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⚠️ Please note that this topic or post has been archived.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. ⚠️

Capture One Pro 7 Performance Problem!

Permanently deleted user
It seems that CO7's performance has taken a significant dive which makes it much less responsive when trying to adjust image settings. The CPU goes to max constantly. It gets worse if it is also processing images to JPEG in the background. There is a delay in showing the adjustment and moving from photo to photo. Sometimes the delay can be a few seconds.

I was processing 169 CR2 images shot by the Canon 5D and 5D Mark II.

System information:

Intel i7-870
16GB RAM
OS drive 240GB SSD
CO7 installed on SSD
Image working directory on SSD
AMD Radeon HD6870 with 1GB memory
OpenCL enabled
Windows 7 Ultimate x64

The same system performs much better with CO6.

Is this a known problem?

Are there any settings that can change to help the performance?
0

Comments

87 comments

Date Votes
  • Cuong Tran
    Agree with Grant's suggestion. We are in the dark as to HW requirements. I could have chosen a much more expensive path (upgrading to 3930K processors) and still am not happy with the solution. Hope that GTX660ti would suffice. But others might have different issues. Geeze. C1 v7 requires serious HW investment.
    0
  • SFA
    [quote="Cuong1" wrote:
    Agree with Grant's suggestion. We are in the dark as to HW requirements. I could have chosen a much more expensive path (upgrading to 3930K processors) and still am not happy with the solution. Hope that GTX660ti would suffice. But others might have different issues. Geeze. C1 v7 requires serious HW investment.



    I tend to agree about the hardware investment from what I have read - but for me I face that anyway as my ageing systems are all 32bit. I don't think it is worth investing to upgrade them - one might just creep in at the lower end of the requirements specification.

    However this is not just a C1 problem.

    For some years I have been using a business application and I have been a beta tester for the provider. We are about to test the next version. Major internal engine changes (sounds familiar?) and in an efort to extend scope (database sizes that can be worked with) and peformance (more memory, etc.) everything points to 64bit and uprated hardware in the medium to long term. In the case of that application there will still be a 32bit version but no one is expecting great performance from it. For some people that may not matter too much - small tasks will likely still be quick enough. But for major projects, large files, high volumes - the needs will be very similar to C1 (except for graphics handling via GPU afaik!)

    It seems that we have just reached a point where software development needs this large jump (rather than incremental changes) to make any sense in the marketplace of the near future. If we want to follow the lead we have to invest. If not, well we can still work but maybe with a little different output or somewhat less speed. That's the choice. It's kind of the same when looking at camera bodies ...

    Timing here is quite good for me. I would prefer not to have to upgrade machines BUT at least I have 2 good reasons for doing so at this time and my last investment two years ago avoided th high cost of a then "top of the range" system which now migh be considered low end spec.!

    Moreover from what I understand so far it seems to make sense, for my purposes, to not worry too much about the GPU spec. for now. If so that makes for a better understanding of where to invest for optimal performance right now and, presumably, an even better return on any later investment should a GPU upgrade prove to be desirable and justifiable.

    However I think we need to better understand the very complex inter-relationships in order to make the wisest investments. And hope that upgrading does not become an annual event!

    If it does then maybe I should take up something requiring less investment - ocean yacht racing perhaps ... 😉


    Grant Perkins

    ETA - I'm interested to hear how you find the 660ti. It's a card I am considering when I get around to the desktop upgrade.
    0
  • Christian Gruner
    [quote="Alain" wrote:
    Christian

    Thanks for the info.

    I have a question: Gives opencl a really noticeable speed improvement when culling pictures?

    More specific I mean choosing between different pictures that where made in a row (for example 5 images from a specific model/pose), where there's a fast switching between pictures and zooming in and out to 100% (or more).

    I do understand that opencl is disabled when a local adjustment is used in a picture.

    Second question: What is the speed improvement when going from a i5 to a comparable i7 (same Mhz). The pure CPU power is about 50% more, but that's not the only factor.

    Alain


    1st question: in short, yes.
    1½: OpenCL works with layers.
    2nd: I have no measurements to relate to, so I can only speculate. Speculation in this case will be that the I7 usually have more cores, and thus more processing power.
    0
  • Christian Gruner
    [quote="SFA" wrote:
    [quote="Cuong1" wrote:
    Agree with Grant's suggestion. We are in the dark as to HW requirements. I could have chosen a much more expensive path (upgrading to 3930K processors) and still am not happy with the solution. Hope that GTX660ti would suffice. But others might have different issues. Geeze. C1 v7 requires serious HW investment.



    I tend to agree about the hardware investment from what I have read - but for me I face that anyway as my ageing systems are all 32bit. I don't think it is worth investing to upgrade them - one might just creep in at the lower end of the requirements specification.

    However this is not just a C1 problem.

    For some years I have been using a business application and I have been a beta tester for the provider. We are about to test the next version. Major internal engine changes (sounds familiar?) and in an efort to extend scope (database sizes that can be worked with) and peformance (more memory, etc.) everything points to 64bit and uprated hardware in the medium to long term. In the case of that application there will still be a 32bit version but no one is expecting great performance from it. For some people that may not matter too much - small tasks will likely still be quick enough. But for major projects, large files, high volumes - the needs will be very similar to C1 (except for graphics handling via GPU afaik!)

    It seems that we have just reached a point where software development needs this large jump (rather than incremental changes) to make any sense in the marketplace of the near future. If we want to follow the lead we have to invest. If not, well we can still work but maybe with a little different output or somewhat less speed. That's the choice. It's kind of the same when looking at camera bodies ...

    Timing here is quite good for me. I would prefer not to have to upgrade machines BUT at least I have 2 good reasons for doing so at this time and my last investment two years ago avoided th high cost of a then "top of the range" system which now migh be considered low end spec.!

    Moreover from what I understand so far it seems to make sense, for my purposes, to not worry too much about the GPU spec. for now. If so that makes for a better understanding of where to invest for optimal performance right now and, presumably, an even better return on any later investment should a GPU upgrade prove to be desirable and justifiable.

    However I think we need to better understand the very complex inter-relationships in order to make the wisest investments. And hope that upgrading does not become an annual event!

    If it does then maybe I should take up something requiring less investment - ocean yacht racing perhaps ... 😉


    Grant Perkins

    ETA - I'm interested to hear how you find the 660ti. It's a card I am considering when I get around to the desktop upgrade.


    HDR HW acceleration will come, but I can't comment on when at the moment.
    But for users not using HDR, they will benefit greatly, performance wise, from a fast graphics card (especially if their CPU is slow)
    0
  • Alain Decamps
    Hi Christian

    First thanks for al the very useful info you give here.

    I used you're info below to do some quick and simple extra/interpolations with the benchmark nr's.

    If
    2896/12119 is about 2x faster (gtx680/3930K)
    then
    2262/6259 is about 3x faster (amd 7870/i5-2500K)

    If the gtx680 gives a twofold improvement on the 3630k I get more than a 3fold improvement by adding a AMD 7870 to my i5-2500K. This is a very simple extrapolation, but is it a valid extrapolation or is it far off?

    [quote="Christian Gr" wrote:

    This is a good source to see what hardware will generate faster performance:
    GPU: http://www.videocardbenchmark.net/directCompute.html
    CPU: http://www.cpubenchmark.net/high_end_cpus.html
    HDD: http://www.harddrivebenchmark.net/ssd.html


    [quote="Christian Gr" wrote:

    I have access to computer with a 3930k cpu and a gtx680. Processing cpu vs gpu, the GPU is around 2 times faster (though CPU is still very fast on its own).
    0
  • Christian Gruner
    [quote="Alain" wrote:
    Hi Christian

    First thanks for al the very useful info you give here.

    I used you're info below to do some quick and simple extra/interpolations with the benchmark nr's.

    If
    2896/12119 is about 2x faster (gtx680/3930K)
    then
    2262/6259 is about 3x faster (amd 7870/i5-2500K)

    If the gtx680 gives a twofold improvement on the 3630k I get more than a 3fold improvement by adding a AMD 7870 to my i5-2500K. This is a very simple extrapolation, but is it a valid extrapolation or is it far off?

    [quote="Christian Gr" wrote:

    This is a good source to see what hardware will generate faster performance:
    GPU: http://www.videocardbenchmark.net/directCompute.html
    CPU: http://www.cpubenchmark.net/high_end_cpus.html
    HDD: http://www.harddrivebenchmark.net/ssd.html


    [quote="Christian Gr" wrote:

    I have access to computer with a 3930k cpu and a gtx680. Processing cpu vs gpu, the GPU is around 2 times faster (though CPU is still very fast on its own).



    You could make that extrapolation, but I dunno if their numbers are linear and so on. But try it out, and remember to tell us your findings.
    0
  • Alain Decamps
    [quote="Christian Gr" wrote:
    [quote="Alain" wrote:
    Hi Christian

    First thanks for al the very useful info you give here.

    I used you're info below to do some quick and simple extra/interpolations with the benchmark nr's.

    If
    2896/12119 is about 2x faster (gtx680/3930K)
    then
    2262/6259 is about 3x faster (amd 7870/i5-2500K)

    If the gtx680 gives a twofold improvement on the 3630k I get more than a 3fold improvement by adding a AMD 7870 to my i5-2500K. This is a very simple extrapolation, but is it a valid extrapolation or is it far off?

    [quote="Christian Gr" wrote:

    This is a good source to see what hardware will generate faster performance:
    GPU: http://www.videocardbenchmark.net/directCompute.html
    CPU: http://www.cpubenchmark.net/high_end_cpus.html
    HDD: http://www.harddrivebenchmark.net/ssd.html


    [quote="Christian Gr" wrote:

    I have access to computer with a 3930k cpu and a gtx680. Processing cpu vs gpu, the GPU is around 2 times faster (though CPU is still very fast on its own).



    You could make that extrapolation, but I dunno if their numbers are linear and so on. But try it out, and remember to tell us your findings.


    Thanks, I hope that I can test it this weekend.
    0
  • Cuong Tran
    Yes TTX 660TI makes a big difference in WB adjustment.
    0
  • Alain Decamps
    Hi

    Installed my brand new 7870, but didn't see a difference if opencl display is on auto or never. Is there a way to see if the opencl is used at all?

    The CPU utilization stayed high and I don't see a difference there to (but this is more difficult to see.)

    I didn't found the "processing" to be faster also.

    One of the tests : new RAW image, zoom to 100%, take the "hand"tool and do a fast move to the side. The screen redraw stays the same (a small two seconds on(2560*1440)).

    Alain
    0
  • Christian Gruner
    [quote="Alain" wrote:
    Hi

    Installed my brand new 7870, but didn't see a difference if opencl display is on auto or never. Is there a way to see if the opencl is used at all?

    The CPU utilization stayed high and I don't see a difference there to (but this is more difficult to see.)

    I didn't found the "processing" to be faster also.

    One of the tests : new RAW image, zoom to 100%, take the "hand"tool and do a fast move to the side. The screen redraw stays the same (a small two seconds on(2560*1440)).

    Alain


    Sounds like you have setting used on the picture that disables OpenCL. A quick way to check is to zoom into 100%, change the contrast.

    - if the picture pixelates slightly, OpenCL is off
    - if the picture remains "sharp", OpenCL is on.

    Try resetting your picture before doing this (Ctrl + R)
    0
  • Raymond4
    I returned a gtx 660ti (gigabyte card - noisy)- but not obviously better than gtx 560ti. Bought an asus 7870 top oc which although was a tad faster on process, had horrible pixelation on 100% hand drag of image. Only tiny twitch with 560ti/660ti. I previously had image corruption of one form or other with radeon drivers, but wanted to try the 7870. Have returned it. I will probably rebuy a 660ti to get 2gb memory or 760ti? when released. I will stick to nvidia. for infomation if anybody fancies trying a 7870 make sure your pc case will take it as they are big!
    0
  • Alain Decamps
    [quote="Christian Gr" wrote:
    [quote="Alain" wrote:
    Hi

    Installed my brand new 7870, but didn't see a difference if opencl display is on auto or never. Is there a way to see if the opencl is used at all?

    The CPU utilization stayed high and I don't see a difference there to (but this is more difficult to see.)

    I didn't found the "processing" to be faster also.

    One of the tests : new RAW image, zoom to 100%, take the "hand"tool and do a fast move to the side. The screen redraw stays the same (a small two seconds on(2560*1440)).

    Alain


    Sounds like you have setting used on the picture that disables OpenCL. A quick way to check is to zoom into 100%, change the contrast.

    - if the picture pixelates slightly, OpenCL is off
    - if the picture remains "sharp", OpenCL is on.

    Try resetting your picture before doing this (Ctrl + R)


    With the contrast I do see a difference between "auto" and never. OpenCL auto is visible faster with a more "smoother" effect.

    When using the "hand" too; and zooming in (to 100%) I don't seem to see a difference.

    Alain
    0
  • Cuong Tran
    For me, the first change of WB is faster but noticeable lag. Subsequently WB change are much more responsive but not instantaneous. Zoom to 100% and move to the furthest side is not instantaneous either - probably 0.3 sec lag. Photo size is 5212x3468. The video card helps and I can live with this level of responsiveness but would love instantaneous response. Upgrading to desk top of 3930K is not a cheap proposition though.

    -Cuong
    0
  • Alain Decamps
    Hi

    I have the impression that my 7870 is only used a little bit when doing display related tasks. When I monitor GPU usage when having an image 100% (on a 2560*1400 display) and moving it with the hand tool from left to right and back I have about 1.5 - 2.0 second delay when the image is completely ok (no big upscaled pixels). The GPU shows (with CCC) that it's only used at abound 28-30% most of the time. I didn't found a display related task that upped the usage significant. The GPU temperature also kept rather low (only a little bit temp increase.

    Unfortunately then the GPU does add a noticeable speed to a CPU only situation.
    Edit : I forgot the not : should be "does not add"

    Is this something other users also see or is there something wrong with my setup (win 7, i5 2500K, 13.1 catalyst, AMD 7870 1Ghz)?

    While processing to jpeg I do see a maximum usage of about 65%.

    BTW. A benchmark luxmark uses the same GPU up to 93% and then the temperatures do rise more.
    0
  • Christian Gruner
    [quote="Alain" wrote:
    Hi

    I have the impression that my 7870 is only used a little bit when doing display related tasks. When I monitor GPU usage when having an image 100% (on a 2560*1400 display) and moving it with the hand tool from left to right and back I have about 1.5 - 2.0 second delay when the image is completely ok (no big upscaled pixels). The GPU shows (with CCC) that it's only used at abound 28-30% most of the time. I didn't found a display related task that upped the usage significant. The GPU temperature also kept rather low (only a little bit temp increase.

    Unfortunately then the GPU does add a noticeable speed to a CPU only situation.

    Is this something other users also see or is there something wrong with my setup (win 7, i5 2500K, 13.1 catalyst, AMD 7870 1Ghz)?

    While processing to jpeg I do see a maximum usage of about 65%.

    BTW. A benchmark luxmark uses the same GPU up to 93% and then the temperatures do rise more.


    We are looking into optimizing the behavior on this, but it should still be noticeably faster than CPU.

    I'm not quite sure understand this part "Unfortunately then the GPU does add a noticeable speed to a CPU only situation" ? Can you elaborate?
    0
  • Alain Decamps
    [quote="Christian Gr" wrote:
    [quote="Alain" wrote:
    Hi

    I have the impression that my 7870 is only used a little bit when doing display related tasks. When I monitor GPU usage when having an image 100% (on a 2560*1400 display) and moving it with the hand tool from left to right and back I have about 1.5 - 2.0 second delay when the image is completely ok (no big upscaled pixels). The GPU shows (with CCC) that it's only used at abound 28-30% most of the time. I didn't found a display related task that upped the usage significant. The GPU temperature also kept rather low (only a little bit temp increase.

    Unfortunately then the GPU does add a noticeable speed to a CPU only situation.

    Is this something other users also see or is there something wrong with my setup (win 7, i5 2500K, 13.1 catalyst, AMD 7870 1Ghz)?

    While processing to jpeg I do see a maximum usage of about 65%.

    BTW. A benchmark luxmark uses the same GPU up to 93% and then the temperatures do rise more.


    We are looking into optimizing the behavior on this, but it should still be noticeably faster than CPU.

    I'm not quite sure understand this part "Unfortunately then the GPU does add a noticeable speed to a CPU only situation" ? Can you elaborate?


    Oeps, I forgot not. And the delay is 1,5 - 2 seconds after I move from left to right and another 1,5 - 2 seconds after I move back from right to left. (I would expect it to be cached, but that's another story.)

    To be clear: I don't feel a speed difference if I use the GPU (display auto) of not (display never). The speed is in both case about the same.
    0
  • Alain Decamps
    Is there something I can do to help?
    Is it usefull to open a support case?
    0
  • Cuong Tran
    Alain. I feel your pain. After I installed GPU perf monitoring, I realized that my video card is barely used by C1. What a freaking waste of money to go out and buy a fast GPU. Very disappointed.

    GPU used by when first clicked on a raw photo: 5-10%.
    GPU used by 100% viewing and then scrolling: 15-20% when I first scrolled. Then CPU is busy.
    GPU used by WB: 5-7%
    GPU used by processing: 3-6%

    Is this normal behavior? What OpenGL version that I should have it installed?

    --Cuong
    0
  • Cuong Tran
    Upgraded my driver to version 3.14.07
    Processing is now much snappier. I saw GPU is 60-80% busy for 1-2 sec (this is the most visible improvement with this driver version). The test that zoom in 100% and scroll from one side to the opposite and then back is snappier too.
    0
  • Christian Gruner
    [quote="Cuong1" wrote:
    Alain. I feel your pain. After I installed GPU perf monitoring, I realized that my video card is barely used by C1. What a freaking waste of money to go out and buy a fast GPU. Very disappointed.

    GPU used by when first clicked on a raw photo: 5-10%.
    GPU used by 100% viewing and then scrolling: 15-20% when I first scrolled. Then CPU is busy.
    GPU used by WB: 5-7%
    GPU used by processing: 3-6%

    Is this normal behavior? What OpenGL version that I should have it installed?

    --Cuong


    Processing is where the real benefit of the GPU comes in. We are always trying to improve on performance, and OpenCL is certainly not an exception.
    0
  • Cuong Tran
    Yes, I am happy camper now. Processing time is 9 sec (from 20 sec). WB change is instantaneous. It's a good move to upgrade to faster video card than to another desktop with 3930K. Brush for local adjustment is still slow and probably this is where 3930k helps but I don't use this feature frequently enough to justify the upgrade to desktop with 3930K. Thx.
    0
  • Alain Decamps
    [quote="Cuong1" wrote:
    Yes, I am happy camper now. Processing time is 9 sec (from 20 sec). WB change is instantaneous. It's a good move to upgrade to faster video card than to another desktop with 3930K. Brush for local adjustment is still slow and probably this is where 3930k helps but I don't use this feature frequently enough to justify the upgrade to desktop with 3930K. Thx.


    Good for you and the're is hope 😉

    At the moment I have the latest non beta driver for my 7870 card, so no direct possibility to upgrade, grrr
    0
  • Alain Decamps
    The newest capture one release 7.1.1 didn't change the behaviour.

    Alain
    0
  • SFA
    Well I have finally got a 64bit system and have been experimenting for a few days now.

    Some general and very subjective observations.

    It's a notebook with a low end Quadro K1000M GPU from which I don't expect much and so far as I have seen on a monitoring gadget it never sees more than about 20% utilisation max.

    i7 3820QM processor running at 2.7Ghz.
    8Gb RAM
    512Gb SSD

    I have so far left the Preview file resolution at 2560 but my screen is max 1920 so a little delay is perceptible when dealing with larger files and re-sizing during edit - but maybe only a second or two at the very longest.

    Basically, using sessions, for almost everything almost all of the time it's a fast as I can use when processing. Someone younger and more skilled may find what little delay there seems to be more distracting. This is for Canon 1D3 files of around 14Mb or 600D files which are larger. If I avoid running a busy web browser (IE9 or Firefox) memory usage seems to run at about 4 to 5 Gb usually. If I run 3 or 4 C1 windows with different sessions it creeps closer to 8Gb total usage but does not seem to greatly affect performance.

    I have not tried catalogues.

    I have, briefly, tried tethering with both 600D and 1D3 and Captured seemed able to take whatever abuse I tried with no obvious problems - but I'm no expert in that area of functionality and I was using a very short cable.

    My Wacom Pre-Bamboo tablet was not entirely happy when first loaded but a Wacom driver installation (the auto install used the Win 7 derived driver) and a re-boot greatly improved things. However I find I am happy using the trackpad for most activity.

    I'm sure the impression of remarkable speed will wear off in time - these things always do - and it I was using a body or back that delivered bigger files I would expect it to be slower but right now I'm thinking I may not be able to justify the extra cost of a desktop machine for whatever advantage it may bring.

    I'm not even sure that my original plan to add more memory is worth following up since it does not seem to be at all memory constrained. (That would probably not be the case if I was running Photoshop or similar at the same time.

    7.1 and 7.1.1 seem to be no different (in subjective terms) in performance on this machine.

    As I said at the start - very subjective and nothing yet to provide comparative measurements but realistically I don't think I could measure processing times, for most things, with any degree of accurate consistency. Batches of imports and processing for output should be measurable of course but how relevant they would be would need to be assessed and a careful test design defined for them to have any comparative meaning. In my opinion.

    I'm happy to field any questions or run comparative tests on a 'standard' file or two if that would interest anyone.


    Grant Perkins
    0
  • mstauss
    Here in our studio I've found it is helpful to turn the "Hardware Acceleration (Use Open CL for)" options off (Never) in the preferences to speed up the overall performance speed of C1 Pro 7.1. We have also upgraded to 7.1.1 patch which seems to fix a lot of the bug issues. Hope this is helpful.
    Cheers
    0
  • Christian Gruner
    [quote="mstauss" wrote:
    Here in our studio I've found it is helpful to turn the "Hardware Acceleration (Use Open CL for)" options off (Never) in the preferences to speed up the overall performance speed of C1 Pro 7.1. We have also upgraded to 7.1.1 patch which seems to fix a lot of the bug issues. Hope this is helpful.
    Cheers


    A slow gfx card compared to the CPU can give this behavior. This is not normal behavior for CO at least.

    My recommendation would be to upgrade your gfx card and you will see a major speed improvement on image-adjustment/processing tasks.
    0
  • Alain Decamps
    [quote="Alain" wrote:
    The newest capture one release 7.1.1 didn't change the behaviour.

    Alain


    I did try the latest beta driver from AMD, but it doesn't change something (visible) the GPU is only 25-28% used.

    OpenGL-version: 6.14.10.12173

    Quite a disappointment.

    Alain
    0
  • Permanently deleted user
    The performance problem still exists in v7.1.1. Turing Open-CL on or off makes not difference as far as screen refresh is concerned. The delay of 1 to 2 seconds when clicking on a new photo or making a exposure change is still there. It appears that CO7 is recreating the preview image each time.

    My computer has two 30" LCD monitors running at 2560x1600. A 1080p TV is connected as the third monitor but is normally off during photo editing. The primary screen is used to run CO7 and the second one is used for image verification after it is processed. The preview size was set to the max, 2880, I believe. Changing it to 2560 made no difference.

    When CO6 came out, it has stability and performance issues. I opened support cases and the problem was eventually fixed. Now we are at this again.

    Is anyone at PhaseOne aware of this problem and working on it?
    0
  • Christian Gruner
    [quote="Nelson1" wrote:
    The performance problem still exists in v7.1.1. Turing Open-CL on or off makes not difference as far as screen refresh is concerned. The delay of 1 to 2 seconds when clicking on a new photo or making a exposure change is still there. It appears that CO7 is recreating the preview image each time.

    My computer has two 30" LCD monitors running at 2560x1600. A 1080p TV is connected as the third monitor but is normally off during photo editing. The primary screen is used to run CO7 and the second one is used for image verification after it is processed. The preview size was set to the max, 2880, I believe. Changing it to 2560 made no difference.

    When CO6 came out, it has stability and performance issues. I opened support cases and the problem was eventually fixed. Now we are at this again.

    Is anyone at PhaseOne aware of this problem and working on it?

    What graphics card do you have ?
    0
  • Permanently deleted user
    [quote="Christian Gr" wrote:

    What graphics card do you have ?


    It's an AMD Radeon HD6870 with 1GB memory.

    The following is the information I posted when I started this thread 5 months ago. Time flies!

    System information:

    Intel i7-870
    16GB RAM
    OS drive 240GB SSD
    CO7 installed on SSD
    Image working directory on SSD
    AMD Radeon HD6870 with 1GB memory
    OpenCL enabled
    Windows 7 Ultimate x64

    The same system performs much better with CO6.
    0

Post is closed for comments.