2022-04-01 11:56 PM
Has anyone run Archicad on a new Mac Studio? If so how was the performance?
2022-04-05 07:18 AM - edited 2022-04-05 07:20 AM
I find it strangely upsetting. What I would like to hear is Graphisoft’s timeline for getting Archicad to run natively on Apple silicon.
No specific date, but later this year.
2022-04-06 08:49 PM
We now have a few days of experience with both Mac Studios (M1 Max with 10-core CPU, 32-core GPU, 64 GB RAM and 1 TB SSD storage).
There is no noticeable performance difference compared to the iMac 24" with M1. The larger projects are exactly as sluggish as on a 2019 iMac i9 (the comparison model in our case).
For example, if a projectile is changed with active transparency pause, you can easily wait 20 seconds or longer (for larger projects)!
Presumably, this was to be feared, since Archicad has not managed to use multiprocessors smartly and thus only the maximum singlespeed is decisive (and this is now unfortunately comparable with the i9 from 2019).
If this is not significantly better in the summer with the native version, Graphisoft probably has no real desire for Apple computers and we will probably have to write off Archicad as a bad investment!
Even a sculpture of Steve Jobs at the headquarters will not help ;-(
So we can only hope that the homework will be done and at least the same speed increases will be realized as VectorWorks has done. Maybe in this case, twice as long to develop also means twice the speed increases?
Fingers crossed 😉
2022-07-10 05:19 PM
I have an Ultra 128MB and I have to say it's like butter compared to my 2014 Intel MAC 4GHz.
I've noticed that, for quick and dirty photo renderings, IradLegacy produces very quick renderings with 20 cores. In that setting, the GI phase doesn't take advantage of cores, but Render portion was quite fast. For low-resolution movies, lack of parallelism hurts during the GI phase. Of course, legacyIrradiance suffers from sparkle-noise in movie renderings, but for still-renderings it creates a cleaner look than the more accurate QMC.
The "Standard 20core" Render in AC25 appears to run in 1/3 the time of the "Redshift w/ internal GPU" at equivalent settings.
2022-07-14 03:38 AM
I just downloaded (again) GRAPHISOFT_Archicad_Sample_Project-S-Office_v1.0_AC25.pla
I run the following....
Model Name: Mac Studio
Model Identifier: Mac13,1
Chip: Apple M1 Max
Total Number of Cores: 10 (8 performance and 2 efficiency)
Memory: 32 GB
Chipset Model: Apple M1 Max
Total Number of Cores: 32
Vendor: Apple (0x106b)
Metal Family: Supported, Metal GPUFamily Apple 7
I opened the above file in 26RC1 (rosetta 2 only)
Went to 3D (default 3D view and settings) and generated a 3D PHOTORENDER using Redshift.
It took about 2min 30 used (80% CPU to start then settled at 30%) Used 1% GPU
Using Standard took about a second longer with similar CPU and GPU loads.
Rotating the whole model in 3D used about 85% CPU snd 45% GPU
Using Detailed shading (shadows) changed the CPU to about 95% and the GPU to 65% (shadows still need to regen after move.
2022-12-08 07:30 PM
Now that GS has released a specific software for Apple Silicon, has anyone tried it on their M-chipped machines? Noticeable difference?
2022-12-14 04:59 PM
Bumping this. Please post experiences from Archicad running natively on Apple Silicon.
2022-12-14 06:20 PM
Everything like refresh, panning and orbiting, opening views, etc. seams a bit faster but rendering in Cineware does not show any real improvement. I'd like to know if a future update will fix this.
2022-12-20 06:09 AM
Thank you for testing this product DS, this reply was very helpful to me.
I guess an AC25 redshift render for a low rise residential home would be done in less time ?
I am looking at one for me to use part time. I always used PC’s in the past but I think I had enough problems with those types of machines.
2022-12-20 08:28 AM
Before I start please note an error in the times above. The photorender using standard took 1 minute longer at 3min 31 sec NOT 2min 31sec. Sorry about that.
I opened the same file “GRAPHISOFT_Archicad_Sample_Project-S-Office_v1.0_AC25.pla” using the Australian version of Archicad 26 build 4024 on the same computer recently upgraded to MacOS 13.1 (22C65) as follows:-
Default 3D window, default Photorender settings using Redshift 1min 32sec compared to 2min 30sec a saving of approx 1 min.
Default 3D window, default Photorender settings using Standard 2min 17sec compared to 3min 30sec a saving of approx 1 min 13sec
I then fired up the INTEL version of Australian Archicad 26 build 4019 opened the same PLA.
These are the result (running through rosetta2)
Default 3D window, default Photorender settings using Redshift 1min 48sec slightly longer than silicon build (but I assume the INTEL version of Archicad uses a silicon build of redshift?)
Default 3D window, default Photorender settings using Standard 3min 07sec again slightly faster than the previous INTEL build (RC1) running under rosetta2
At some point I will fire up my 2014 MacBook Pro and run the same tests
In essence (and just using this average size project and a very basic test) it can be seen that there is an approximately 30% speed increase. This is significant but actually irrelevant because Apple have now completed their transition to their own Apple silicon chips. No more Macs with INTEL chipsets.
2022-12-22 07:24 AM
Bit more info following on from my previous post (2022-12-19 11:28 PM)This time using my 2014 MacBook Pro (INTEL)
same file GRAPHISOFT_Archicad_Sample_Project-S-Office_v1.0_AC25.pla” using the Australian version of Archicad 26 build 4019
Default 3D window, default Photorender settings using Standard 4min 46sec (compared to 3min 07sec on Mac Studio MacOS 13.1 above)
Default 3D window, default Photorender settings using Redshift. Redshift not supported by previous OS "Big Sur"
Times to open project 2014 MacBook Pro:- 29 sec to library dialogue 55sec in total
Times to open project Mac Studio Max:- 15 sec to library dialogue 26sec in total
Times to open 3D widow 2014 MacBook Pro:-7.25 secs
Times to open 3D widow Mac Studio Max:-4.07 secs
All times are approximate.