Jump to content
You must now use your email address to sign in [click for more info] ×

Poor Performance on MacOS and Apple M1


Recommended Posts

Take a look if your Ryzen 7 CPU or another nearly equal one is listed here and compare with the M1 then ...

... there is still a difference between many Desktop- and Mobile-CPUs due to clock rates and TDP etc. Mobile-CPUs in mobile devices have to fight with the generated heat when performance stressed, these quickly throttle/tack down because otherwise they cannot be cooled enough in the long run on such slim cases like Notebooks or Mini-PCs, they would otherwise always be/get too loud and in worst cases die the heat-death.

15 minutes ago, the_tux said:

... all the "positive" benchmarks from youtubers seem to be unrealistic and now I have the feeling that it was/is just marketing.

Well one has to differentiate here, they are good performers among the mobile CPU device segment, but can't yet compete with the full blown higher segment Desktop CPUs.

☛ Affinity Designer 1.10.8 ◆ Affinity Photo 1.10.8 ◆ Affinity Publisher 1.10.8 ◆ OSX El Capitan
☛ Affinity V2.3 apps ◆ MacOS Sonoma 14.2 ◆ iPad OS 17.2

Link to comment
Share on other sites

@GaryLearnTech

I hope you don't take it the wrong way, but I think your efforts are really great and  I think it's great how you're trying to solve the problem. If I could get the RAW files to 50 seconds processing time I would be super happy. 
I'm currently pissed off at Apple and all the youtubers. They present a performance that is simply not even close to being sustainable.
Maybe it's also because I don't use Adobe programmes.
I use the same settings as in your screenshot. 
The first step is also super fast for me. The second processing step, HDR Merge, takes a very long time for me and takes about 95% of the total processing time.
According to App Tamer, I have a CPU load of 5% to 70%. With you it goes over 500% and an average of 395%.
Even my maximum CPU load is far below your average.

Is it possible that the CPU has a problem?
 

Link to comment
Share on other sites

Do you already installed and run the latest Big Sur update versions?

☛ Affinity Designer 1.10.8 ◆ Affinity Photo 1.10.8 ◆ Affinity Publisher 1.10.8 ◆ OSX El Capitan
☛ Affinity V2.3 apps ◆ MacOS Sonoma 14.2 ◆ iPad OS 17.2

Link to comment
Share on other sites

@the_tux 

On 3/3/2021 at 4:59 PM, the_tux said:

The installed Version is 1.9.1 and the result of lipo: Architectures in the fat file are x86_64 arm64

I attached the 9 files in a zip file.

Affinity_Photo_Benchmark.zip 154.77 MB · 7 downloads

 

33 Seconds

Note: I do not have Chrome the stealer installed by the way.

Screenshot 2021-03-06 at 22.20.46.png

Screenshot 2021-03-06 at 22.21.02.png

Screenshot 2021-03-06 at 22.32.36.png

 

My dad always told me, a bad workman always blames their tools….

Just waiting for Ronny Pickering…..

Affinity Photo, Designer, Publisher 1.10 and 2.4 on macOS Sonoma 14 on M1 Mac Mini 16GB 1TB
Affinity Photo, Designer, Publisher 1.10 and 2.4 on Windows 10 Pro. Deceased
Affinity Photo, Designer, Publisher 2.4 on M1 iPad Pro 11” on iPadOS 17.4 
 

https://www.facebook.com/groups/AffinityForiPad

https://www.facebook.com/groups/AffinityPhoto/

The hardest link to find https://affinity.help

Link to comment
Share on other sites

@Paul Mudditt  Some minor differences in the preferences, version of macOS, etc, but probably nothing too significant.  What time do you get if you tick the Automatically Remove Ghosts option?  (I got a similar score time to yours when I left that option off.)  Since our two systems are fairly similar, I'd expect your time to come out around 45-50 seconds too.

—— Gary ——

Photo/Designer/Publisher: Affinity Store, v2.1.1 release

Mac mini (M1, 2020), 16GB/2TB, macOS Ventura 13.4.1(c) • MacBook Pro (Intel), macOS Ventura • Windows 10 via VMware Fusion • iOS: current release

Link to comment
Share on other sites

8 hours ago, GaryLearnTech said:

@Paul Mudditt  Some minor differences in the preferences, version of macOS, etc, but probably nothing too significant.  What time do you get if you tick the Automatically Remove Ghosts option?  (I got a similar score time to yours when I left that option off.)  Since our two systems are fairly similar, I'd expect your time to come out around 45-50 seconds too.

5-10 seconds more for automatic ghost removal.
 

 

My dad always told me, a bad workman always blames their tools….

Just waiting for Ronny Pickering…..

Affinity Photo, Designer, Publisher 1.10 and 2.4 on macOS Sonoma 14 on M1 Mac Mini 16GB 1TB
Affinity Photo, Designer, Publisher 1.10 and 2.4 on Windows 10 Pro. Deceased
Affinity Photo, Designer, Publisher 2.4 on M1 iPad Pro 11” on iPadOS 17.4 
 

https://www.facebook.com/groups/AffinityForiPad

https://www.facebook.com/groups/AffinityPhoto/

The hardest link to find https://affinity.help

Link to comment
Share on other sites

@Paul Mudditt  That's good, your time was in line with my timings for this more complete test.  

@the_tux  I got curious about how much the environment was impacting on Affinity Photo's performance.  You know, having iTunes playing in the background, OneDrive running both my personal and my work accounts, App Tamer measuring everything (albeit normally every 5 seconds, not the 1s I used in my earlier video), 1Password, Typinator - and a stack of others, too many to mention.

To test this, I decided to run the benchmarks while booted in Safe Mode.  In case you're unfamiliar with Safe Mode, it boots the machine with pretty much none of the third party software loaded.  On top of that, I held down the shift key as I logged in to stop the remaining third-party apps from my Login Items running.  That should take it as close to pain vanilla Big Sur configuration as I can easily manage.  In olden days you'd simply hold down the shift key at startup to boot into Safe Mode.  It's a little different on Apple Silicon: https://support.apple.com/en-gb/HT201262  

After running the booted-into-Safe-Mode benchmarks, I rebooted as normal and ran a fresh set of benchmarks and …  [cue drumroll] … there's very little difference.  I pulled your earlier screenshot and you can see the three side by side for easy comparison.

2021-03-06_22-49-40.thumb.png.34d0b0bac45c234d4c7625309f87d2bd.png

Your MacBook Pro has the same spec of M1 CPU as my Mac mini.  (It's the MBP's small brother - the MacBook Air that has a slightly lower specced M1.)

Your benchmark scores come out in the same ballpark as mine, which is to be expected.  So when you ask "Is it possible that the CPU has a problem?", I reckon probably not.  (I follow Mac-related news on a couple of sites, most days, and would have noticed if there were reports of faulty M1s being discovered.)

What is different between the two models is the RAM - you've gone for 8GB, while I went for 16GB.  I think it's possible that this might be the source of the difference.  You said "8GB RAM should not be a problem or bottleneck" but, I'm sorry, I'm not so sure we can quickly dismiss it for this HDR task.  I'd be curious to learn what time you record if you use rerun the test using just three or four of the original images, instead of all nine.   Does that make a big difference from your original 216 seconds?  I tried three (from the centre of your range, to keep things vaguely realistic) and the time came down to just under 20 seconds (inc ghostbusters).

2021-03-06_23-42-50.png.d892edc6b62632bb2118d48adedb0b48.png

If that works better for you, as it might if RAM is significant here, then repeat but try increasing by one image at a time.  At some point, I think you might well see a jump in the time you record.

The other possible difference that might be playing into this is your free disk space.  You said it was a new laptop, but not what size of SSD it has.  But more importantly, how much free space does it have?

To go back and answer your original question - Will there be any optimisation for the Apple M1 from Serif? - I'd imagine that the answer is absolutely yes.  If they identify problems in certain areas and can resolve them, or otherwise just work out faster algorithms, then yes we'll see optimisations.  I don't think it's been said anywhere that that this is it fully optimised and there's no room for any more improvement.

—— Gary ——

Photo/Designer/Publisher: Affinity Store, v2.1.1 release

Mac mini (M1, 2020), 16GB/2TB, macOS Ventura 13.4.1(c) • MacBook Pro (Intel), macOS Ventura • Windows 10 via VMware Fusion • iOS: current release

Link to comment
Share on other sites

7 hours ago, GaryLearnTech said:

 

To go back and answer your original question - Will there be any optimisation for the Apple M1 from Serif? - I'd imagine that the answer is absolutely yes.  If they identify problems in certain areas and can resolve them, or otherwise just work out faster algorithms, then yes we'll see optimisations.  I don't think it's been said anywhere that that this is it fully optimised and there's no room for any more improvement.

Affinity is already optimised with V1.8.6 for the M1, that was why Serif appeared at the Apple M1 product launch.

Versions prior to V1.8.6 were not optimised for M1, might try and install that version to see how much it slows down.

I would be interested in what other apps have been installed, ant-virus perhaps or I have seen a few people who reported slow performance with Big Sur benefit from uninstalling Keystone which gets installed rather secretly by Chrome. See Chromeisbad.com for more details on why you should never install Chrome on a Mac.

 

My dad always told me, a bad workman always blames their tools….

Just waiting for Ronny Pickering…..

Affinity Photo, Designer, Publisher 1.10 and 2.4 on macOS Sonoma 14 on M1 Mac Mini 16GB 1TB
Affinity Photo, Designer, Publisher 1.10 and 2.4 on Windows 10 Pro. Deceased
Affinity Photo, Designer, Publisher 2.4 on M1 iPad Pro 11” on iPadOS 17.4 
 

https://www.facebook.com/groups/AffinityForiPad

https://www.facebook.com/groups/AffinityPhoto/

The hardest link to find https://affinity.help

Link to comment
Share on other sites

@GaryLearnTech Thank you for the detailed report and the benchmarks. After my MBPro Mid 2015 took 250 seconds to render the images, I am considering buying a new MacBook with the M1 Chip.. But I have to be careful about the price.  

I'm leaning towards either an MBPro Air 16GB Ram /512GB SSD(my favourite) or an MBPro 8GB Rar/512GB SSD.  
I rarely use HDR. I mainly develop and edit single RAW files. 
are there big differences between the two models?  What would you recommend?

Many thanks for your help.

Nikon D7500 - Win 10 - Mac - Affinity Photo/Publisher (Betas if available)

Link to comment
Share on other sites

@Paul Mudditt

I also use Windows, but I also know Linux very well. And MacOs clearly has Linux genes ;)
I have not installed a virus scanner or Chrome. The MBP is currently in an absolutely out-of-the-box state.

@v_kyr

All Mac OS Big Sur updates are installed and only Affinity Photo is installed. Otherwise only the Apple accessories.

@GaryLearnTech

My MacBookPro has an SSD with 256GB. There are currently a good 230GB free. The HDR combination with the selection of 3 images takes me 57 seconds.
I also tested your tip with safe mode. Unfortunately, there are only marginal improvements here, which are in the lower one second range.

@all

I would like to summarise the current results:

Hardware:
- MacBookPro vs Mac Mini
- 8 GB vs. 16GB
- 256 GB SSD vs. 1TB / 2TB SSD

Software:
- Affinity Photo Version 1.9.1
- Settings for HDR combination are identical
- Settings in the Performance section of Affinity Photo are identical.
- The image material for the HDR combination is identical.
- All accessories (Chrome, virus scanner, etc.) are not active or installed.
- Installed from the AppStore

Result:
- MacBookPro takes 261 seconds, Mac Mini between 20 and 40 seconds

Summary.

Yesterday I tested various test runs with different image materials and the performance on the MBP is always worse than on a Windows device. I can't get anywhere near your test results on a Mac Mini with M1. The Mac Mini definitely seems faster with the same hardware.

The whole of yesterday was spent testing various software.  Even Filmora (video editing) is classes slower on the current MBP than under Windows.

I will send the MacBookPro back to amazon. At the end of the day, I am really disappointed with the Apple M1 product in a MacBookPro. 

Link to comment
Share on other sites

Ideally we could do with somebody with the same MacBook as you but with the 16 GB RAM to try your files to see what time they get. 

 

My dad always told me, a bad workman always blames their tools….

Just waiting for Ronny Pickering…..

Affinity Photo, Designer, Publisher 1.10 and 2.4 on macOS Sonoma 14 on M1 Mac Mini 16GB 1TB
Affinity Photo, Designer, Publisher 1.10 and 2.4 on Windows 10 Pro. Deceased
Affinity Photo, Designer, Publisher 2.4 on M1 iPad Pro 11” on iPadOS 17.4 
 

https://www.facebook.com/groups/AffinityForiPad

https://www.facebook.com/groups/AffinityPhoto/

The hardest link to find https://affinity.help

Link to comment
Share on other sites

3 hours ago, the_tux said:

I also use Windows, but I also know Linux very well. And MacOs clearly has Linux genes

Not really, both are Unix systems and MacOS evolved from NeXTstep/OpenStep which in turn had BSD Unix roots with a Mach kernel. So nowadays MacOS is more similar to FreeBSD than Linux and the superset of all of this is Unix.

As I said before, the different processing time results for the M1 Notebooks and Minis are more related to RAM sizes and SSD performance and probably to the fact that the notebooks due to their device construction do throttle much more under higher load in order to compensate heat. - Other than that, Big Sur itself is evolving and so a construction site, which still has quirks and bugs (same as Affinity).

☛ Affinity Designer 1.10.8 ◆ Affinity Photo 1.10.8 ◆ Affinity Publisher 1.10.8 ◆ OSX El Capitan
☛ Affinity V2.3 apps ◆ MacOS Sonoma 14.2 ◆ iPad OS 17.2

Link to comment
Share on other sites

  • Staff

@the_tux I would certainly lean towards RAM being the issue. From profiling the GPU, it can use roughly 5.5GB of allocated memory during the merging process with all of your files selected, and will actually shoot up temporarily to over 8GB during the initial alignment process whilst the images are being decoded and held in memory. If I select just three files (picking a dark, middle and bright exposure) that usage reduces to around 2GB.

Because of the unified memory architecture (as opposed to GPU having its own pool of dedicated memory), you would have to factor in other apps, general overheads and other requirements that will easily push the memory over 8GB and therefore into swap. I've generally found the swap is very efficient with M1—in most cases I haven't even realised it's being used—but it appears there is likely a bottleneck when using the GPU for this task.

I'm sorry your experience with M1 has been frustrating—I got one a couple of weeks ago and I was very skeptical, but was happy to be proven wrong. It's quiet, fast and my hugely expensive MPB 16" is now gathering dust! Apart from a few scenarios with CLI apps that are going through the translation layer, everything is basically quicker and snappier, especially in the Affinity apps.

2 hours ago, the_tux said:

Even Filmora (video editing) is classes slower on the current MBP than under Windows.

Again, I do wonder whether this is either an optimisation issue or a memory issue. Have you tried FCP at all? For me, it's just as fast (if not faster) than when it's being used on a high-end MPB, and without obnoxiously loud fans too which is a welcome change.

Product Expert (Affinity Photo) & Product Expert Team Leader

@JamesR_Affinity for tutorial sneak peeks and more
Official Affinity Photo tutorials

Link to comment
Share on other sites

@James Ritson

Thank you very much for the comments. 

Since I have been using Affinity Photo on the PC and iPad for years, I thought, after the many positive videos about Apple's M1 MacBooks on youtube and in the press, that I could finally switch to a Mac Book. It is well known that Intel has not always had a happy hand when it comes to performance in recent years. Especially in MacBooks, the performance of Intel CPUs was always throttled. The common statement in the tests was always that the M1s are very performant and that the 16GB only come into play when editing 8K material. What surprises me is that the HDR combination is faster under Windows with 8GB.

I ran HDR combinations again today with Affinity Photo on the 2018 IPad Pro. And the iPad is faster than the MacBook Pro. The iPad definitely has less RAM (4GB) than the MacBookPro (8GB).  I think for normal editing of RAW material, the performance (CPU and RAM) is also not that much of an issue.

I also tested DXO today. It is faster (130 seconds) than Affinity when creating an HDR combination from the 9 images. But that may be due to different algorithms. But DXO is also slower on the M1 than on the Ryzen.

So there seemed to be some potential for optimisation in Apple's M1 and the ARM architecture. 

It seems that there will be no solution to the problem on the MacBook Pro in the near future. And the Mac Mini seems to have better performance with the same hardware.

Link to comment
Share on other sites

  • 2 months later...
On 3/7/2021 at 6:44 AM, the_tux said:

Thank you very much. Unfortunately, all the "positive" benchmarks from youtubers seem to be unrealistic and now I have the feeling that it was/is just marketing.

I don’t have my M1 Mac Mini yet, but in looking at early reviews, as with a lot of photography Youtubers, they obsess about video editing performance, seeing as the are Youtubers, it’s what they do now, not so much photography. I saw this earlier in iPad Pro reviews (pre-M1) too with Luma Fusion performance outperforming large PC editing rigs with smooth 4K playback with lots of edits and export times by quite a way. That all comes down to optimised decoders in the Apple silicon, not a reflection of raw CPU or GPU in general.
And so the early M1 Mac Mini and MacBook reviews tested video performance endlessly too, with a few doing Lightroom export tests and very simple Photoshop tests, both before and after optimisation, but nothing too taxing. And of course all the synthetic benchmark tests.
The thing to remember here is that the M1 is Apple’s very first desktop CPU, and it is put into all the low-end Macs. None of the machines they replaced were particularly powerful, and all had integrated Intel Irix graphics. What has been amazing to many of the reviewers is that these low-end M1 Macs outperform earlier Pro Macs, even Mac Pros, in certain tasks not involving a dedicated GPU. GPU-heavy tasks, be it on a Mac or PC, always do far better on a system with a dedicated GPU than any integrated graphics chip, like the M1.
Again, the machines being replaced all had integrated graphics, like the M1, and in that context it is far superior to what it replaces (even the 2020 Intel 13” MacBook Pro only had Intel Irix graphics). The more powerful Pro chip(s) are about to be announced at WWDC, it is conceivable they might even do a separate GPU for thermal reasons. 
So yeah, I’m not surprised your Ryzen PC outperforms the M1 with 8Gb in certain tasks. If you were doing a video editing and exporting task, it would likely be a different story.

And yes, 8Gb vs 16Gb of unified RAM could make a difference; in your PC (don’t remember what you said it’s specs were), your graphics card (if you have one) has it’s own dedicated RAM, separate from the 8Gb system RAM.

we shall soon see what Apple has planned for their Pro machines, replacing the current Intel Macs with a graphics card. That is bound to be a major point of differentiation with the M1.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...

Important Information

Terms of Use | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.