If you can get some debug information about the crash, then yes, open an issue.
If you can’t, then it probably isn’t worth it, because it likely won’t get fixed with the only knowledge being “it crashes with Color Zones”.
also make sure you disable OpenCL and see if it crashes.
I’ll do some more testing. What options on darktable command line should I use to try and get more information?
Launch from the terminal with the option -d all
Thought I’d revive this thread to share a couple more findings. I recently picked up my own M1 Air and finally got around to some more testing. This time using my own XMP that has a more modern workflow including profiled denoise, two instances of diffuse and sharpen, sigmoid, and tone eq. Here’s what I ended up with:
M1 Air 16GB OpenCL: 14.72s
GTX 1050 2GB: 77.83s
I’ll echo @Antonio.guerreiro and say the editing experience on the macbook is pretty great. Generally, edits occur almost in real time until D&S is active. Luckily I have a couple of fairly consistent setups for D&S so what I usually do is apply D&S as an export style so I never actually experience the performance impact.
Maybe it’s a little off topic but I risk the question :
Given the reported performances here, is a mac mini m1 or such the best per/$(€ or other) performance machine for DT ?
I have the impression that even for the price of a brand new unit it’s not clear to me that second hand PC market have best to offer …
As basis to my reaction here is what I found quick researching :
In my country the apple website displays the newer M2 mini unit at a lower price than older M1 … that is 700€ for that price on second hand market I think I’d get an i-5 9th gen with 16Gb of ram an SSD and a gtx 1060 or such…
Yeah, specially since the benchmarks with the 2GB 1050 are probably a worst case scenario due to the limited VRAM. Same goes for the X1 Carbon with built in graphics.
It’s good to remember Apple is doing everything they can to kill OpenCL and only support Metal. So it might be a gamble in the long term.
I was more counting on Asahi linux 
Using GNU/linux as main OS since 2004 I did not imagine daily driving anything else !
That said the main drive of the M1 and M2 being non swappable could be an issue in the long run …
My system is probably CPU bottlenecked but I achieve far worse results with a 8GB GTX 1650 …
That said @hatsnp my original post is a question, not as an assertion 
Interesting results 
Fair, I meant my reply more as a comment than anything else I’m not the right person to give an answer to it
Definitely interesting, I would’ve expected the 1650 to be a ton faster than my 1050.
I’m also quite excited to see Asahi mature. The performance they’re getting out of GPU compute seems very promising. Once that becomes generally available Mac minis/airs definitely do look like solid value (assuming you’re ok with Apple’s hardware choices).
Last year, I bought a brand new gaming PC mostly for image processing (and originally gaming, but then the Steam Deck happened…), with a fancy i5-10400 and a NVIDIA 3060. This year, I thought I’d try an M1 machine, so as to have less energy consumption and fan noise, a Mac Mini M1 base model with 16 Gb RAM.
3060: [dev_process_image] takes 0,735 secs (3,959 CPU)
M1: [dev_process_image] takes 0,652 secs (0,867 CPU)
3060: [dev_process_preview] takes 0.284 secs (1,384 CPU)
M1: [dev_process_preview] takes 0,156 secs (0,242 CPU)
Sigmoid, a CPU-bound module, is a full 2x faster on the M1 than the intel CPU. That’s astonishing.
Wow it’s kind of shocking, it’d be nice to have that correlated with other tests …
I’m not a mac addict at all (anti trade and standard breaking/user jailing practices, anti right to repair practices… really pushed me away from that brand not counting the silly marketing 
) but given the power consumption and reasonable price/performance when it comes to DT, I’d consider getting one eventually (when my machine breaks; and shoving asahi linux on it’s drive…)
PS to be precise, which of these tests are ran on OSX and which on Asahi linux ?
Mine ran on M1/macOS, and Intel/Kubuntu. I’d be happy to run some more repeatable benchmarks, but so far the M1 really does look shockingly good.
Hi there,
I searched the forum and found this thread. Maybe there are also some experiences with the m2?
I‘m thinking of getting a new Mac mini m2 base model 8gb ram and 256gb. Coming from an iMac 14,2 late 2013 i5 3,2 ghz with a GeForce get 755M (1gb ram) I guess it would be way faster, especially when using diffuse and sharpen.
Has someone tried 8gb vs 16gb ram directly? Would the upgrade help when developing or just make an export faster?
And would upgrade the disc space to 512gb be better, because of the speed gain? Just from the size it doesn’t make sense to me, since external drives are way cheaper and I could swop it between MacBook and the mini. But maybe it still would be better to have more space available, when Asahi or another Linux is ready.
How are the experiences with Ventura? Still problems with it, like @dorescu suggested earlier?
Has apple made an announcement, when they want to drop OpenCL finally? Couldn’t find anything to OpenCL on Sonoma.
Happy to hear from you and your thoughts.
Hi @semperit!
Has someone tried 8gb vs 16gb ram directly?
Yes, sort of – but not on a Mac, on a Ryzen CPU,
and it was many versions ago… (I even compared
4 * 8 gig with 2 * 16 gig RAM sticks.)
Have fun!
Claes in Lund, Sweden
That gtk issue was already fixed upstream so fine since 4.2.1
Apple drops opencl if they decide to do so - and I doubt they will announce in advance 
at least there are no indications they will do it soon … but don’t expect support for opencl > 1.2
Hi Claes,
What were your findings of the test?
Alright, thanks for the answer.