New Apple MBP M1 + Resolve 17 - Is it worth the price?

5K$ is quite a lot. But desktop is a desktop, not a laptop which you freely move. There is also screen needed, which with good spec can cost 5K itself.
You always buy per your needs, not just because it's Apple etc. Well, at least I do it this way.

The dude lost me the moment he said he was going to be comparing the new Macs w/o optimization against the older one playing in proxy mode 1/4 resolution. That guy’s really pressing on the scales!

I’m picking up the M1 Max 16” as soon as it’s available here. And a phenomenal machine configured like so is only $3,500.00 - no need to spend five grand to get a state of the art laptop.

Apple M1 Max 10-Core Chip
32GB Unified RAM | 1TB SSD
16.2" 3456 x 2234 Liquid Retina XDR Screen
32-Core GPU | 16-Core Neural Engine

configured like so is only $3,500.00 - no need to spend five grand to get a state of the art laptop.

Hi.

In Europe do you normally tell the price inclusive of sales tax. As I remember is it between 19% and 25%.

In the US you normally tell the price without sales tax. As it is different from state to state. And American people know, to add their local sales tax to the price.

Apple have different pricelist in different part of the world, as the cost of running the local Apple shop is dirrerent from country to country. But Apple also have some price difference in healthy countrys I don't understand.

Regards Carsten.

configured like so is only $3,500.00 - no need to spend five grand to get a state of the art laptop.

Hi.

In Europe do you normally tell the price inclusive of sales tax. As I remember is it between 19% and 25%.

In the US you normally tell the price without sales tax. As it is different from state to state. And American people know, to add their local sales tax to the price.

Apple have different pricelist in different part of the world, as the cost of running the local Apple shop is dirrerent from country to country. But Apple also have some price difference in healthy countrys I don't understand.

Regards Carsten.

Everything sells for different prices around the world. For example, a Teranex 8K converter that runs $1,295 in the States costs $1,633 here in Vietnam.

OTOH, the Honda PCX scooter is about half the price in Indonesia vs Europe.
There are taxes, national politics and business politics…

I won't get that M1 Max for 3.500 $ in Europe, not even for the same amount in Euro.

Watch this first.

The new Apple MBP M1 MAX, fully configured, is impressive with Resolve. I think with optimization turned on, it will perform pretty well for even demanding projects. However, to get it to such configuration, I heard him say it's about $5000USD.

I don't mind the laptop form factor but I would have preferred it on a Mac Mini version. I also think that for $3000USD, I can build me a super duper Windows workstation with an RTX 3090 discrete GPU, a decent CPU, and 64Gb of RAM with a DecLink 4K mini monitor. If I spend $5000USD, I could have either added a 27" HDR monitor to it, or instead of an HDR monitor, I could have a Threadripper ad possibly a decent monitor. And regardless which of these build I would do, it surely will run circles against the MBP M1 MAX, or not?

Is this a matter of hardware solution or could Resolve be developed so it will be more performant even on a base computer too?

What do you guys think?

Is this just some kind of new age thought experiment or what? I think you should put your money where your mouth is and buy that PC, just as I'm going to walk the walk and pick up the Mac the instant it hits shelves here, knowing that I'm buying the best laptop in the world, one that will hold its value better than any other laptop on earth.

If I wanted to buy a laptop, it would be this one. I use the original M1 for general computing, and the platform has a lot to offer. Outside of specific use cases where you need a ton of non-specific GPU or CPU compute, it would be a very capable machine for almost all users, especially if the ProRes acceleration is nice and responsive on the timeline. I think the "debate" around it is really a reaction to the misleading performance reports more than the actual utility of the computer/processors.

I didn't find the reports very informative regarding Fusion. So, I tested my shiny new MacBook M1 Pro with the "Spaceship" project, which used to be linked on BM's product page for Fusion. It's not there any more, but I found it in one of my backups. I hope it's OK to link it here, after it was displayed for download for such a long time: http://www.dropbox.com/s/2i1vm5rjv3kqr4 ... t.zip?dl=0
(moderators, please inform me if I should take it down)

This project grinded to a halt on my wife's MB Air M1 with 8 GB VRAM. Now, on my new laptop it is rendered in a tad under one hour (58:30). For comparison, my old 2017 iMac 27" needs two and a half hours. All CPU cores are working full steam, but other than the older Intel iMac, the M1 laptop doesn't break any sweat and only one fan is slowly rotating at all (the iMac is getting really noisy).

It would be interesting if others could test this on a really strong PC.

This is an interesting test (and 'real'). Not a 30 seconds file playing test.

Here's at least a video that acknowledges that just rendering a standard video clip in Resolve doesn't tell you much. He put a bit of work in.

Yep, another 'proper' test and results are as you would expect ( at least for me). I'm actually surprised that PC desktop wasn't better in Resolve, but that just show how powerful M1 Max is. In the same time this desktop is powerful, but it's not a high-end workstation, so if you think what much better CPU+GPU could do you have quite clear image where M1 seats.

It would be interesting if others could test this on a really strong PC.

I do love a benchmark! I'm on a pretty strong PC, but running macOS.

Fusion Studio:


Resolve:


I noticed that the comp was almost entirely CPU bound. The GPU kicked in now and again, but generally the GPU was between 2 - 5% usage while CPU was up around 70-80% most of the time. Render performance degraded steadily over time. In Fusion Studio I was seeing ~2.5 seconds/frame average at the start, but this was down to 4.5 seconds/frame by halfway through. It stabilised around that figure for the remainder of the render.

59min M1 Air Pro vs. 42min 32min for fairly good workstation. If I were to say what impressed me more than it has to be M1 (with its weak GPU). Of course this is just for this specific project and fact that GPU is not used much gives explanation for those render times. Sounds okish.

59min M1 Air vs. 42min for fairly good workstation. If I were to say what impressed me more than it has to be M1 (with its weak GPU). Of course this is just for this specific project.

That was the Resolve test - it was 32:53 in Fusion Studio, which I expect is what Uli used also (I had to convert the comp to run in Resolve).

But yes, for a laptop the result is pretty good. And Uli has the 32GB Pro, not the 64GB Max, which may have performed even better.

(He didn't get 59 mins on the M1 Air - he said that ground to a halt and never completed )

Oh, ok. Then M1 Pro looks bit less impressive.

"Spaceship" project...

This project grinded to a halt on my wife's MB Air M1 with 8 GB VRAM. Now, on my new laptop it is rendered in a tad under one hour (58:30). For comparison, my old 2017 iMac 27" needs two and a half hours. All CPU cores are working full steam, but other than the older Intel iMac, the M1 laptop doesn't break any sweat and only one fan is slowly rotating at all (the iMac is getting really noisy).

It would be interesting if others could test this on a really strong PC.

With AMD Threadripper 3960X, 2080Ti and 128GB RAM it took 32 minutes under Win10. Didn't check how much was cpu how much gpu based, will run another pass with gpu use disabled. I must say 58 mins from new M1 laptop sounds pretty good.

But yes, for a laptop the result is pretty good. And Uli has the 32GB Pro, not the 64GB Max, which may have performed even better.

Max has shown being throttled to 70% when hammered on so this is questionable.

Regardless of possible throttling GPU power is much higher in Max. If project uses CPU mostly then timings would be about the same.
Also for those test you should be probably running in Power mode.

Not if you want to include titles.
Putting a simple Title+ in your timeline will destroy your frame rate.

Now that Zebra Zone guy does something serious. I can understand that he doesn't want to publish his projects as sample projects, so we can't compare with other workstations or PC laptops. It would be interesting how far a Hackintosh can hold its own on his projects where the GPU gets much more involved than in the Spaceship.

Yes, the Spaceship project is very CPU heavy, so it doesn't tell much about potential throttling when done on the M1 Max (which would probably not be much faster in this case). Regarding power mode, it might rather be a marketing gimmick. That MaxTech guy hardly found any difference.

If I find some time today, I'll prepare a small Resolve project based on footage that is public.

OK, here you go. I slapped a few demanding effects onto some clips you can download. These are in 8K, since my older iMac with only 8 GB VRAM didn't want to swallow 12K (the M1 doesn't care).

The project is here: https://www.dropbox.com/s/9wf1tl92bd7qq ... p.zip?dl=0
You'll just need to relink the clips after downloading them from BM's gallery for the 12K under "James Perse - Robertson House". (the author may forgive me for just trying some nonsense on these, aesthetically)

My trusty old iMac with an eGPU, both internal and external Radeon 580 with 8 GB VRAM, needs 44:37 to render them into ProRes 422 HQ. My mid-level MacBook M1 Pro needs 25:09. Interestingly, the GPUs in the MBP are running full speed all of the time, while the GPUs in the iMac are not as active on the last clip with the Speed Warp. As expected, the eGPU does a little bit less, like 10-15%, on the iMac. After all, Thunderbolt 3 is not that fast. And its fans are blowing full speed, while the MBP again doesn't brake a sweat and has one fan rotating lazily.

I hope someone will test this on an MBP M1 Max, which should perform better. Might be interesting to see if any thermal throttling is going on and if the power mode does anything.

P.S. Sigh, my trusty old workhorse may need to find a new job.

My trusty old iMac with an eGPU, both internal and external Radeon 580 with 8 GB VRAM, needs 44:37 to render them into ProRes 422 HQ. My mid-level MacBook M1 Pro needs 25:09.

Good test, Uli. Here's my results - 06:19:



In tests like this my system can be considered like a Mac Pro 2019 with third-party AMD 6900XT 16GB. A Mac Pro 2019 with official W6900X would also be similar, but maybe a little quicker by virtue of the double VRAM. (In CPU bound tests my system will beat the 16-core Xeon Mac Pro 2019, but not sure versus the 24-core or 28-core.)

Interestingly, the GPUs in the MBP are running full speed all of the time, while the GPUs in the iMac are not as active on the last clip with the Speed Warp.

During the render I generally saw around 75% GPU total, however interestingly it was divided between Resolve (~60%) and WindowServer (~15%):


I am not quite sure why this is. I had "Updates During Renders" = Off in Resolve, so nothing was really happening on the monitors. I am wondering if Resolve is using some GPU API that shows up under WindowServer for some reason?

As you found, during the final clip doing Speed Warp, GPU usage was much lower. Around 30% Resolve, nothing in WindowServer.

P.S. Sigh, my trusty old workhorse may need to find a new job.

If you do much Fusion work, then I've found older machines like this can still be very useful for Fusion Studio network renderers. Yes they're much slower, but they still do a useful job. Even if my main machine is 3x faster, using the secondary machine still lowers the overall time by 25%. Though of course, it might not be that great when overall power consumption is an important factor

I've currently got a 2019 16" MacBook Pro with 2.3 GHz 8-core Intel i9 processor, 32GB memory, 1TB SSD and AMD Radeon Pro 5500M and my machine crawls with Resolve Studio 17. Screenshots of a typical tiny project with minimal noise reduction, film emulation plugin and grain. 3.5FPS is typical for both playback and rendering speeds (ProRes 4444).

Screen Shot 2021-11-12 at 2.47.33 PM.jpg (825.9 KiB) Viewed 11597 times

Screen Shot 2021-11-12 at 2.49.01 PM.jpg (554.75 KiB) Viewed 11597 times

Uli, I think I see why the final clip showed so much less GPU - it doesn't have any Color effects at all

EDIT: I just realised that it's not meant to have color effects, as it uses Speed Warp. I misread 'speed warp' as 'color warp' for some reason!

That's a nice machine you've built there, Tom, thanks for the immediate report.
I wonder if what I see regarding GPUs in the M1 is activity of the so-called 'neural engines'. Isn't Speed Warp using the GPU too?

On the MBP the WindowServer is using around 5%. The two efficiency cores are around 50%, two of the power cores are lazily working around 20%, the rest is dozing off. Only the GPUs are at 100%.

And then, you're right, the iMac might still serve as a background renderer.
At least until better Mac minis are coming.

Here's at least a video that acknowledges that just rendering a standard video clip in Resolve doesn't tell you much. He put a bit of work in.

Thanks for that!

Screen Shot 2021-11-12 at 6.37.53 PM.jpg (499.32 KiB) Viewed 11533 times

Screen Shot 2021-11-12 at 6.37.19 PM.jpg (562.2 KiB) Viewed 11533 times

My 16" 24 core 64GB ram arrived two days ago. If I could be of assistance I will but I will need a bit of hand holding along the way as I'm new...even though I'm old.

My 16" 24 core 64GB ram arrived two days ago. If I could be of assistance I will but I will need a bit of hand holding along the way as I'm new...even though I'm old.

Would be cool if you could download Uli's project and then head over to BMD website and download the two media files:

https://www.blackmagicdesign.com/ca/pro ... ro/gallery

(The James Perse - Robertson House section - has two files to download - one is 2.04 GB and the other is 788 MB)

Okay I downloaded the files but I'm not sure of something. At first I put the two files on the timeline then when I opened the test file from Uli and linked it back I'm not sure if it left the files I put on the timeline before opening the dpr file. I deleted the clips from before the glitch portion of the test.

Assuming everything is okay what is the next step? btw playback was a bit choppy at the beginning then went to 23.976fps but still was a bit choppy on some of the movements.

If it's time to render this out I'd like to know what settings to use and I'll run it.

Chris

Don't make any new project or put any files on a timeline. Just do this:

0. Close Resolve, if it's open
1. Open Resolve
2. At the Project Manager screen, the first window you see when you open Resolve that says "Projects" at the top, right-click on the background and choose "Import Project"
3. Browse to the DRP from Uli
4. Give the project a new name and click OK
5. Open that new project
6. In the Media Pool, relink the two missing files
7. Go to Deliver
8. Change the "Location" to a suitable disk location for you
9. Click Add To Render Queue
10. In the Render Queue, click Render All
11. At the end, make a note of how long it took to render - or ideally, take a screenshot

That's a nice machine you've built there, Tom

Thanks, yeah I'm quite pleased with it. Saved up for a while to finally upgrade, a year ago. Before that I'd run with the same machine for 10 years

I wonder if what I see regarding GPUs in the M1 is activity of the so-called 'neural engines'. Isn't Speed Warp using the GPU too?

The last clip does use the GPU for me, but much lower. Maybe this effect isn't as optimised or something.

Uli, I think I see why the final clip showed so much less GPU - it doesn't have any Color effects at all

My bad, I realised it was meant not to have any Color - I was looking for 'Color Warp' not 'Speed Warp'

here the result of my 11 years old (2010) classic mac pro 5,1 (specs in my signature)

1.jpg (10.13 KiB) Viewed 11398 times

it plays everything in real time (incl. external video display via HDMI in UHD)

title

2.jpg (98.17 KiB) Viewed 11398 times

last clip

3.jpg (61.24 KiB) Viewed 11398 times

but also mean, you are happy that the export is so fast and then comes the last clip +g*


.

The last 16" Intel MBP with a 5600M and 32GB of ram pulled it off in 35:39. The CPU usage was around 75%, GPU ~25%. Timeline playback was 3-4fps, and basically unusable but the render is clean.

Good Luck

Screen Shot 2021-11-12 at 12.09.03 PM.png (21.42 KiB) Viewed 11382 times

Tom,
Thanks for the detailed description on to do it.

Untitled.jpg (715.28 KiB) Viewed 11377 times

Here is the screenshot. With about 1:30 left I heard the fans spin up but not much and by the end they were silent again.

All results are just a reminder that Intel's mobile CPUs were/are simply crap. Desktop wise they seems to be ok (or maybe also not, but we lack a reference point).

Has anyone managed to run Uli's test on an Nvidia GPU where it can't offload to system RAM when needed? My two GeForce TITAN X cards with 12GB and no tensor cores chokes on the last clip and halts the render. Not sure if it's a memory issue or if the effect used prefers tensor cores.

The DaVinci Resolve Neural Engine uses tensor cores if available as I understand it but should still work without them but slower.

Tom,
Thanks for the detailed description on to do it.

Untitled.jpg

Here is the screenshot. With about 1:30 left I heard the fans spin up but not much and by the end they were silent again.

what is your soft-/ hardware specs?

MacBook Pro 16” Max 24 core gpu, 1T ssd 64GB ram.
Running latest OS

MacBook Pro 16” Max 24 core gpu, 1T ssd 64GB ram.
Running latest OS

THX

close on my old cheesecrater

MacBook Pro 16” Max 24 core gpu, 1T ssd 64GB ram.
Running latest OS

THX

close on my old cheesecrater

Better than my old MBP from 2012, I could only try to run DR 15. A 3 minute clip took 20 minutes and this was only with some minor color adjustments.

When it comes to video editing and rendering in DaVinci Resolve, the 14" M1 Max MacBook Pro obliterates this $5,077 computer with 12th gen Core I9-12900K Alder Lake from Intel paired with a RTX 3080.

The 14" M1 Max used in the comparison only runs around $3,700.

In the video world, MacBook Pro is at the pinnacle.

@OP I've yet to see any $3,000 build 'running circles' around the 16" M1 Max for video editing and rendering in DaVinci Resolve.

Screen Shot 2021-11-13 at 7.52.56 AM.jpg (776.05 KiB) Viewed 11288 times

Screen Shot 2021-11-13 at 7.57.47 AM.jpg (389.91 KiB) Viewed 11288 times

And when compared to the $5,800 HP Fury G8, with Intel i9-11950H 8-core processor, 15.6-inch DreamColor display, 32GB DDR 3200 SODIMM memory (upgradable to 128GB), 1TB SSD, and Nvidia RTX A5000 GPU with 16GB of dedicated GDDR6 memory, the M1 Max once more destroyed the competition in DaVinci Resolve.

https://postperspective.com/review-appl ... -pro-2021/

Has anyone managed to run Uli's test on an Nvidia GPU where it can't offload to system RAM when needed? My two GeForce TITAN X cards with 12GB and no tensor cores chokes on the last clip and halts the render. Not sure if it's a memory issue or if the effect used prefers tensor cores.

The DaVinci Resolve Neural Engine uses tensor cores if available as I understand it but should still work without them but slower.

It shouldn't be the VRAM, since my old iMac with only 8 GB VRAM handled it (that's why I didn't use 12K for the source). Is your driver up-to-date?

Agree with all your assessments. Still think that a fully configured MBP M1 Max will be $5K. .

'Fully configured' can mean whatever you want it to mean. A 16" M1 Max with 32GB RAM runs just $3,500 and trounces PCs running over $5,000. And it runs whisper quiet.

Apple M1 Max chip with 10-core CPU, 32-core GPU, and 16-core Neural Engine
32GB unified memory
1TB SSD storage
16-inch Liquid Retina XDR display
Three Thunderbolt 4 ports, HDMI port, SDXC card slot, MagSafe 3 port
140W USB-C Power Adapter
Backlit Magic Keyboard with Touch ID - US English

Second that. You don't need to max out the internal SSD, it will never be enough for all your video anyway.
Even the RAM could be fine at 32 GB, as far as I can see from the tests by MaxTech.

BTW, I made a PC version: https://www.dropbox.com/s/afvx4zismgwes ... C.drp?dl=0

The only change is the output setting to DNxHR, since PCs don't encode ProRes in DR. Of course, the hardware encoding of ProRes in the new Macs can make a difference here.

'Fully configured' can mean whatever you want it to mean. A 16" M1 Max with 32GB RAM runs just $3,500 and trounces PCs running over $5,000. And it runs whisper quiet.

The M1 machines are very impressive for laptops, but we've seen in this thread that in GPU bound tests, they're still some way from beating a higher end desktop GPU.

My Hackintosh was about the same price as a 64GB M1 Max and it's twice as fast in Uli's test as the M1 Max 64GB, and four times faster than the M1 Pro 32GB. In the earlier Fusion composition test, which is primarily CPU-bound, it was just under twice as fast as the 32GB M1 Pro. (We don't have an M1 Max test for that, but as it's CPU bound I would expect similar results to the Pro).

Mark's 10+ year old Mac Pro with 2+ years old Radeon VII is a couple of minutes faster than the M1 Max 64 in this GPU-bound test, and 2.5x the speed of the M1 Pro 32GB.

The new M1 machines are great laptops, but from all I've seen I think it's hyperbole to say they're workstation beaters.

One of the reasons I was keen to participate in this benchmark is that I built my machine a year ago, and when the M1 Max system came out and the first benchmarks appeared I started to wonder if I should have waited a year, and got one of those instead. Then I could have had portability and top class performance. Based on these results, I'm now re-assured in my decision to build a year ago. If portability was a primary concern, the M1 Max would have been very interesting to me. But as a workstation user, I'm still glad to have a full desktop - which could be upgraded even further if I wanted (more RAM, second GPU).

We'll have to wait until next year to see what Apple does in the workstation space, like a new iMac Pro and Mac Pro.

From my experience I second that. Those fanboys should try a reality check.
They are great laptops, but they don’t beat a strong desktop, neither under Windows nor as a Hackintosh.

59min M1 Air Pro vs. 42min 32min for fairly good workstation. If I were to say what impressed me more than it has to be M1 (with its weak GPU). Of course this is just for this specific project and fact that GPU is not used much gives explanation for those render times. Sounds okish.

I think we've left reality if a two generation out of date consumer desktop (Yes an I7 is a consumer chip) is considered a fairly good workstation.

When it comes to video editing and rendering in DaVinci Resolve, the 14" M1 Max MacBook Pro obliterates this $5,077 computer with 12th gen Core I9-12900K Alder Lake from Intel paired with a RTX 3080.

I'd love to see the settings from those tests.

Or go look at the Learn Color Grading last video. He puts an older laptop against a M1 MAX.

If you really think in the real world the Max is slower than an out of date laptop but faster than an Gen 12 desktop

I think we've left reality if a two generation out of date consumer desktop (Yes an I7 is a consumer chip) is considered a fairly good workstation.

I have an i9-10980XE, not an i7. And yes it is technically a consumer chip, however it's overclocked and in general CPU benchmarking beats the 16-core Xeon W found in a Mac Pro 2019.

Also, Hendrik got the same 32 minute figure with his 24-core AMD Threadripper.

But it's still not relevant to anybody buying a new machine today.

That was a 2019 release wasn't it?