PERFORMANCE ON M2 TERRIBLE !!!! – FIX PLEASE!
THE PERFORMANCE OF CAPTURE ONE ON BRAND NEW MAC M2 MAX IS APPALLING!!!! I HAVE MAXED OUT MACBOOK PRO AND THE PROGRAM IS WORKING LIKE ON WINDOS 95!!!
YOU GUYS NEED TO DO SOMETHING ABOUT IT, IT'S UNBELIEVABLE HOW BAD IT IS ON 6 GRAND 96GB RAM MACBINE MACHINE.
THANKS
-
Official comment
We now have two Support tickets regarding slow slider movement on Apple Silicon Macs. We are looking into it. If you file a support ticket, please add information on size of viewed photo (pixel dimensions and bit depth), as well as your screen setup, as that is relevant for performance in this context.
- Number and native resolution of screens
- Scaling setting (take a screenshot of System Settings > Displays)Regarding the main other issue that is discussed in this thread, JPG export speed: If you are unhappy with it, please make an improvement request in https://support.captureone.com/hc/en-us/community/topics/6431248131613?sort_by=votes and vote it up.
-
If you're referring to the memory leak bug that we're actively working on fixing, we released an update an hour ago that should address this bug. Download it here: https://www.captureone.com/en/account/download
You were also not the only one affected by this bug. See here: https://support.captureone.com/hc/en-us/community/posts/10401192573085
Download the update and let us know in the post above whether you experience better performance.
0 -
Overall, I'm disappointed in the performance of Capture One Pro compared to Lightroom Classic on the new Apple Silicone chips. Apple doesn't even have a 4-core Mac available in it's line-up anymore, they're all 8-core and up, and Capture One is refusing to utilize multiple cores (at all) along with their GPU acceleration to improve it's overall performance in terms of speed and smoothness.
It's just lazy...2 -
Not being multi-threaded and unable to use multiple cores or efficiently use the GPU would detract from performance but can someone add some details as to the specifics of where performance is affected?
What in Capture One performs poorly? In what way? On what device? Configured how? Are there objective measures of performance in comparing Capture One to Lightroom, or Capture One to itself on M1, M2, or Intel Macs of various configuration for RAM, GPU and mass storage?
Knowing the statistics would help those of us contemplating buying a new Mac better than a general impression.
5 -
+1 for some specifics.
I'm still using a 2013 MacPro. Capture One works fine on this, at least when adjusting images, although it sometimes stalls for very strange reasons (like the first time I try to load a 3rd party profile after launch. It goes unresponsive for several minutes while apparently trying to read a ColorSync profile). And the library, even on internal SSD, is not always lightning fast. But generally it's quite ok. I'm considering anyway upgrading to an M2 Mac Studio, but if it is demonstrably true that Capture One is actually worse than on a 4 core Intel Xeon, then I won't bother...
0 -
My personal experience with silicon chips and Capture One has only ever been a pleasant one. I can't comment on the individual experiences you're all having, but you should absolutely make our support team aware about this.
However, the 16.2.2 update we released on Tuesday should address the memory leak issue. Download the update and if you're still experiencing issues, let the team know.
Have any of you also tried running Capture One via Rosetta? Do you see an increase in performance? If so, then there's definitely something there that indicates Capture One isn't running as it should on M1/M2 chips. You can open an app using Rosetta by right-clicking on it, selecting "Get Info" and then ticking the box which says "Open using Rosetta"
1 -
ArtisRight on YouTube probably has the most detailed reviews based on standardized tests of C1, LR, PS etc. on Intel and Apple Silicon Macs. He continuously points out the lack of optimization of C1 for Apple hardware. One of his remarks is that he would love Capture One being optimized for non-tethered use (via a preference in the settings I suppose); i.e., he suspects that some performace penalties arise from its roots.
Maybe C1 is already in touch with this guy, but if not, he could provide constructive feedback to help improve certain issues I am sure.
I have a Late 2015 27" iMac 4 GHz i7 quad core with 40 GB RAM and SSD and performance is decent although sluggish sometimes when working on an image (I don't care so much for import speeds). Seeing the lack of progress in performance in the above mentioned reviews I am not inclined in investing 3 or 4k euros in an M2 Pro Mac Mini or Max Studio plus a 27" monitor to see only minor improvements.
It's not that Adobe and Apple are best palls either – but Adobe, and, more to the point, other much smaller app developers, manage to squeeze much more performance out of the SoC-based Macs. We're nearing the 3rd generation of Apple chips and if things remain as they are, and as much as I prefer C1 over the other options, I'll regrettably have to jump ship...
1 -
Back History:
I used an intel i7 x7820x 8-core Windows PC with a Sapphire Pulse Radeon RX 580 8GB Video card. That was the smoothest and most efficient performance of Capture One I've ever used - back in late 2017 through early 2020. It was significantly smoother and faster than my old 2009 Mac Pro and 2015 MacBook Pro - not surprising.When I moved back to Apple with the 8-core 5K iMac, with the Radeon Pro 580X 8GB GPU, it was literally almost spec for spec identical with the Windows 7820x workstation, I was expecting similar performance. It wasn't. It was slower in performance in terms of rendering thumbnails, and slower at exporting files. It was also jittery - the sliders were not nearly as smooth and effortless as my custom build Windows 10 Pro workstation. The sliders were....jumpy.
Now that I have Mac Studio, the same issues persist as far as smoothness. It still feels jumpy. The sliders aren't smooth. If you move the exposure slider, you have to wait for the effect of the slider movement to catch-up. The PC sliders (especially noted on the exposure sliders) on the Intel i7 7820x where instantaneous and buttery smooth.
Pre-Apple on Silicone, Lightroom used to be much slower in terms of imports, rendering and exporting files. It never made efficient use of GPU acceleration, much less multiple threads/multiple cores. Lightroom and Photoshop were all based on single or double core processing with high clock speeds. Capture One had a definitive performance gain over Lightroom with their "GPU hardware acceleration".
Now, with Apple Silicone, Adobe is not only making use of GPU acceleration, they're also leveraging multiple cores - and scaling them appropriately.
If you don't follow Art Is Right on Youtube, you probably should.
Real world bench marks from ArtIsRight...
MacOS Venture 13.4
Lightroom Classic (version 12.3/12.4) Export - full acceleration
1000 Nikon D850 (45MP) RAW to JPEG - GPU and CPU based task
M1 Max 10/24, 32GB RAM, 512GB SSD = 9m 4s
M1 Ultra 20/48, 64GB RAM, 1TB SSD = 4m 48s
M2 Max 12/30, 32GB RAM, 512GB SSD = 6m 57s
M2 Ultra 24/60, 64GB RAM, 1TB SSD = 4m 3s
Lightroom CC (version 6.3.1 and 6.4) Export
1000 Nikon D850 (45MP) RAW to JPEG - GPU and CPU based task
M1 Max 10/24, 32GB RAM, 512GB SSD = 7m 42s
M1 Ultra 20/48, 64GB RAM, 1TB SSD = 4m 8s
M2 Max 12/30, 32GB RAM, 512GB SSD = 5m 50s
M2 Ultra 24/60, 64GB RAM, 1TB SSD = 3m 45s
Capture One (Version 16.2.13) Export
1000 Nikon D850 (45MP) RAW to JPEG (fullsize) - Hardware Acceleration/GPU Export
M1 Max 10/24, 32GB RAM, 512GB SSD = 22m 27s
M1 Ultra 20/48, 64GB RAM, 1TB SSD = 18m 17s
M2 Max 12/30, 32GB RAM, 512GB SSD = 18m
M2 Ultra 24/60, 64GB RAM, 1TB SSD = 16m 57s
....Barely a minute difference here, between the M2 Max and M2 Ultra options. That's ridiculous.
Not only is Capture One significantly slower in performance, Capture One doesn't scale remotely close from 24core GPU to 48 core GPU... I won’t even bother suggesting anything about the CPU, since that doesn’t seem to be in play at all with C1P.
When the M1 Max base Mac Studio with LR out-performs the M2 Ultra with Capture One Pro by nearly 9 minutes, that should be embarrassing….
4 -
A few months later nothing has changed. On my M2 Max Studio only 1 or 2 cores are used during C1 image operations =/
2 -
Shame on me for updating my machine and using the latest software updates too ...
I can't edit a single image in C1 without at least a half second lag when adjusting a simple slider like exposure. I'm a wedding photographer and you can imagine how infuriating it would be to try to adjust multiple sliders in a single image, sitting and waiting for the screen to update after I move the slider that long for each adjustment is appalling. And then try to do that across 850+ images for a single wedding gallery. No thanks, I'm out!
Something must be done to fix this issue - I'm back to Lightroom (again) until this can be sorted out.
My machine / specs:
MacbookPro M2 Max, 32gb memory, 2tb ssd, Sonoma 14.0
Capture One Pro, build 16.3.0.76, using Sessions, applying a custom Style which includes an ICC profile from The Archetype Process, and editing the image from there.
Happy to provide more details on what I'm doing workflow-wise if that's helpful.
1 -
I bet you expected, justifiably, that Capture One would work better on the M2 MacBook Pro than it did on your previous setup. If it is worse, something is wrong that should be correctable. This is a user form and all we can do is tell you what we are experiencing.
I am running Capture One 16.2 with a 65k+ catalog of images using an M2 Mac Studio Ultra with OS 14.1 with 128 GB RAM and 2 TB SSD internal storage with my Capture One database and RAW image files on an external OWC Pro SX Thunderbolt SSD.
I do not have any lags when moving sliders using the Mac Studio. Moving the adjustment sliders was very smooth. Going from one thumbnail to the next in an album takes less than a second to fully resolve. Since Capture One loads the next image in the background, this is practically unnoticeable unless you move very quickly from image to image.
This is quite an improvement from my Intel Macs with Capture One version 23 (16.2). I should note that that the jerkiness appeared only after I upgraded to version 22 and remained with version 23. Specifically, after upgrading to version 22, this became a problem with my iMac Pro with OS 14.0 (and previous OS versions) with 64 MB RAM and the same storage setup as the Mac Studio or my MacBook Pro with 32 GB RAM. With the iMac and MBP, sliders were jerky, much as you describe and moving from thumbnail to thumbnail took 2-3 seconds to show a fully resolved image on the viewer. This was no better with my test catalog with less than 100 images.
My advice would be to create a support ticket and let the support folks work this out. They are much better prepared to figure out the cause. Some of it may have to do with the coding changes, but your M2 MacBook Pro should be able to handle it much better.
0 -
A follow-up on my post above. The following is all on Mac.
2020 Mac-Mini i7 Hexacore with 64GB RAM and using an eGPU AMD RX6700. Capture One 22 Pro v15.434
Apply lens corrections to (say) 60 images. While it is doing its thing, open Activity Monitor app and notice that all the cores are being used.
2023 Mac Studio M2-Max with 64GB RAM. Same operating system (Mac OS 14.1) and same version of Capture One. Same task. Open Activity Monitor and this time only 1-2 cores are active, the others unused.
Disappointing to say the least.
0 -
I've also seen somewhat spotty performance when adjusting sliders on my M1 Air (16GB RAM, 7 GPU Cores), but only when editing plugged into my 4K external monitor. These issues never arise when I'm editing solely on the built-in display, so I had assumed it was simply a limitation of my system—benchmark-wise, I know my system has the lowest GPU performance of any Apple Silicon system—but if similar issues still present themselves on higher-end systems, I'm a little worried. I plan to upgrade to either an M3 or M3 Pro system soon primarily for a better display and more storage; it would be frustrating to see Capture One performance not scaling somewhat with the significant increase in CPU + GPU power.
0 -
My CPU history shows all (24) CPU cores performing when adjusting exposure sliders and when applying exposure changes to 15 images. It appears that Capture One 16.2 is using all CPU cores on this system for this task. I have not tested other tasks. Also, activity monitor will only show an amalgamated graph of GPU usage.
My system: Capture One 16.2 with a 65k+ catalog of images using an M2 Mac Studio Ultra with OS 14.1 with 128 GB RAM and 2 TB SSD internal storage, 24 CPU, 60 core GPU with my Capture One database and RAW image files on an external OWC Pro SX Thunderbolt SSD and a Mac Studio 5k monitor and a 4k Dell monitor.
0 -
Just downloaded the trial version of Mac C1 Pro v16.3.0.76
Noticed it runs a bit faster, but again with Activity Monitor open only 1 or 2 of the CPU cores (out of 8+4) are being fully used. I have the "unbinned" version of the M2-Max chip with 38 graphics cores. I notice these are being heavily utilised however.
Appears it's time to upgrade to 16.3.x =/
0 -
I have the same issues on a MacStudio M2 Ultra and my MacBook Pro M1 Max 16". Very slow and jumpy with sliders especially on GFX 102 mpx files. It is unusable! I keep getting the runaround from anyone I ask who works for C1. It's an issue and a big one at that. Lightroom Classic is smooth on both machines. Do I have to use Lightroom Classic again? All of my shoots are in C1 Sessions. C1 keeps ignoring the issue which tells me they probably don't care or aren't even optimizing for multicore Apple chips. Oh and yes I made 2 support tickets which were both ignored.
0 -
See 2 entries, above. Capture One 14.2 is quite snappy on my Mac Studio Ultra. No sluggishness with sliders. 2 seconds to search All Images with one keyword in the filter. Half a second to show a resolved image in viewer when clicking on random thumbnails. Imports at the rate of 2/sec for ~50 MPx images.
Why results vary so much between similar systems is not readily explainable. If the software runs well on one Mac Studio Ultra and not on another similarly configured one, it seems that the problem relates to the interaction of other process competing for resources with Capture One. While this may not let Capture One off the hook, the problem seems more likely to be within the computer.
Have you tried creating another user not running any other processes than the default ones, installing Capture One for just that user, and seeing if this helps? Have you tried a complete uninstall and reinstall of Capture One? Of course the advice of users is no substitute for a response from support staff.
0 -
John Arborio Your support ticket was not ignored. We have *never* ignored a support request and never will.
I can only see one support case made on your account, which was solved within 12 hours.
Please contact our technical support team and describe the issues you are facing here: https://support.captureone.com/hc/en-us/requests/new
I encourage anybody else who is having performance issues to do the same. I have been using Apple's silicon chips ever since they came out with Capture One and have experienced nothing but great things – so there must be something local on the machines in question that we need to look into.
Also, I would encourage anybody running an old version to try the latest version (16.3) and see if there are any issues there. We have made multiple performance improvements since some of the versions mentioned in this thread.
0 -
I thought I submitted 2 tickets. The issue was never resolved. I had a M2 Ultra but now consolidated to a MBPM1 Max 16”. The same issue persists on both machines set up new without any other software open. I’m not sure what it could be. I didn’t bother to upgrade to the M3 Max because I don’t think C1 will run any better.
0 -
Jacky Jacky - Is this thread being ignored by the developers because the users haven't submitted a ticket - even though this is a Capture One "Support forum". Then what's the point of the messageboard?
And you're saying ArtIsRight's performance issues in comparison with either LR option is wrong?
There upwards of 24 cores - soon to be 32 cores on these new Apple M chips, and Capture One Pro is saturating the 2 efficiency cores and about 70% saturation of a single performance core on 10-core M1 Max chip.
Does this seem like an efficient way to use the resources of both the GPU cores and the CPU cores.The export below is from 1000+ Nikon Z6 II files.
Edit: Just updated to 16.2.6.7 and rebooted my Mac Studio - and this update is less efficient at exporting than 16.0.1.17 - (it's slower)Art Is Right's performance numbers (as posted above)...
Lightroom CC (version 6.3.1 and 6.4) Export
1000 Nikon D850 (45MP) RAW to JPEG - GPU and CPU based taskM1 Max 10/24, 32GB RAM, 512GB SSD = 7m 42s
M1 Ultra 20/48, 64GB RAM, 1TB SSD = 4m 8s
Capture One (Version 16.2.13) Export
1000 Nikon D850 (45MP) RAW to JPEG (fullsize) - Hardware Acceleration/GPU ExportM1 Max 10/24, 32GB RAM, 512GB SSD = 22m 27s
M1 Ultra 20/48, 64GB RAM, 1TB SSD = 18m 17s
Lightroom is far more efficient than Capture One Pro at import and export. ...it would be fantastic if we could always export files at night, and walk away. But some of us need to export multiple large batches during the day. I'd rather it take 5-7 minutes than 20-25 minutes.1 -
They know their software is lagging far behind and won’t admit it. It would me more encouraging if they said they are working on it instead of making me go through steps they have nothing to do with fixing the problem.
0 -
Others and I have posted references to ArtisRight's testing in the past as well. A returning remark from Art is that he wonders whether the performance issue may be due to its origins in tethering. Almost every video he suggests C1 might add a toggle to set-up the program as a batch program vs a tethered one. I am not a developer, so I can't judge if this makes sense from an architectural point of view. I also wonder whether intel-era architectural decisions are not hampering the software to exploit Apple silicon to its fullest potential. Another question is whether support for older, i.e., intel-based, Macs isn't hampering both the speed of development as well as the performance (on Apple silicon). Also, at what point is C1 going to stop supporting non-Apple SoC with future releases? (If there is a performance and/or economic penalty for keeping intel-Macs supported in the latest version, isn't this a potential win?)
I also agree with craig stodola's remark:
Is this thread is being ignored by the developers because the users haven't submitted a ticket - even though this is a Capture One "Support forum". Then what's the point of the messageboard?
We have had to wait a long time before getting a clear answer on the potential support for Hasselblad, and although it's not satisfactory to some of us, at least it has brought clarity. I suggest C1 would have transparent and timely communication concerning these performance issues as well as the roadmap for Apple hardware support. This (perceived or factual?) non-communication is further fueling frustration and will do more harm than good, e.g., causing users to lose trust and increase abandonment of the platform. In a previous post, I already suggested getting in touch with ArtisRight. He seems very knowledgeable about photography tech, and he is a no-BS in-depth tester of the major apps on Apple SoC.
Maybe C1 could also be supported by some of us acting as beta testers?
Just my 2c
Erik
1 -
Art Is Right - Latest export benchmarks with updated software, including the M3 Max chips
Lightroom CC 7 (using CPU + GPU + RAM) Export
1000 - Nikon D850 (45MP) RAW to JPEG fullsize (240ppi)
16MBP M1 Max 10/32 64GB - 6m 57s
MS M1 Ultra 20/48 64GB - 4m 10s
16MBP M3 Max 16/40 48GB - 4m 34s
Capture One 16.3.0.76 - GPU acceleration/hardware export
1000 - Nikon D850 (45MP) RAW to JPEG fullsize (240ppi)
16MBP M1 Max 10/32 64GB - 22m 27s
MS M1 Ultra 20/48 64GB - 21m 6s
16MBP M3 Max 16/40 48GB - 18m 46s
There's so much good with Capture One, then they totally sandbag it when it comes to multiple core utilization. ...and they charge $200 for a loyalty user upgrade.
I've been using C1P since version 5. And their best perpetual license "loyalty program" extends to those who bought the latest license within the last year - aka, even a 1-time buyer. Not a loyal 14 year user.1 -
craig stodola It's a user to user forum, not necessarily a support forum. If you need support then you submit a support request and detail your findings there, so that they can be actioned on with the audit team and eventually the developers.
We don't have the staff (it's just me and Denis Huk in an official capacity) to provide technical support, but I do provide solutions where I can. I've nothing to offer here at such a technical and hardware specific level, so it's better that your issue is handled formally and by the right people. I hope that's agreeable.
-2 -
Have any of you tried running Capture One using Rosetta to see if there's any impact on performance? That's a quick and easy way to see if performance on silicon chips is sub-optimal.
As mentioned above, please contact our support team with your findings, so that a bug can be logged and reproduced by the QA and eventually fixed by the dev team.
Posting on the forum does great things for discussion, but if there is an issue then we need it logged and worked on in the correct way.
-1 -
@JackW - It's not a matter of the Developers providing technical support here. It's a matter of them using this forum as a resource and being aware of user issues. Technically Capture One still works.
The performance of Capture One sucks compared to it's top competitor - especially when it comes to exporting images. It shouldn't be exceedingly difficult for them to figure out that they're not utilizing 7 of the 8-performance cores and address this disadvantage without a support ticket.
If they literally need users to spell this out for them, then it's an epic awareness failure on their part. ...Not to mention it's an epic failure on the Marketing department, considering they should be monitoring their top competitor's performance standards, and figure out where they have shortfalls so they can stay competitive.
Even with ArtIsRight's benchmarks, which are a fantastic resource in and of itself; At the very least, Capture One should own a subscription to Adobe Lightroom, and be able to test your software vs. theirs to find performance shortcomings and or learn new features, or figure out what you're both missing which you can add to make Capture One a better solution.
I do want to add: From a marketing stand point, Art Is Right's YouTube Channel is growing. His performance tests are now grabbing between 20k and 40k views. ...and he's comparing Lightroom and Capture One side by side. And he's showing Adobe is tripling Capture One terms of exporting performance. This may not seem like it matters, but there are upwards of 40 thousand users who being swayed which software to buy/switch to as much as deciding on which computer to buy as a result of Art's performance tests.
....I feel like Capture One should be paying me for this information...2 -
craig stodola I've made the relevant teams aware of this post and the discussion around performance on M2 devices.
We still need to document the behavior you and others are experiencing – I've outlined the way the communicate that formally to us above.
We haven't experienced any of the performance issues outlined in this thread in-house, nor have I heard of performance issues like this anecdotally. If 40,000 users (a huge chunk of our user base) were being affected then it would certainly show up in our analytics even before a support case was made. Hence why I (and the rest of the team) need those experiencing this behavior to communicate it through the right channels.
-1 -
Jack W We're all on the same team here, just trying to sort out a long-standing issue.
Others and I have referred several times to Youtuber ArtisRight. He's arguably the most in-depth tester of CaptureOne, Lightroom, Photoshop, and Final Cut Pro. He recently realigned his testing procedures, including many M1, M2, and now M3 machines. The depth and breadth of testing for photo apps on Apple silicon is unmatched. He's been reporting the same issues for a few years: suboptimal use of extra cores, almost no scaling with increased cores, etc. in C1. The export stats for 1000 photos show this very clearly. He is using 15+ Macs in various configurations. This and user reports in this forum should count for something. May I suggest the developers contact this guy, at least?
It worries me to read your "40000 users don't seem to have a problem" argument. Firstly, not every user is a pro, and even if they are, they might be on intel, or they may not have to perform tasks in volumes where bad performance and non-scaling are an issue. Secondly, not every user is willing to spend a few hours to give feedback in user forums. Lastly, are you saying that you collect very broad user stats all the time for all tasks that show export times – to refer to just this one clear problem – that are in line with Lightroom (or better) and that for those users, many more or all cores are being used all the time? In other words, are the reports here and the test from ArtisRight outliers?
Even if this were the case, and if I were in marketing at C1, I'd contact ArtisRight mighty fast to help him correct these freak-outlier bugs on his dozen test machines because I'm sure many more people see his tests than these forum posts. And they do so to decide whether or not, to commit to C1.
4 -
I’ve wasted enough time trying to use Capture One for anything other than a tethering app on my M1 Max, M1 Ultra and M2 Ultra machines. It’s a shame you won’t admit the issues and fix them. I don’t have time to screen record and become a beta tester for you. Anyone with an Apple Silicon machine can repeat this issue. How about you go buy one, throw a 200 mb raw file into it and experience it for yourselves.
0 -
I have submitted a ticket. With this info. Including this screen capture. I haven't heard a thing a word...
If you're telling me, they're looking at this screen grab, and not seeing the problem, or they're running tests, seeing this same grab, and pretending it's not a problem. Then THAT'S the biggest problem.2
Please sign in to leave a comment.
Comments
94 comments