Am 02.12.18 um 16:42 schrieb Michael Rasmussen:
On Sun, 2 Dec 2018 15:52:02 +0100
Volker Lenhardt <[email protected]> wrote:
See also 'opencl_avoid_atomics' here:
https://www.darktable.org/usermanual/en/darktable_and_opencl_amd.html
Did you see above?
Sorry for not having told you. Changing made no difference.
Am 02.12.18 um 15:59 schrieb Ulrich Pegelow:
>
> For this case I would like to see the full output of 'darktable -d
> opencl -d perf' with activated OpenCL from program start to where you
> process the image with profiled denoise.
>
I've found a wrong option setting: I had the central option (in German)
"OpenCL-Scheduler-Profil" set to "mehrere GPUs". I don't know why. Now
it is set to "Standard". The processing time decreased to less than a
fourth.
Profiled denoise:
144,177185 [dev_process_image] pixel pipeline processing took 4,058 secs
(1,591 CPU)
Is this to be tolerated? Do you still need the long output? Then please
wait until tomorrow, I'm on the run at the moment.
There's still the problem of too slow response with shifting a zoomed in
image to have e.g. a quick look at the corners to detect chromatic
aberration. And cropping an image is slow, too. All of this rating is
compared to what I was used to under Linux with less confined equipment.
Volker
____________________________________________________________________________
darktable user mailing list
to unsubscribe send a mail to [email protected]
