Author Topic: CRITICAL SPEED ISSUE WITH APPLE M SYSTEMS  (Read 10183 times)

2023-12-01, 00:45:23
Reply #75

frv

  • Active Users
  • **
  • Posts: 344
    • View Profile
Thanks CambridgeCreativ
The node editor is handy at times and I do use it. Mostly to use a single map for several materials. Still, I have no idea why I should worry about wether it is a CR bitmap or just a dragged a dropped map from the finder.

The asset browser does not accept CR bitmaps btw. Although I am no longer using the asset browser because of the many problems with CR assets and their materials.


2023-12-03, 02:02:30
Reply #76

jojorender

  • Active Users
  • **
  • Posts: 246
    • View Profile
Although I am no longer using the asset browser because of the many problems with CR assets and their materials.

hey frv, what are you using instead of the asset browser? Don't know of any alternatives...

CR team,
I find it strange that only some cor bitmap shader problems get fixed and the rest ignored again...
Lost all hope for a real fix. Corona was fun while it lasted.

Peace!


2023-12-03, 22:10:14
Reply #77

Stefan-L

  • Active Users
  • **
  • Posts: 514
    • View Profile
i use the asset browser all time for all, works actually well for us .

the only thing is i not store the textures inside the browser database, but keep texture referenced as external files on disc/server. this is more reliable and also faster.

2023-12-03, 22:50:15
Reply #78

frv

  • Active Users
  • **
  • Posts: 344
    • View Profile
The asset browser work great but I have to many assets that are not working properly. Especially with layered and CR bitmaps.
I have set up files with assets that belong together and just copy paste them in my working files.
Another thing with the asset browser is that it duplicates all CR materials. For instance, many different maxtree assets can share the same 2 or 3 materials but when placed with the asset browser all assets have their own materials.

2024-01-15, 04:11:15
Reply #79

BigAl3D

  • Active Users
  • **
  • Posts: 881
    • View Profile
I wanted to throw this into the fire if I may. We acquired a Mac Studio M2 Max with 64 GG RAM and a 500 GB drive. This wasn't mean to really be a workstation, but to control our server and be an occasional workstation. Anyway, a decently powered system nonetheless. I have a simple scene with three products (a deodorant stick, a tube and a small spray) set up in Corona. Simple materials and graphic for the labels. Renders look amazing as usual. A couple of lights and one EXR in the Corona Sky object. Just three object with nothing else. I rendered an image at 4k, 15 passes and Intel Denoise. It will be delivered in a video at HD. I don't think I can show the image here though, yet.

This one frame took 3 min. 23 sec. on my 2017 iMac Pro with 18 cores. Not too bad. I will have 3 to 5 seconds to render. So I rendered this same frame on the M2 Max. expecting at least a 30% time reduction. It rendered in 3 min. and 40 sec.! What in the hell is that about? I was expecting 3 min., maybe 2 min. 30 sec. on this new box, based on my other testing.

I rendered the C4D Grapes scene I like to do for simple CPU speed and got 2 min. 37 sec. for the M2 Max. The same iMac Pro got 3 min. 19 sec.

Looks like Corona doe not take advantage of the new chips, as others have been complaining about. So disappointing. I'm rendering at 2017 speeds.

I mean the new features are great and all, but the real selling point is fast rendering. Either Chaos isn't supporting the Mac fully, or Apple isn't providing the dev support to Chaos. It's a mystery.

2024-01-15, 10:32:56
Reply #80

Philw

  • Active Users
  • **
  • Posts: 267
    • View Profile
@BigAl3D Interested if its the C4D Bitmap vs Corona Bitmap factor?

2024-01-15, 10:39:54
Reply #81

frv

  • Active Users
  • **
  • Posts: 344
    • View Profile
My work is about 80% design and modeling in Vectorworks and C4D, photoshop, lightroom. Maybe 20% rendering.  I do see a huge diffrence with the newer Apple M systems. I went from a late 2015 4core intel mac with 32Gb Ram to a 20 core Mac Studio Ultra M1 128Gb ram. The MacStudio is a fantastic machine but for purely rendering only about as much faster as the increase in cores indicate.

Its my experience that what matters most is the number of cores or threads and enough RAM. Newer computers have better and faster memory(management). But that shows almost only in parcing the scene before render. I wouldn't be surprised if this applies to PC as well for CPU rendering.


2024-01-15, 18:53:24
Reply #82

BigAl3D

  • Active Users
  • **
  • Posts: 881
    • View Profile
@PhilW That's a great suggestion. Turns out the three materials did use C4D Bitmaps. I swapped them with Corona Bitmaps and it rendered :05 sec. slower, so that's nice. There are also Normal maps, but they already use Corona's Normal Map nodes.

@frv Yes, I too have always known that C4D would reliable scale depending on the cores. This iMac Pro has 18 and I just saw my M2 Max has 12 cores (8 performance and 4 efficiency). I would fully expect the new tech in these chips would easily make up the difference in cores, if not surpass it.

As I mention earlier, this Mac Studio wasn't spec-ed out specifically for this purpose, but I expected more. I made a graphic showing my four test just for giggles. Not sure if that info sheds any light on this issue. Since the C4D Grapes test shows a modest 21% speed increase, seems like this is a Corona-related issue. I use After Effects every day and the few jobs I've done on this Mac Studio have blown me away with how smooth, responsive and fast AE was. I did a bunch of tracking and used the new Rotobrush 3.o and it crushed it. Sigh.

2024-01-15, 18:59:28
Reply #83

Philw

  • Active Users
  • **
  • Posts: 267
    • View Profile
This is why I’m holding out for an m3 ultra studio. Should have 32 cores….  Plus raytracing hardware on the GPU for my Redshift life.
« Last Edit: 2024-01-15, 20:03:19 by Philw »

2024-01-15, 19:02:34
Reply #84

BigAl3D

  • Active Users
  • **
  • Posts: 881
    • View Profile
Thanks for that info Phil. So how would that new hardware impact Corona, since you know, you're here in the Corona forums?

2024-01-15, 19:06:47
Reply #85

Philw

  • Active Users
  • **
  • Posts: 267
    • View Profile
Just sheer number of cores on the cpu - so best of both worlds… in theory, of course!

2024-01-15, 19:09:01
Reply #86

BigAl3D

  • Active Users
  • **
  • Posts: 881
    • View Profile
I really thought these integrated chips would be faster simply because since everything is right on the same chip. Wonder if there is any un-locked power in these chips that just need a key to open them up.
« Last Edit: 2024-01-16, 00:51:37 by BigAl3D »

2024-01-16, 08:32:18
Reply #87

Philw

  • Active Users
  • **
  • Posts: 267
    • View Profile
I honestly would say those results aren't bad - a 12 core pretty much head to head with an 18 core machine does show some modern gains as clock speeds aren't massively different on the new stuff.
« Last Edit: 2024-01-16, 12:47:51 by Philw »

2024-01-16, 12:45:21
Reply #88

frv

  • Active Users
  • **
  • Posts: 344
    • View Profile
For me though pure render speed is not much of a thing. Rendering iterations at low res is much more. I really like the Corona lister for switching off displacements and scatters. I use it all the time. In my workflow it has cut my time spend rendering iterations by a lot.
It would even be better if you could switch displacement and scatters off by a button in the VFB. Maybe a quick preview button at a lower res as well switching everything off on the fly.  The lister works but it's another window to deal with.

2024-02-07, 22:37:57
Reply #89

handz

  • Active Users
  • **
  • Posts: 12
    • View Profile
So sad to see that nothing is still solved, seems like Corona does not care about Mac users at all. I really do not want to migrate to PC after 12 years... can anyone let us know, there is something being done about this issue? M chips are super fast for anything from 2d graphics to video editing,  seems bit weird they would be THAT bad for rendering. Worse than 4+ years old Intels...