MP&XnConvert: No multi-threading when encoding AVIF

[Stock]

Creating an avif takes a looong time.
To show what I mean, I'm converting a big 3000x3000 picture from jpg to avif using an AMD 5600X 6-core/12-thread processor.
Here are my conversion settings:

avif multi bug settings.png (19.51 KiB) Viewed 1562 times

Multi-threading is selected and set to use all 12 threads.

Here are the results:

time1.png (10.59 KiB) Viewed 1592 times

The whole affair took 1 minute and 5 seconds (sic!)
but during the entire time, the CPU never went above 9% load.

To confirm, I ran it one more time with multi-threading turned OFF:

time2.png (10.23 KiB) Viewed 1578 times

...and it took the very same 1 minute and 5 seconds.


A friend of mine used AOM's libavif and was able to successfully multi-thread the encoding.
With his 4-core/4-thread Phenom processor he cut down the required time by almost 75% after adding the

Code: Select all

--jobs

parameter. See also: https://web.dev/compress-images-avif/

Speeding up the encoder
One parameter that may be good to change depending on how many cores you have on your machine is the --jobs parameter. This parameter sets how many threads avifenc will use to create AVIF images. Try running this at the command line. [...]
[Using] 8 threads when creating the AVIF image [...] speeds up AVIF encoding by roughly 5x.

Imagine that: five times as fast! For my example picture, encoding time would go down from 65 seconds to just 13 seconds!

If anything needs multi-threading, it's AVIF.

[cday]

You presumably have multi-threading selected on the Output tab?

[Stock]

multi-threading selected?

[cday]

Alles klar...

[masterjp]

You are funny.

XnViewMP Multi-threading setting use many CPUs cores, but it is important to convert more than 1 file.

[Stock]

You are funny.

Thanks.

Avifenc can multi-thread the encoding of a single file.
If the functionality is already included and ready to be unleashed simply by slapping the --jobs parameter on it, why not go for it?

As I've shown, there are huge savings in encoding time to be had and most of the time I don't have 12 files at hand, but only a couple of files.