Problem with Batch Queue multi core processing

[Robert Zeller]

Thank you very much: the prerelease 5.8 works on my system as it should
! All six cores running parallel delivering perfect jpegs !


On 24.09.2017 08:33, Gilles Caulier wrote:
> You can test the pre release bundle available here :
>
> https://drive.google.com/drive/folders/0BzeiVr-byqt5Y0tIRWVWelRJenM
>
>
>
> 2017-09-23 23:30 GMT+02:00 Robert Zeller
> <robert.zeller at robert-zeller.org
> <mailto:robert.zeller at robert-zeller.org>>:
>
>     Thanks for the information, Maik! So I am waiting for DK 5.8
>
>     Robert
>
>
>     On 23.09.2017 23:25, Maik Qualmann wrote:
>>     This problem is fixed in digiKam-5.8.0. The thread support for libraw was not 
>>     activated, it only affects NEF files.
>>
>>     See here:
>>
>>     http://digikam.1695700.n4.nabble.com/Batch-RAW-Conversion-to-jpg-leads-to-unusable-jpg-tiff-when-using-all-kernels-td4703791.html
>>     <http://digikam.1695700.n4.nabble.com/Batch-RAW-Conversion-to-jpg-leads-to-unusable-jpg-tiff-when-using-all-kernels-td4703791.html>
>>
>>     Maik
>>
>>     Am Samstag, 23. September 2017, 22:44:21 CEST schrieb Robert Zeller:
>>>     Gilles,
>>>
>>>     My system has a 6 core CPU with 16 GB of main memory; OS is openSUSE
>>>     Leap 42.3, I am running DK 5.7.0-178.7 as provided by SUSE distribution.
>>>
>>>     I am encountering a problem with multi core processing in BQM when
>>>     converting NIKON raw files and using all 6 cores; BQM is using the
>>>     following workflow:
>>>
>>>     Color Auto-correction
>>>
>>>     Sharpen Image
>>>
>>>     Color Effects
>>>
>>>     Convert to JPEG
>>>
>>>     This works fine with just ONE image in the run-Queue and "Work on all
>>>     processor cores" enabled. With more than one image in the queue I only
>>>     get a pattern of fine lines in RGB colors in the resulting jpegs .
>>>     Disabling "Work on all processor cores" I get correct results for all
>>>     pictures, but clearly at the prize of loosing all the possible speedup
>>>     of the 6 cores.
>>>
>>>     With all cores enabled I can see that every core is allocating its one
>>>     memory, and there is definitely enough memory on my system.
>>>
>>>     Could it be, that for some reason the memory gets mixed up among the six
>>>     running jobs? Or am I doing something  wrong?
>>>
>>>     Best regards,
>>>
>>>     Robert Zeller
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/digikam-users/attachments/20170924/221d85ad/attachment.html>