Skip to main content

⚠️ Please note that this topic or post has been archived. The information contained here may no longer be accurate or up-to-date. ⚠️

Performance slowdown

Rüdiger5
Hey guys!

First of all I have no big issue, bug etc. to report. But over the past two years I noticed a massive slowdown of the general performance (including render, editing etc.) Back in the days I got my brand-new MacPro and used Capture One 8.x, Sony 7R and Nikon D800. It was nearly impossible to for example measure the speed of rendering jpegs cause it was so fast. But today everything feels much more sluggish and rendering of one jpeg takes about four seconds …


My setup these days:
http://www.nehmzow.de/files/screenshot_forum.png
Capture One 9.1.1 (latest build)
Sony 7R II
for example: JPG-render 4 seconds

Monitoring says that just 10/15% of the hardware is used.
http://www.nehmzow.de/files/screenshot_forum2.png

What happened?! Where is all the great performance I had two years ago?

Hope you guys can help me. Any advice to increase the performance is welcome. 💡

Thanks in advance,
Rüdiger
0

Comments

8 comments

Date Votes
  • HansB
    I see 60% to 80% CPU usage with openCL OFF. And, just like you, I see 10% to 15% CPU usage with openCL ON because the work is done by GPU. So I assume your openCL is ON.

    Did you already compare openCL ON against openCL OFF? I am curious about the results.
    The GPU is slightly faster here than the CPU, but I'm on a 2014 Macbook Pro, not a 2013 Mac Pro.

    It might be a good idea to remove ICOCL.bin from /Users/Shared/Capture One/ImageCore and therefore force CO to build a new one. CO compiles this file for openCL access to your hardware.


    Regards,
    Hans
    0
  • Wesley
    How much space left in your HD?
    0
  • Rüdiger5
    ok ...I removed IOCL.bin but no noticeable change in performance.

    My volumes >> http://www.nehmzow.de/files/hd.png
    - HD, boot volume containing all applications
    - work, is holding all capture one sessions


    Anybody some suggestions? Thanks in advance !
    0
  • HansB
    [quote="HansB" wrote:
    I see 60% to 80% CPU usage with openCL OFF. And, just like you, I see 10% to 15% CPU usage with openCL ON because the work is done by GPU. So I assume your openCL is ON.

    Did you already compare openCL ON against openCL OFF? I am curious about the results.
    ...


    Just to make sure it's not the GPU that slows your system down. Please set openCL in the CO preferences to 'never', 'never', restart CO and try again. OpenCL support increased a lot during CO and OS X releases. Could be it became slower the more openCL was used.


    Regards,
    Hans
    0
  • dredlew
    How long ago since you had a clean install? I don't think it would make a 4sec difference but if it's been several years since the drive was wiped, it may be fairly fragmented. Given it's an SSD though, it shouldn't be that noticeable.

    What is more likely, and that's just my assumption, is that your catalog has significantly increased in size. This initially used to be an issue with Aperture and iPhoto as well, the larger the catalog, the more it would slow down the whole app. Back then, it was advisable to split up the catalog in multiple smaller ones to maintain performance. However, Apple fixed that at some point and since then, catalog size doesn't matter anymore. Whether you have a few hundred or a few hundred thousand images, it will not affect performance.
    C1 on the other hand, still has a long way to go when it comes to catalog performance. And based on various reports on this forum, large catalogs can not just bring the app but the whole computer to a crawl. There are also several reports of catalog corruption with large catalogs...

    Personally, I haven't yet experienced this much, since after reading these reports, I've been creating many smaller catalogs instead of one big one. So far, so good. What is the size of your catalog?
    0
  • Rüdiger5
    Thanks everybody for commenting so far... But:
    - its a quite fast SSD inside a mac pro, so there should be no defragmenting...
    - I don't use catalogs, only sessions not that big, less than 2000 images each session
    - turning open cl off leads to no reasonable difference in previewing or snappiness of the application, but rendering takes about 50% longer... so this is no gain, except all processor are used now 😉

    My assumption is, that many changes under the hood in OS X handling open CL together with changes at CO 8 > 9 > 9.1.1 and maybe changes at Sony 7R2 raws are responsible, but officially nobody is talking about performance decrease like that. So ...
    HELLO PHASEONE !!! What do you think??
    0
  • Christian Gruner
    Please contact our support. Be sure to provide:

    - Screencapture video of CO 8 vs CO 9 doing the same adjustments and processing. This way, our Supporters can see your exact flow and details.


    A note on the CPU load: when using OpenCL (GPU), the CPU is only used for loading/decompression and saving. All render work is done by the GPU, which OSX does not have any native tools to show the load on.
    0
  • SFA
    I know nothing about Macs but I have a Windows PC with 2 SSDs installed.

    Firstly, having done some SSD performance reading, the smaller drives are generally not as fast as the larger ones.

    Secondly the rated performance may not extend very far into a slightly large batch before things slow down - it's the way files are usually handled and the automatic disk management and in-line "defragmenting" where attempts are made (or should be made) to extend the life of the disks by spreading the write load.

    And from my own experience with the 500Gb drive (Samsung 930) on SATA3 and the mSATA 1TB drive in a spare comms slot using only SATA2 (half maximum rated speed) output file writing performance seems fine BUT improves noticeably if the disks have more than about 20% free space. This is most evident for the system disk.

    I am using smaller files so direct comparisons are not easy and the the performance difference may not be so obvious. My outputs, even producing multiple output files of different sizes, do not take 4 seconds per file - more like 1 second or less - but there does seem to be a difference when more free space is available. Of course what has been done in the editing is also a factor - so I have no comparative test results to offer precision.

    You might be able to spot any really large differences by analysing the appropriate log file.


    HTH.


    Grant
    0

Post is closed for comments.