Cycles Apple Metal device feedback

Does anyone have experiences they can share for Cycles Metal performance related to:

Single Radeon Pro Vega II vs. Two Single Radeon Pro Vega II (or a single Duo card).

Much appreciated!

Hey everyone!
This is a little unrelated, but i’d love it if blender could integrate the Apple Object Capture API. It uses the LiFAR scanner on the newer iPhones and iPads to create a 3D photo-scan of any object!

1 Like

A couple of things don’t make sense.

  1. Blender isn’t available on iOS so how would the implementation work?
  2. There actually is already software that can do this, but it is paid. It works by taking a lot of photos of an object and creates a 3D model of it.
  3. If there is an app on iOS that do a photo scan of an object, you probably could just scan, export it to OBJ or something, and import that into blender. I don’t own an iPhone/iPad with a LiDAR, so I couldn’t test it.
2 Likes

Blender and Apple, thank you for the hard work. Been testing macOS 12.3 beta and Blender 3.1 release candidates on my iMac (Retina 5K, 27-inch 2019) with i9 and Vega 48 and cycles is very responsive in the viewport, and between metal support and denoise render times have dropped like a rock (sorry that this is qualitative and not quantitative.)

I am very intrigued how this will perform on a Mac Studio with an M1 Ultra - I’ve already seen a tiny hint of this with metal on my iMac, but I was wondering, will/does Blender’s implementation of OpenImageDenoise take advantage of the M1 neural engine, or the hush-hush AMX coprocessor? If so, I can imagine truly brutal performance coming out of these machines,

7 Likes

The Apple object capture API runs on a Mac. It’s a photogrammetry engine built into the OS that can make use of the extra depth information from photos taken on iOS devices.

Since it’s OS-specific, I imagine it would fit better as a Blender plugin than a built-in integration.

4 Likes

Hi,

I just upgraded my MacBook Pro to the latest OS (12.3) and it has a AMD 5500M. I saw the Blender 3.1.0 release note saying that the AMD chips can now be used in Cycles but I don’t find that to be the case. When I clicked the check box next to my graphic card in the system tab in the preference, the GPU option indeed lit up in the property’s menu. But when I press the render button in the viewport, no rendering shows up but a line of text showing that first time loading render kernel may take a few minutes. I waited for about 1 minute or so and still nothing shows up. The results is the same when I press the F12 key. I also tried restarting the computer and the problem still exist. Now I am not sure if the Blender 3.1.0 could use AMD chips with Cycles. Could any of the guys in the development team take a look into this problem, please?

Thanks

Please wait 4-10 minutes for Blender to finish kernel compilation. Every time you update Blender or macOS, you need to recompile the kernel. Be patient.

2 Likes

Technical question:

Has multi GPU via Infinity Fabric Link been implemented? If not, will it be?

And: is it even relevant for render compute performance? For simulations?

The answer to all your questions is very likely: no.
Rendering on multiple identical Graphic Cards should scale well even without Infinity Fabric Link.
If you have a multi GPU setup in your Mac Pro, you should be able to try it out. I would be interested to see the render times of the BMW scene on one card and on two cards. Or does Blender only see one GPU?
Simulations run on the CPU, afaik there is no use of the graphic card right now.
I will later post the BMW-render time of my iMac Pro Vega64.

Roger, thanks.

I might be able to pick up a used Pro Vega II that has been removed from another Mac Pro. Unfortunately, when you buy a Mac Pro with a single card, you don’t get the Infinity connector… and Apple doesn’t sell them individually.

If you buy an additional card from Apple, you get the connector with the card.

But over time, there will be a “large” 2nd hand market for used single cards without connectors that we will want to install in other single card Mac Pros. I feel that Apple needs to offer the connector for cheap to those of us who buy used MPX modules.

Anyway, I might be able to get the card as a short term loaner to evaluate performance. Will report back if that happens.

1 Like

Just got a MacBook Pro M1 Pro and the speed is great! 3:11 in Classroom scene. And now hearing that the times still could be halved is mind-blowing! One thing I have been wondering is that could the Neural Engine used for Motion Tracking. Tracking times in Final Cut Pro are great as it uses the Neural Engine.

1 Like

Hello,

I took the new Blender 3.1 (apple version) (did not know it was out yet) and tried it on my Macbook Air M1 base model. This is the scene:

Then I made a render on the CPU to see if it is optimized or something and sadly instead of 8min (Blender 3.0) it took 8:31 (Blender 3.1)

Then I went into Preferences, System and enabled M1 GPU and M1 CPU (both)
I pushed on render GPU Supported and after 5m20s it finished the whole render (actually it kept doing a portion, then disappearing then again etc and after 5min it showed me that all 128 samples were finished (and the result, well it is only 25% of it done and even that one is looking like 1 sample).

It was really funny to get this result, seeing the many positive comments.
So I hit Experimental, GPU, and after 14m 59s it showed me this result (I think 25% missing, but at least it looks like 64samples instead of the 128)

Can someone tell me where my technique is bad, or was there a box I forgot to check?

If I render one hexagon, it renders about twice as fast on GPU+CPU than on CPU alone. But if I render all hexagons, it fails on GPU Supported and it does only a portion on CPU+GPU in twice the time than on CPU alone. I can share more details if needed.

2 Likes

Yes, the latest Apple chip (M1 Ultra) links two GPUs and CPUs using “Ultra Fusion” to pool resources.

Edit: Sorry, you’re talking about dGPUs on the MacPro. :nerd_face:

iMac Pro 8core Vega64 16GB:
BMW Scene
CPU: 3:14 Min - Peak Memory 147.5 MB
GPU: 0:57 Min - Peak Memory 768.55 MB ???
CPU+GPU: 0:56 Min - Peak Memory 136,25 MB

Classroom:
CPU: 7:49 Min - Peak Memory 630,15 MB
GPU: 1:55 Min - Peak Memory 1359,80 MB ???

Finally I can use my GPU, thank you!
Still, the high memory usage on GPU seems a bit strange.

1 Like

This seems like a bug. And bug reports should be made by following these steps:

  1. Open Blender.
  2. Select from the top of Blender Help -> Report a bug
  3. Fill out the form that opens in your web browser with all the relevant information related to the issue you’re experiencing, including stuff like RAM configuration of your Mac, any error messages you’re getting, the issue you’re experiencing, steps to reproduce the issue, a .blend file if the issue is limited to a specific .blend file, etc.
2 Likes

Hi, same for me : Cycles Apple Metal device feedback - #232 by fantomac
768.45 M Peak Mem.

I was following this thread holding my thumbs.
2019 MacBook Pro 16, BMW:
CPU - I7: 05.28.39
Radeon Pro 5500m 8GB: 01.26.20
RX 6900xt eGPU: 00.21

This is really, really amazing, just wanted to point to one thing that 12.3 introduced: it broke many setups with high end AMD gpus (eGPUs, Mac PRO), so unless Apple fixes it, I think I will revert to Monterey 12.2.
Static rendering is fine, but all all other work is stuttering and lagging :frowning:

1 Like

Wanted to share some Blender 3.1 performance numbers for 2017 8-core iMac Pro (details below). Very close to @martin.ackerl, but his Vega 64 is faster. GPU is 3x faster for me.

•	BMW CPU 03:30, GPU+CPU 01:04
•	Classroom CPU 07:50, CPU+GPU 02:20
•	Junk Shop CPU 01:18, CPU+GPU 00:44

Note that all my Blender 3.1 CPU render times were higher than the 3.0 scores. iStat Menus said no other process was taking CPU efford.

Details: 3,2 GHz 8-Core Intel Xeon W, 64 GB 2666 MHz DDR4, Radeon Pro Vega 56 8 GB. Blender 3.1 release, MacOS 12.3.

2 Likes

Eagerly awaiting the first Cycles test results on a Mac Studio with M1 Ultra.

3 Likes

RX 6900xt eGPU: 00.21

That is truly amazing. Just a few seconds behind the RTX 3090.

Radeon Pro 5500m 8GB: 01.26.20

This is super weird. I got 1.26 as well (although on previous tests I got 1.17). What’s weird is that I’ve a Radeon RX 5500-XT (Desktop). According to GPU UserBenchmark, my GPU is faster than yours. Could I be getting a CPU bottleneck? My CPU is far less powerful than the one in your machine.