I read that the latest beta has optimizations for M1/Pro/Max, so hopefully that gets released soon and we see some nice speed boosts on those new machines.
Edit: I just realized you’re talking about Studio 2 and had to double check which forum I’m in… pretty sure all of the non-AI apps aren’t being updated anymore.
Hopefully they will make a Native version of Topaz Video Enhance AI for Apple Silicon, the new M1 Pro and M1 Max processors, and also the inclusion of the neural engines can be a real game changer in terms of performance. I use the new MBP M1 Pro with 16GB of RAM, it is crazy fast on optimized apps, but on this app it is still quite slow…
the speed on the M1 Pro chip is okay at best, I can see with the help of some software that this program is not yet optimized to use all graphic and neural cores to its advantage. On top of that, It is not a native app, so there is a significant performance penalty for using Rosetta 2.
specs:
m1 max 32 core gpu
32gb unified ram
14-inch model
video:
h.264 highly compressed 1080p video
upscale 200% (1080p to 4K) using artemis medium quality
0.41 frames per second
Please name the version of the software used.
topaz video enhance 2.6. the latest version with native M1 support. it’s a bit buggy tho.
rosetta version (2.4.0) was the same 0.41 to 0.43 frames per second.
it looks like M1 native support version (2.6) uses less resources. I went from using 8GB of RAM to 2GB of RAM and not much utilization of the CPU/GPU with the native M1 update.
Due to the high bandwidth to the cpu and gpu, you don’t need as much memory.
In order for the gpus to deliver good performance on desktops, you have to write a lot to the videoram, otherwise the PCI-E will limit you.
But this only works if you know what should go in.
With ray tracing this works great, with videos you don’t know what’s coming next so it’s a random workload and here high bandwidths are worth a lot.
At least that’s how I understood it.
Anyone noticed that the encoding goes up sky high, once the window is not visible or the displays are turned off? At least thats the case with Monterey. Dont know if its an energy saver thing, but I have turned everything off or if it is a bug with VEAI.
I bought a couple of mac mini with m1 chip… for 850 a piece… they both run 1080p to 4K tasks with 2.6 at around .42 sec average
the new more powerful M1 Pro and M1 Max chipped machines do not seem to run faster… so
it seems to be the cheaper solution at the moment to get a couple of these mac mini boxes 
Oh good to know! I was thinking of buying a MacBook with the M1 Pro or Max, but if it doesn’t give 3x to 4x the speed performance of a cheap Mac Mini M1, then it would not be worth paying $2k to $3k.
that is what ive discovered so far… someone please prove me wrong in here if that is not the case
Have you had the chance to compare with the M1 Pro and M1 Max? How do you know they perform all the same?
read what others have posted here… the differences are not big… not big enough to justify the cost… at least to me
I’m wondering if anyone can provide some insights into the VEAI performance (2.6.1) with the cheaper Mac Mini M1 with 8Gb of ram and 256Gb SSD. Rather than plowing the money into upgrading my GTX-1650 to a RTX series card since I am not a gamer, the M1 option for me seemed to be a better use of funds and since I can also upgrade my other Topaz AI apps such as Gigapixel and Sharpen to native M1 and get speed boosts over my Mac Pro. Right now the still image apps are running on my Mac Pro 5,1 intel.
I’m really clueless as to what sort of performance gains I will be getting if I get just a Mac Mini M1 as a headless render server for VEAI like upscaling 480p to 4k or upscaling 1080p to 4K as well for Gigapixel and Sharpen AI.
Thanks in advance for any insights you can give me…
I’m using an 8Gb M1, with 256Gb SSD which is mainly used using Screen Sharing for Topaz and ffmpeg.
As with all things Topaz all thing depend on the image/video size and settings, Denoise usually takes a 2-3 seconds for a 5.7k image with is what I normally work with.
VEAI for PAL (720 x 544) to 4k using Proteus is usually in the 0.12 seconds per frame.
Happy to do a few comparisons for you if you want to pass a few small example files with settings to compare to equivalent runs on your Mac Pro, I just have a 7 year old iMac to compare too so its a
serious step up for me
Personally though I’d cough up the extra for 16Gb for a bit of breathing room, although maybe not necessary if you’re going to be using it exclusively for Topaz.
The Mac Mini was a game changer for me with the M1. I only render 720p 59.94 fps and use the Artemis Low Quality setting (as I am restoring old analog video). My render times are always .14 sec / frame.
I originally used an iMac 3.7 GHz 6-Core Intel i5 with 40GB RAM – same settings and render times were .28 sec / frame.
The only odd thing I am running into is after the video is rendered (and then I believe the software creates a new file with the audio merged with the video – which is necessary) – when quitting the software, it seems to want to create this merged file again before quitting.
I always have to terminate the application – as re-creating a 150GB file is little time consuming and redundant. I am using the ProRes 422 HQ setting. I wish it had the ProRes 422 setting. HQ is overkill for me and just creates larger video files.
Regardless of this – VEAI is quite amazing. I plan to buy a large number of Mac Minis M1 to shorten my rendering time. Curious what the supposed M2 chip will do with this. I have read that the other M chips are not providing much of an improvement compared to just the M1.
I do have 16GB of RAM – but I get the feeling it is not necessary for this software.
My understanding is that VEAI uses the Neural Engine, M1’s AI accelerator, and that didn’t change in the Pro or Max, you’d properly get a small boost from the video compression but that is easily outweighed by the time required to render the frame.
