HW assisted Synology performance

[3djc 5]

I'm currently having a (not so old) DS1815+, which serves nicely as an emby server, as long as no transcoding is involved. I'm considering upgrading it to either a DS918+ (Apollolake Celeron J3455) or DS1819+ (Deverton Atom C3538). They both have pro and cons in my situation, so I was wondering how we can compare transcoding performance now that hw acceleration is there.

 

It is really hard to relate to somebody else experience to compare those, as each video file is different. Is there a way to 'extract' a kind of standard emby transcoding performance index that Syno (or even better if all types of emby servers are supported) owner could run and allow the community to have better view of transcoding perf ? (say one for 1080p and one for 4k for exemple)

 

Best regards

 

jc

[Luke 36999]

@@softworkz might be able to suggest some benchmarks. Thanks.

[softworkz 3325]

This is hitting something that is actually on my private wish list: Being able to run transcodings based on fixed configurations.

At the moment, emby is extremely flexible trying to adapt to each individual playback situation - which is a very good thing.

 

The problem - when it comes to benchmarking - is, that it is very difficult to replicate a certain transcoding setup on different systems. 

So at the moment I'd have to say no, that's currently not possible in a reasonable way.

[3djc 5]

Hum,

 

Thanks for the quick response. That’s a bummer, I was hoping we could run a transcode of a reference file to a reference hypothetical device and benchmark that. Sure that could not be used to say if a file could be transcoded on system x, but it could allow to say overall, this one is roughly 10% faster at this than this other

 

3djc

Edited November 30, 2018 by 3djc

[softworkz 3325]

I didn't want to say that It's impossible, there'S just no straightforward way to do so.

 

You can of course try to setup both systems in the same way, with the same settings and use the same client and same file for testing, but you would still need to carefully compare the ffmpeg command line parameters to be sure that it's the same setup.

[3djc 5]

Then I guess at the end, what matters is ffmpeg performance on that box, so maybe a shell script with a predefined file (or maybe 2, a typical x264 1080p and one x265 4k) and 'mainstream' command line param would do it ? Is there anything I need to be aware off to be sure to be running with HW acceleration ?

 

3djc

Edited December 1, 2018 by 3djc

[solabc16 379]

Hello

 

Agree with the sentiments above, and of course if you need transcoding then the DS918+ (or similiar) is your only option.

 

As noted, neither the DS1815+ or DS1819+ support h/w transcoding.

 

The raw processing power is not sufficiently different across these two platforms to have a meaningful difference in the context of software transcoding capability. Speaking very generally, they tend to be good a good solution if you only need to transcode SD content.

 

Best

- James

Edited December 2, 2018 by solabc16

[3djc 5]

Thanks for the advice, it is for the occasional situation where transcode is happening for me (occurs once in a while when wife uses iPad/AppleTV), which might very well happen on 4k and/or x265 content.

 

I’ll go for a 918+, i’ll try to squeeze 16gb ram in it (not officially supported but usually works fine).

 

3djc