Mac Mini 2012 : should I switch on OpenCL or not?
Hi!
My machine is a Mac Mini 2012 with i7 quad-core processor running at 2,3 GHz and Intel HD Graphics 4000, OS X 10.11.3.
I'm using CO9 Express for Sony: do you think I should tick the boxes for OpenCL acceleration in Preferences? Will CO9 Express make use of my (admittedly weak) GPU to speed up anything or not? If not, is it OK to leave these options to "Auto" or should I choose "Never"?
Thx!
My machine is a Mac Mini 2012 with i7 quad-core processor running at 2,3 GHz and Intel HD Graphics 4000, OS X 10.11.3.
I'm using CO9 Express for Sony: do you think I should tick the boxes for OpenCL acceleration in Preferences? Will CO9 Express make use of my (admittedly weak) GPU to speed up anything or not? If not, is it OK to leave these options to "Auto" or should I choose "Never"?
Thx!
0
-
[quote="John Doe" wrote:
Hi!
My machine is a Mac Mini 2012 with i7 quad-core processor running at 2,3 GHz and Intel HD Graphics 4000, OS X 10.11.3.
I'm using CO9 Express for Sony: do you think I should tick the boxes for OpenCL acceleration in Preferences? Will CO9 Express make use of my (admittedly weak) GPU to speed up anything or not? If not, is it OK to leave these options to "Auto" or should I choose "Never"?
Thx!
I have a 2015 Mini (dual core i7) and OpenCl sometimes causes my raw images to go black when I am adjusting them and sometimes causes my output images to become black when processing them to disk and I had to turn OpenCL off. On the other hand I have a 2013 MacBook Pro with a quad core i7 and NVidea graphics card and found OpenCL to be faster.
You are somewhere in between those two since you have a quad core i7, but no separate graphics card. My suggestion would be for you to turn it off and see how it works, turn it on and see how that works and then decide based on how long it takes to adjust your images, how long it takes to process them and how they look on the screen.
I would be interested in knowing your results.0 -
I have the same system and when I turn OpenCL on it is much slower and laggier to draw masks than with it off. I set OpenCL to Never for Display and everything runs pretty smoothly. 0 -
@MikeFromMesa: I haven't noticed any clear-cut difference yet, hence my question.
@meanwhile: I don't use masks. Did you notice laggy performance with other tools too? I understand you set OpenCL to "never" for display and to "auto" for the other setting?
Thanks to both of you for your replies. 😊0 -
[quote="John Doe" wrote:
@MikeFromMesa: I haven't noticed any clear-cut difference yet, hence my question.
)
The purpose of OpenCL, as I understand it, is the provide parallel functionality to the software to allow faster and smoother adjustment and processing of the images. That is, instead of the main software having to do all of the work, some of the processing and adjustment work could be handed over to the graphics card processor to do while the main software is doing something else. Hence a faster image adjustment and/or faster processing to disk.
If you are not noticing any difference, then perhaps you are not getting any benefit and so it makes no difference. However there are two tests I would perform to verify this, provided your workflow would accommodate them.
1) In my workflow I adjust all of the images I want to keep and only then, after all adjustments are complete, do I select them and process them. If your workflow is similar, then there is a simple test. Open a catalog/session you have already adjusted (presumably with at least 30 or 40 raw images to process), turn OpenCL off for processing and time how long processing to disk takes. Then turn OpenCL back on and do the same thing, again timing how long it takes and then compare the two times.
2) If you have any of the tools that are available to display how many of your cores are being used when processing to disk, you can repeat the above two tests seeing how many cores are in use. If one test shows fewer cores in use (and hence more available for other things) you might want to select that OpenCL setting. Alternately you can use the Activity Monitor to see if you are using more or fewer memory/CPU cycles and then choose that setting.
My guess is that if you don't notice any difference, there probably is not any noticeable difference and the setting does not matter. In that case I would turn it off as there is no benefit from having the software go through the process of trying to hand the work off to something else that is not doing what it should do.
Just my opinion.0 -
Good idea. Here's what I did. I selected a collection of 128 pictures and processed them to JPEG.
without OpenCL (neither Display nor Processing) : 2 minutes 54 seconds
with OpenCL (both Display and Processing) : 2 minutes 30 seconds
CPUs run at ~80% in both cases but as the times show, CO performs faster with OpenCL activated.
As regards drawing masks (cf. "meanwhile"'s post), I activated the trial Pro version (local adjustments are not available in the Express version) and actually found the opposite : it /seems/ a bit smoother with OpenCL activated.
So I guess I'll leave OpenCL activated on my machine, as it seems to offer a slight performance boost.
Thanks!0 -
[quote="John Doe" wrote:
As regards drawing masks ... I ... actually found the opposite : it /seems/ a bit smoother with OpenCL activated.
Does "smoother" mean "not as sharp", and is it something positive which you mean? (if you mean performance then it is not the opposite)
Given the performance difference is not that big I would go with the setting with the better image quality, whichever this is...0 -
What I meant is that drawing a mask with OpenCL activated seems less "jerky" than without OpenCL on my system, contrary to what "meanwhile" experienced (he/she said it was slower and laggier with OpenCL activated on his/her Mac).
Sorry if it wasn't clear, English is not my native language.0 -
[quote="John Doe" wrote:
CO performs faster with OpenCL activated.
...
So I guess I'll leave OpenCL activated on my machine, as it seems to offer a slight performance boost.
Given the figures you got that is what I would do. Glad you got it figured out.0 -
OpenCL works fine on my 2012 quad core i7 Mini. Thumbnail scrolling, adjustment slider dragging, and image panning are all satisfyingly smooth. No image blackouts. I’ve never noticed any differences with image exporting. 0 -
[quote="markalanthomas" wrote:
OpenCL works fine on my 2012 quad core i7 Mini. Thumbnail scrolling, adjustment slider dragging, and image panning are all satisfyingly smooth. No image blackouts. I’ve never noticed any differences with image exporting.
I guess that is the difference between the quad-core i7 chips and the dual-core i7 chips. I have the latter and I have both image blackouts when adjusting and black images when exporting if OpenCL is turned on. A two-fer. Lucky me.0 -
Or maybe Intel HD Graphics 4000 vs. 5000 ? 0 -
OpenCL should be GPU dependent, no? 0 -
Yes, although designed for heterogenous computing devices, OPENCL is only used for the GPU(s) in C1. [quote="Christian Gruner" wrote:
Regarding different rendering on CPU vs GPU, then yes they are different, as 2 different programming languages are used. However, we run extensive checks to ensure they remain identical in output.
source:
So, "hardware acceleration" in the preferences could be renamed to "GPU acceleration".0 -
Well I can speak for myself, OpenCL doesn't work very well on my late 2015 iMac (4Ghz i7, 24GB Ram, AMD M395x 4GB GPU)
If I set OpenCL as Auto, drawing a mask is laggy, slow and not consistent, but then my CPU usage on Activity Monitor is very low. If I do the same while OpenCL is set to never, then drawing mask is very smooth but then CPU Usage jumps to 400% or more.0 -
For reference, the HD 4000 in my Mini 2012 scores a paltry 20.587599 according to the CaptureOneICP.log benchmark: 2016-02-03 17:14:49> OpenCL benchMark : 20.587599
0
Post ist für Kommentare geschlossen.
Kommentare
15 Kommentare