Help wanted with testing on M1 (Apple Silicon) Mac

Anything and everything to do with DCP-o-matic.
Stern
Posts: 3
Joined: Sun Mar 14, 2021 9:43 pm

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by Stern »

First of all, thanks for the quick fix! It works like a charm!

I did 4 benchmarks, one for Big and Sentinel on the internal SSD. Big did it at 17fps, Sentinel at 11fps. I did another one for Sentinel on my Promise R4 raid0. As for fps, there was no difference with SSD. I did a 4th Sentinel benchmark, with the source on SSD and the DCP output to the external raid. That turned out to be slower at 10fps. Not sure if these are even useful comparisons, but I was kinda curious myself.
Carsten
Posts: 2804
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by Carsten »

During the final phase of the conversion, the DCP has to be read back for hash calculation. That is probably somewhat slower from the RAID than from the very fast internal SSD. The J2K conversion itself, as the main task of the conversion, was probably exactly same speed for both runs.
ArthouseIthaca
Posts: 1
Joined: Wed May 19, 2021 9:58 pm

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by ArthouseIthaca »

I just downloaded version 2.15.149 and am attempting to transcode a DCP from a 1080p MKV on my MacBook Air w/ M1 chip.

Previous test version worked on this machine, but new test version is crashing with as soon as I start the encode process. Tried adjusting color range and JPEG200 bandwidth settings. Also adjusted number of threads. Still crashes as soon as "Make DCP" button is clicked.

Detailed crash report is attached to assist with development:

DOMCrash_2021-05-19.txt
DOMCrash_2021-05-19.txt
You do not have the required permissions to view the files attached to this post.
carl
Site Admin
Posts: 2548
Joined: Thu Nov 14, 2013 2:53 pm

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by carl »

Thank you very much for the bug report!

I made a note on the tracker and I'll take a look as soon as I can.

carl
Site Admin
Posts: 2548
Joined: Thu Nov 14, 2013 2:53 pm

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by carl »

Please could you try again with alpha 2 and let me know how you get on? :)
briancallahan
Posts: 4
Joined: Wed Jul 07, 2021 5:48 pm

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by briancallahan »

Hi Carl,

I’ve take a look through a few threads, including this one to try to work out if the newest alpha version might be stable enough for production (I mean, I know it’s an alpha version).

I’m looking at getting a newer Mac Mini to help with making DCPs, either as a standalone unit or as an additional server to chain with my main machine.

Does the current stable version run reasonably on the M1, too?

If I used the latest Alpha on the new M1 silicon as a processing server, would I be able to use the current stable release on an older Mac as the master, or do I need to use the same version on both machines?

Thanks for your time!
carl
Site Admin
Posts: 2548
Joined: Thu Nov 14, 2013 2:53 pm

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by carl »

I'm not aware of any terrible bugs in the 2.15.x alpha - the main thing blocking release of 2.16.0 is trying to get the OpenGL rendering backend to work better.

I don't personally have an M1 machine but reports suggest that 2.15.x works fine. 2.14.x will only run via Rosetta.

Unfortunately if you want to use 2.15.x on an encoding server you would also need to upgrade your master.
Carsten
Posts: 2804
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by Carsten »

While the M1 Macs in general certainly can perform nicely with certain optimised apps, they are currently not the best choice for DCP-o-matic DCP encoding. Certainly not as an encoding server. Of course, they do work, and if you want to buy one for other reasons as well, go ahead. But for an encoding server, there are certainly better or cheaper choices (e.g. an AMD Ryzen 9). Even on a completely different CPU architecture as on an M1, it's the number of hardware computing cores that is most important. And the current M1 systems are optimised for low power, not highest performance. That will certainly change with future systems.

https://dcpomatic.com/benchmarks/fps
IoannisSyrogiannis
Posts: 185
Joined: Mon Nov 13, 2017 8:40 pm

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by IoannisSyrogiannis »

In the new site layout I can't see values for fps on benchmarks.
Do the alternate columns/stripes suggest a frame per second each, or is it intentional to have a relative but not numeric representation?
...or is it just an overlook?
carl
Site Admin
Posts: 2548
Joined: Thu Nov 14, 2013 2:53 pm

Re: Help wanted with testing on M1 (Apple Silicon) Mac

Post by carl »

Sorry, that's a mistake - it should be fixed now.