Looks like it, but instead of an egpu, you'll soon be able to attach a bare bones PC with a 3090 via ethernet and use Octane X 2022 which will allow mixed Metal and Cuda rendering.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
MP 7,1 GPU rendering on the new Mac Pro. News, updates, chat.
- Thread starter vel0city
- Start date
- Sort by reaction score
??? I'm not sure I understand what you writing, it's a simple benchmark comparing the previous top of the line vs the new top of the line and runs 4x the speed, that's about it.. can't be more apple vs apple than this.
I'm not expecting the new GPU to perform 4x better in every test (synthetic or real world) and I'm still waiting for more data.
Cinebench R20 results shouldn’t be compared to R24 results. Similar with Geekbench 4 to 5 . Apple isn’t comparing to the norm redshift benchmarks. The old and new MBP are matched up on benchmarks version numbers , but detached from almost everything else .
Apple tend to drift into narrow siloed comparisons. i was speaking more to the attached graph and jumping around to map the ‘4x‘ back into a comparative number in the graph. You have the 4x but tough to attach that other than just “4”.
Good luck with OTOY delivering on that promise in the next 200y. They consistently overpromise and underdeliver. Just check the Octane X rollout — what a mess
Yeah its a shame isn't it. I'd love that dream to be a reality, that you could plug in a bare bones PC over ethernet to help out on Mac rendering. what a dream. but unfortunately, with otoy, until its out its best not to think that its a future possibility, and even then will be mighty crashy. I am so used to saving every minute when I work with octane its funny not funny.Good luck with OTOY delivering on that promise in the next 200y. They consistently overpromise and underdeliver. Just check the Octane X rollout — what a mess
Last edited:
This is very exciting (and also very sad for AMD, poor them not having a future in Macs). Can't wait to see real world bench marks of after effects rendering, C4D CPU rendering, Red Shift & Octane.
But in a way what makes me even more excited is the thought of the iMac desktop M1, hoping that can be even faster CPU and more GPU. I have a 2019 16" MacBook Pro, so gonna hang onto that and start saving for iMac replacement. Very much hoping it comes with 30" XDR screen. but sorry this is off topic, and just my head rambling. ?
The other thing Otoy have been promising for a long time is that when Octane does crash (and it does, continually), it will not crash the host DCC along with it. That would make the crashes a little less annoying.
The rumors is 20 and 40 (although there would be some 'binned' in-between offerings in there like they have done with M1 Pro and Max ). The idea is to multiply the Max 2 and 4 times . Look at the die shots though , it is probably a different die for the 2 but the same baseline functionality as the Max (Jade) one. More added to make combining the dies into combinations, but not as visible to the end user or the applications. ( or just way more expensive monolithic dies and prices. )
If the cap the larger screen iMac at less than a 100W ( capped on Max) then quite likely will be thinning it out (perhaps to lessor degree than the 24" ) .
The problem with trying to add the M1 Pro to the larger screen iMac is that it is only covering mobile mid range dGPU ; not desktop ones. Even the Max is only covering a substantively underclocked 3070 like performance. That would much closer to a horizontal move on the GPU side.
Max also backslides a bit on ports ( presuming they leave a Ethernet jack to compete with SD-card and more Type A/C ports ). A 20C-64G SoC would allow them to do a better "iMac Pro". I don't think it will be exclusive to the Mac Pro. Mac Pro volumes are just way too low to support two dies ( the quad set up would need a "mirror" die to place the inter-die at a 'four corners" meeting place. )
Does anyone have any experience using the W6800X with the Mac Pro? I realized that Redshift doesn't list it as a supported AMD GPU on their website, and I just purchased one. I've seen lots of people mentioning using the Duo though?
I'm in the same boat as you, Vel0city, freelancer working mostly on AE / C4D / Octane on a 5,1 MP with 3 X 980ti.
15 days ago, after a long long time of thinking, I finally purchased a refurbished 16core 7,1, thinking I cannot wait one or more year stuck again in my old system, with the 5,1 struggling on heavy projects, with no possiblities to upgrade most of my softwares
I put a rx6900XT inside, this renders 3X faster than my 3X980ti, on Octane X, planning to get a 2nd one so it would be 6X faster renders than on my 5,1
Now those M1Max MBP...
Today Apple gave me 15 more days to decide if I turn the 7,1 back and get a M1 Max
I work 95% of the time at home on my desktop
I use an old laptop occasionally when needed, but currently with no possibility to render 3D
So I know I will buy a M1 Max MBP one day, to render 3D when not home, but this is not at all a priority.
I know it makes more sense to turn back the 7,1, and spend one+ year working on a M1 max MBP, with external drives to have access to the 15TB of stuff I need daily for different tasks, and then get a promising Silicon MacPro next year
But I'm not a laptop guy, I like to have all my drives accessible inside the machine, also 2X6900xt still seem faster than one M1 Max chip if I'm not wrong, and well, I kinda like this 7,1
And if I buy a M1 Max now it will have to be the top 32core / 64Gb to be my daily machine
Then when I get a Silicon MacPro, my M1Max expensive laptop will stay unused most of the time....
Well I'm a bit hesitating, even if I know it might be a bad idea to keep a 10k Intel machine, that might be difficult to sell in 1 or 2 years...
15 days ago, after a long long time of thinking, I finally purchased a refurbished 16core 7,1, thinking I cannot wait one or more year stuck again in my old system, with the 5,1 struggling on heavy projects, with no possiblities to upgrade most of my softwares
I put a rx6900XT inside, this renders 3X faster than my 3X980ti, on Octane X, planning to get a 2nd one so it would be 6X faster renders than on my 5,1
Now those M1Max MBP...
Today Apple gave me 15 more days to decide if I turn the 7,1 back and get a M1 Max
I work 95% of the time at home on my desktop
I use an old laptop occasionally when needed, but currently with no possibility to render 3D
So I know I will buy a M1 Max MBP one day, to render 3D when not home, but this is not at all a priority.
I know it makes more sense to turn back the 7,1, and spend one+ year working on a M1 max MBP, with external drives to have access to the 15TB of stuff I need daily for different tasks, and then get a promising Silicon MacPro next year
But I'm not a laptop guy, I like to have all my drives accessible inside the machine, also 2X6900xt still seem faster than one M1 Max chip if I'm not wrong, and well, I kinda like this 7,1
And if I buy a M1 Max now it will have to be the top 32core / 64Gb to be my daily machine
Then when I get a Silicon MacPro, my M1Max expensive laptop will stay unused most of the time....
Well I'm a bit hesitating, even if I know it might be a bad idea to keep a 10k Intel machine, that might be difficult to sell in 1 or 2 years...
Last edited:
I don’t know the answer to your dilemma, but happy to hear octane x is working well for you. Is that inside of C4D? And how stable do you find it?
In one way it is a weird time to get a 2019 Mac Pro, but surely with 2 x 6900xt will be faster at GPU rendering than m1 max MacBook Pro. And if you are making money with that tool and you happy to pay cost of mac vs pc then it sounds like for you this refurb Mac Pro makes sense.
I’m hoping iMac Pro makes a return next year and can get one of these. Although I’m very sad to see m1 macs don’t support External GPUs. So no possible way to increase render power later.
Actually I'm not properly working on Octane X yet. I'm in the middle of a project, so that is not the right time to switch. Also from the tests I have done, Octane X does not seem reliable yet for production. It does render old projects fine, but as soon as you start playing with the nodes, crashes happen often...
So that is another problem, I may have to switch to another render engine (as Vel0city did here), if Otoy can't deliver a more stable solution. But it takes some time to master an engine. We'll see how it turns with Octane X on macOS 12. Until now, Octane V4 on High Sierra has been quite stable, some crashes from time to time, but OK for production.
I haven't made my decision yet regarding the 7,1. Main thing is the poor confidence I have in selling this machine in the future. Expensive and very "niche".
If you're using Redshift in C4D don't upgrade to Monterey. Lots of crashing with forced hard reboots even with the newest builds. Had to move project over to Win system to keep going. C4D itself, including Arnold, seems to work fine.
Have you reported that crashing to the devs? They are usually really good with bug fixes.
Is Arnold M1 native yet?
Reported via Maxon Support and the Redshift Forums where someone had already started a thread.
Arnold not M1 Native yet, but has better support for the new node system than Redshift. And infinitely subsurface scattering.
Haha well...meh
No!
I decided to keep the 7,1 anyway
Looking forward to see what AS MacPro will be. A beast for sure.
And I'll probably get one of those MBP on the refurb next year
very much looking forward to seeing Redshift and Octane benchmarks on MacBook Pro M1Max. Thanks for doing this @vel0city. Has it come yet?
I don't think the YouTube benchmarkers really know how to test for GPU rendering. If you are able to test them in a way to see how they compare to the PC side aswell that would be great.
I feel excited again for the future of GPU rendering on Mac. It's all rumours but dreaming about the Mac Pro version of the M1 with its 128 GPU cores ?. And personally hoping they make an updated iMacPro early next year, cos thats what I am going to save up for.
Screenshotted these from a YouTube video, speculating what future Apple Silicon for Mac Pro. Yes please.
I don't think the YouTube benchmarkers really know how to test for GPU rendering. If you are able to test them in a way to see how they compare to the PC side aswell that would be great.
I feel excited again for the future of GPU rendering on Mac. It's all rumours but dreaming about the Mac Pro version of the M1 with its 128 GPU cores ?. And personally hoping they make an updated iMacPro early next year, cos thats what I am going to save up for.
Screenshotted these from a YouTube video, speculating what future Apple Silicon for Mac Pro. Yes please.
Should get it the week of the 8th.
I don't have a PC here but can test against a 2009 Mac Pro with 64GB Ram and a 1080ti. Great computer that has been sadly unstable since I added the 1080. But that card is so great that I've tolerated the almost daily KPs. My M1 Max has 64GB as well, will be interesting to compare.
I'm really looking forward to seeing how this Mac works out as a lookdev machine. I want something where I can fully push X-Particles and the new Insydium Meshtools, and work on scenes loaded with displacement textures, assets from Quixel Bridge, volume builder objects etc, without wondering if I'm going to get trapped in an infinite beachball loop.
From Redshift I'm hoping for snappy feedback with a reduced wait for the IPR to refresh. The SoC should do that. I'm not expecting realtime feedback, just a responsive and stable IPR that gives me faster times to first pixel. Final renders I'm not that bothered about, I farm most of my final frames these days anyway.
Lemme know if there's anything specific you'd like me to test.
Octane still isn't M1 native, but I do have an active subscription with Otoy and will test in Rosetta.
I'll give Arnold a try too, sadly not native though so will have to run C4D in Rosetta. Arnold on my M1 Mini is almost as fast as the 2009 Mac Pro. Blew me away when I first tried it.
I'm a lot more optimistic about the Mac and rendering now. CPU and GPU are both getting a significant boost. I don't think it's too much of a dream to fantasise about 128 GPU cores. It will come.
Exciting times.
Just keep in mind the above a basically "block diagrams" to see what is in the SoC, but not the actual layout of the planned silicon.
Rumors have Jade 2C (M1 Max Duo) & Jade 4C (M1 Max Quadro) as SiPs (System in Package), which would be two or four M1 Max SoCs "lashed together" to perform as if it were a single larger unit?
I am not an engineer or anything, so the ins & outs of it are beyond me...
Exciting times, [Teal'c] indeed [/Teal'c]...! ;^p
I am hopeful for a fully Metalified Blender, to run on a M1 Max (10c/32c/64GB/1TB) Mac mini!
I was about to keep my new 7,1 (16core / 48Gb RAM at purchase), with a RX6900xt (and a 2nd to come) inside for 3D rendering.
This was before I tested working in After Effetcs. What the hell is going on? I knew AE was not optimized for the 7,1, but man, this is absolutely unusable.
Spinning wheel all the time, interface sluggy as f***.
If I open a very light project with a few layers and effects, it's perfectly fine. But all the other projects I opened, from Normally heavy to Heavy, they all make the spinning wheel appear at any click. Way more sluggy than on my 5,1.
Even while doing nothing, I juste stare at Pointer / Wheel / Pointer / Wheel... every 2sec
So frustrating because beside the laggy interface, rendering on AE 22 with Multi-Frame is freakin' fast
Same problems on AE 18.x and AE 17.5
Anyone here able to work on AE on a 7,1?
Very disappointing
I have only 2 days left to send back the 7,1 to Apple
Not very excited to go laptop, even if M1 Max are very promising. At least they are here now so that should be an alternate choice while waiting for AS MacPro.
But in the meantime that means goodbye to fast 3D rendering that I would get with 2x RX6900xt...
I've been waiting for years to upgrade from my 5,1, and now this...
Gosh
This was before I tested working in After Effetcs. What the hell is going on? I knew AE was not optimized for the 7,1, but man, this is absolutely unusable.
Spinning wheel all the time, interface sluggy as f***.
If I open a very light project with a few layers and effects, it's perfectly fine. But all the other projects I opened, from Normally heavy to Heavy, they all make the spinning wheel appear at any click. Way more sluggy than on my 5,1.
Even while doing nothing, I juste stare at Pointer / Wheel / Pointer / Wheel... every 2sec
So frustrating because beside the laggy interface, rendering on AE 22 with Multi-Frame is freakin' fast
Same problems on AE 18.x and AE 17.5
Anyone here able to work on AE on a 7,1?
Very disappointing
I have only 2 days left to send back the 7,1 to Apple
Not very excited to go laptop, even if M1 Max are very promising. At least they are here now so that should be an alternate choice while waiting for AS MacPro.
But in the meantime that means goodbye to fast 3D rendering that I would get with 2x RX6900xt...
I've been waiting for years to upgrade from my 5,1, and now this...
Gosh
I have been using a 7,1 since march, 16 Core, 32Gb RAM at purchase + 128GB Aftermarket, Radeon Pro 580X 8 GB.
I mainly do 3D animation and rendering on it, we produce about 6-8 minutes of 3d animation every month.
As much as I'd like to use the GPU for rendering, I see pointless to update the video card to something more powerful as our pipeline is Maya + Arnold (CPU only in Mac). Been thinking to get and Nvidia and boot in windows for rendering in GPU, or learn redshift and move the postproduction to use the GPU in macOS. Also at least here AE doesn't use any processing on the GPU, just eats the VRAM and only some third party plugins use the GPU processing (I always keep and eye on what's going on wit iStat menus).
Anyway I do all the post processing in AE and I understand very well your frustration as I encounter these problems almost every day, AE has been very sluggish and nothing different from using a MacBook Pro 15" 2019.
At least for AE 2022 keep ing mind that many plugins are still single core and will do a bottleneck for all the multithreading.
I've found that working with .exr sequences is super slow as AE has to re-render everything every time, I always try to render all the sequences (beauty, albedo, indirect lighting, color id, sss, z depth, etc) of the timeline as .movs without any compression and work with them as movie files to create a more manageable situation. It gets super slow when I'm starting to add Depth of field effects, glows or anything that has to tax the cpu. Also I connect a 500 GB samsung SSD T5 for the cache files, I have to purge it and ram every 30-40 mins, if not AE gets even slower.
Today I had to export a 4 min video and it took more than 2 hours with media encoder, even when I have everything the most optimized possible.
So yeah that's my experience, and I always think why Adobe didn't create something better... even in AE 2022 I haven't been able to use the full CPU, just about a jump from 10-15% to 30-35% CPU usage tops... not 95-100%. I wonder if it would be better to move everything to Nuke or Davinci.