
Annullato (core dump creato)
jenshannoschwalm commented 4 months ago

You tried with resource setting to small? Your cl device memory is pretty small so if you run other cl using apps like firefox or libreoffice you might be in trouble? Could you possibly check with current master?

Germano0 commented 4 months ago

It crashes also with darktable resources: small

gi-man commented 4 months ago

@Germano0 Did you try with the current rocm? I think the current is 6.1.0 vs 5.7.1. I know Karol Herbst fixed a few issues with rocm had with memory (eg. https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28122)

Also, is this a iGPU or a real GPU card?

piratenpanda commented 4 months ago

rocm is not rusticl :) gfx1035 is the R680 https://www.techpowerup.com/gpu-specs/radeon-680m.c3871

jenshannoschwalm commented 4 months ago
  1. It's difficult to tell from here but the small graphics mem might very well be evident even with "small" resources setting as a) the kernel code b) gnome desktop and c) apps like firefox - they all eat cl memory. I checked the code code again for memleaks, none found. Maybe we find something with -d opencl -d memory. Can you somehow monitor your graphics memory?
  2. Have you always found dt crashing with "inpaint opposed"? That might hint to something specific with your driver.
  3. rocm opencl for gfx1035 might have more problems, see https://github.com/ROCm/ROCm/issues/2048
  4. Unfortunately we have had so many issues with amd drivers depending on version and hardware, simply not possible to keep tracking on them if the AMD people themselves cant do neither.
Germano0 commented 4 months ago

Hello, concerning 6.1.0, I am waiting for it to be released in Fedora. For the moment I can provide a newer trace made with rocm 6.0. output of darktable -d opencl -d pipe -d memory darktable_gdb_rocm_6.txt

Concerning GPU RAM usage, let me attach two screenshots of radeontop:

darktable closed (after a darktable crash), Firefox running, Plasma 6.0.3 running

radeontop_idle

darktable exporting, Firefox running, Plasma 6.0.3 running

radeontop_darktable_exporting

gi-man commented 4 months ago

Hanno, I had some time to look at the latest log output. The drivers still say: dedicated mem GLOBAL MEM SIZE: 1024 MB. I dont think this is correct since I think it shares the system memory. To me the drivers are still an issue.

Regardless this part was odd to me: 44.2717 [opencl memory] device 0: 1505903796 bytes (1436.1 MB) in use Is it trying to use more memory than available?

jenshannoschwalm commented 4 months ago

Hanno, I had some time to look at the latest log output. The drivers still say: dedicated mem GLOBAL MEM SIZE: 1024 MB. I dont think this is correct since I think it shares the system memory. To me the drivers are still an issue.

I will have another look at the way we test for that but i think it's correct and suspect the driver to be reporting wrongfully. Anyway, that shouldn't matter how the code works :-)

Regardless this part was odd to me: 44.2717 [opencl memory] device 0: 1505903796 bytes (1436.1 MB) in use Is it trying to use more memory than available?

Whow - congrats - propably a really good catch! That would indeed very likely mean "problems ahead"! Difficult to spot on larger cards even while using the simulation modes, we don't have a test for that yet.

Why don't you go for it? A debugging path would be

  1. Isolate the module that's taking too much
  2. check there for proper calculation of required cl memory in tiling_callback() - we had quite a number of wrong-doings here
  3. or possibly some blending cl requirements
  4. A big fat warning about "took more cl memory than allowed" would be very good!
github-actions[bot] commented 2 months ago

This issue has been marked as stale due to inactivity for the last 60 days. It will be automatically closed in 300 days if no update occurs. Please check if the master branch has fixed it and report again or close the issue.

Germano0 commented 2 months ago

no longer experiencing this bug on 4.6.1 and rocm 6.1.0. @gi-man concerning ~1000 MB GPU RAM, I booted Windows and I saw that it's the same RAM amount gpuz

But in Linux I noticed that there are two values: VRAM and GTT. The latter should be the amount of CPU RAM shared with the GPU. I don't know why VRAM is setted to a fixed value of ~1000 MB. On Linux I tried to run a heavy videogame and the VRAM used amount sometimes exceeded the maximum VRAM value. I suppose that GTT is a kind of compensation tank for VRAM...

after