slow batch conversion

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

slow batch conversion

fotografie
Hello,,

I had the same problem with digikan 5.4.

The batch converison worked really slow. But I found an entry in the
menu to tell digikam to use all prozessor kernels. And that speeded it
up. Now I have the problem again with digikam 5.6 and 5.7. Where can I
set digkam up to use all cpus?

Thanks in advance for your help.

Best regards Ulf

Reply | Threaded
Open this post in threaded view
|

Re: slow batch conversion

Gilles Caulier-4
Hi,

There are few explanation to give here about multi threading.

In digiKam all is multi-threaded when this can be and when it's give
improvements of speed :

- mostly all image editor tools use multi-core.
- mostly all BQM tools use multi-core (same code than editor in fact).
- BQM queue can be multi-threaded (more than one at the same time).
One core or multi-core (optional)
- Maintenance tools : same as BQM queue.
- The database access is now ported to multi-cores/multi-threads (KIO
slaves was used before). It use always all core to process, but some
bottleneck can appear with database if too much query are queue in
interface, especially with sqlite. Mysql is better in this case.
- Face management use multi-core or not.
- Assistants as panorama, emailimages, pseudoHDR, HTMLGallery,
Videoslideshow : all are mutithreaded, and some tools use multi-core,
some not, as videoslideshow which serialize a video flux from images
(encoding with ffmpeg use multicore).

What do not use multicore :

- Scan items : even if it run in a separated thread, it do not use
multicores yet.
- Download items from camera : there is a bottleneck with concurrent
media access. It use a separate thread at least.

Note there is a bug about multi-core support fixed in next 5.7.0 which
can introduce huge time latency in previous versions. This fix is a
big step in stabilization of application.

Best

Gilles Caulier




2017-08-20 0:03 GMT+02:00 fotografie <[hidden email]>:

> Hello,,
>
> I had the same problem with digikan 5.4.
>
> The batch converison worked really slow. But I found an entry in the
> menu to tell digikam to use all prozessor kernels. And that speeded it
> up. Now I have the problem again with digikam 5.6 and 5.7. Where can I
> set digkam up to use all cpus?
>
> Thanks in advance for your help.
>
> Best regards Ulf
>
Reply | Threaded
Open this post in threaded view
|

Re: slow batch conversion

fotografie
Hello,

thanks for the answer.

Let me give a short overview of what I did (digikam 5.6 and earlier) and
where my fault was:
- i selected the pics for batch conversion
- in batch conversion at the bottom you have the batch conversion tabs
and the control list.
- in the tab named "behaviour" (or so, I use the german version,
therefore I don't know whether the name is correct) there you can
select: "use multicore" - and that I did not do, so the process of
converting 500 pics was slowed down.

Have a good week
Ulf


Am 20.08.2017 um 11:03 schrieb Gilles Caulier:

> Hi,
>
> There are few explanation to give here about multi threading.
>
> In digiKam all is multi-threaded when this can be and when it's give
> improvements of speed :
>
> - mostly all image editor tools use multi-core.
> - mostly all BQM tools use multi-core (same code than editor in fact).
> - BQM queue can be multi-threaded (more than one at the same time).
> One core or multi-core (optional)
> - Maintenance tools : same as BQM queue.
> - The database access is now ported to multi-cores/multi-threads (KIO
> slaves was used before). It use always all core to process, but some
> bottleneck can appear with database if too much query are queue in
> interface, especially with sqlite. Mysql is better in this case.
> - Face management use multi-core or not.
> - Assistants as panorama, emailimages, pseudoHDR, HTMLGallery,
> Videoslideshow : all are mutithreaded, and some tools use multi-core,
> some not, as videoslideshow which serialize a video flux from images
> (encoding with ffmpeg use multicore).
>
> What do not use multicore :
>
> - Scan items : even if it run in a separated thread, it do not use
> multicores yet.
> - Download items from camera : there is a bottleneck with concurrent
> media access. It use a separate thread at least.
>
> Note there is a bug about multi-core support fixed in next 5.7.0 which
> can introduce huge time latency in previous versions. This fix is a
> big step in stabilization of application.
>
> Best
>
> Gilles Caulier
>
>
>
>
> 2017-08-20 0:03 GMT+02:00 fotografie <[hidden email]>:
>> Hello,,
>>
>> I had the same problem with digikan 5.4.
>>
>> The batch converison worked really slow. But I found an entry in the
>> menu to tell digikam to use all prozessor kernels. And that speeded it
>> up. Now I have the problem again with digikam 5.6 and 5.7. Where can I
>> set digkam up to use all cpus?
>>
>> Thanks in advance for your help.
>>
>> Best regards Ulf
>>

--
Mit freundlichem Gruß

Ulf Richter

Kindle-E-Books:
Fotografie - leicht und verständlich erklärt. Dein Weg zum besseren Bild
http://amzn.to/20Mhz8L 

Linux VServer Setup Leicht und einfach erklärt
http://amzn.to/18kfJuI 

[hidden email]
mobil: 0170 295 8780