Hi,
did some tests:
Intel® Core™ i5-2500 CPU @ 3.30GHz × 4
GeForce GTX 750 Ti, 2GB,128-bit
Ubuntu 15.10 64-Bit, 16GB
opencl - m1
[dev_process_export] pixel pipeline processing took 1,075 secs (2,872 CPU)
opencl - m3
[dev_process_export] pixel pipeline processing took 2,001 secs (6,548 CPU)
no-opencl - m1
[dev_process_export] pixel pipeline processing took 1,289 secs (3,892 CPU)
no-opencl - m3
[dev_process_export] pixel pipeline processing took 2,289 secs (7,528 CPU)
Details attached
With opencl activated darkroom mode is much more snappy.
If you need more information, let me know.
thorsten
Am 19.01.2016, 21:42 Uhr, schrieb Ulrich Pegelow
<ulrich.pege...@tongareva.de>:
Hi,
I just merged my OpenCL implementation of the Markesteijn demosaicing
algorithm into master. Markesteijn with one or three passes ("-1" and
"-3", respectively) is darktable's preferred method for demosaicing
images of cameras with Fuji's X-Trans sensor.
The algorithm is rather complex and has a significant memory overhead.
Therefore the performance advantage over the already well established
CPU codepath will depend a lot on the GPU hardware. In my case (AMD
Radeon HD7950) I get roughly a 2 times faster processing compared to my
i7-2600@3.4GHz.
But that's just one system. Slower GPUs might even underperform versus
the CPU. I would like to gather more benchmarking data in order to
decide if the OpenCL codepath for Markesteijn should be enabled by
default or not.
So if you are running master and if you have a working OpenCL system I
am highly interested in your comparison of OpenCL versus CPU speed.
Please start darktable with '-d opencl -d perf' and do a few test:
1) with OpenCL enabled in preferences
2) with OpenCL disabled in preferences
Please perform:
a) exporting an X-Trans image with default history stack
b) as above but now with demosaic set to Markesteijn-3
Don't forget to tell me your hardware setup (Graphics Card, GPU memory,
CPU type, system memory).
I am also interested how reactive darktable acts in darkroom mode if you
pan the image or change some module parameters (take exposure correction
as an example). Please zoom into the image with a zoom level of 67% or a
bit more. This test is more about "look and feel" rather than an exact
measurement. Does darktable react more "snappy" with OpenCL or just the
opposite?
If you don't have an X-Trans sensor raw at hand you can download one
from here:
https://www.dropbox.com/s/gef9qkapkktrjn5/DSCF6768.RAF?dl=0
Thanks for your support!
Ulrich
___________________________________________________________________________
darktable developer mailing list
to unsubscribe send a mail to
darktable-dev+unsubscr...@lists.darktable.org
___________________________________________________________________________
darktable developer mailing list
to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
Intel® Core⢠i5-2500 CPU @ 3.30GHz à 4
GeForce GTX 750 Ti, 2GB,128-bit
Ubuntu 15.10 64-Bit, 16GB
[opencl_init] OpenCL successfully initialized.
[opencl_init] here are the internal numbers and names of OpenCL devices
available to darktable:
[opencl_init] 0 'GeForce GTX 750 Ti'
opencl - m1
[export] creating pixelpipe took 0,129 secs (0,200 CPU)
[pixelpipe_process] [export] using device 0
[dev_pixelpipe] took 0,000 secs (0,000 CPU) initing base buffer [export]
[dev_pixelpipe] took 0,010 secs (0,016 CPU) processed `Raw-Schwarz-/WeiÃpunkt'
on GPU, blended on GPU [export]
[dev_pixelpipe] took 0,102 secs (0,060 CPU) processed `WeiÃabgleich' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,008 secs (0,020 CPU) processed
`Spitzlicht-Rekonstruktion' on CPU, blended on CPU [export]
[dev_pixelpipe] took 0,736 secs (2,524 CPU) processed `Entrastern' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,069 secs (0,052 CPU) processed `Basiskurve' on GPU,
blended on GPU [export]
[dev_pixelpipe] took 0,011 secs (0,008 CPU) processed `Eingabefarbprofil' on
GPU, blended on GPU [export]
[dev_pixelpipe] took 0,043 secs (0,032 CPU) processed `Schärfen' on GPU,
blended on GPU [export]
[dev_pixelpipe] took 0,022 secs (0,016 CPU) processed `Ausgabefarbprofil' on
GPU, blended on GPU [export]
[dev_pixelpipe] took 0,073 secs (0,144 CPU) processed `Gamma' on CPU, blended
on CPU [export]
[opencl_profiling] spent 0,0613 seconds in [Write Image (from host to device)]
[opencl_profiling] spent 0,0025 seconds in rawprepare_1f
[opencl_profiling] spent 0,1012 seconds in [Read Image (from device to host)]
[opencl_profiling] spent 0,0137 seconds in basecurve
[opencl_profiling] spent 0,0098 seconds in colorin_unbound
[opencl_profiling] spent 0,0126 seconds in sharpen_hblur
[opencl_profiling] spent 0,0141 seconds in sharpen_vblur
[opencl_profiling] spent 0,0126 seconds in sharpen_mix
[opencl_profiling] spent 0,0210 seconds in colorout
[opencl_profiling] spent 0,2489 seconds totally in command queue (with 0
events missing)
[dev_process_export] pixel pipeline processing took 1,075 secs (2,872 CPU)
opencl - m3
[export] creating pixelpipe took 0,118 secs (0,172 CPU)
[pixelpipe_process] [export] using device 0
[dev_pixelpipe] took 0,000 secs (-0,000 CPU) initing base buffer [export]
[dev_pixelpipe] took 0,017 secs (0,020 CPU) processed `Raw-Schwarz-/WeiÃpunkt'
on GPU, blended on GPU [export]
[dev_pixelpipe] took 0,045 secs (0,084 CPU) processed `WeiÃabgleich' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,008 secs (0,024 CPU) processed
`Spitzlicht-Rekonstruktion' on CPU, blended on CPU [export]
[dev_pixelpipe] took 1,705 secs (6,140 CPU) processed `Entrastern' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,067 secs (0,084 CPU) processed `Basiskurve' on GPU,
blended on GPU [export]
[dev_pixelpipe] took 0,013 secs (0,008 CPU) processed `Eingabefarbprofil' on
GPU, blended on GPU [export]
[dev_pixelpipe] took 0,040 secs (0,028 CPU) processed `Schärfen' on GPU,
blended on GPU [export]
[dev_pixelpipe] took 0,020 secs (0,020 CPU) processed `Ausgabefarbprofil' on
GPU, blended on GPU [export]
[dev_pixelpipe] took 0,085 secs (0,140 CPU) processed `Gamma' on CPU, blended
on CPU [export]
[opencl_profiling] spent 0,0633 seconds in [Write Image (from host to device)]
[opencl_profiling] spent 0,0032 seconds in rawprepare_1f
[opencl_profiling] spent 0,0640 seconds in [Read Image (from device to host)]
[opencl_profiling] spent 0,0145 seconds in basecurve
[opencl_profiling] spent 0,0112 seconds in colorin_unbound
[opencl_profiling] spent 0,0118 seconds in sharpen_hblur
[opencl_profiling] spent 0,0127 seconds in sharpen_vblur
[opencl_profiling] spent 0,0132 seconds in sharpen_mix
[opencl_profiling] spent 0,0184 seconds in colorout
[opencl_profiling] spent 0,2123 seconds totally in command queue (with 0
events missing)
[dev_process_export] pixel pipeline processing took 2,001 secs (6,548 CPU)
[export_job] exported to `/home/fritz/Bilder/Test DSLR/DSCF6768opencl - m3.jpg'
no-opencl - m1
[export] creating pixelpipe took 0,136 secs (0,216 CPU)
[pixelpipe_process] [export] using device -1
[dev_pixelpipe] took 0,000 secs (0,000 CPU) initing base buffer [export]
[dev_pixelpipe] took 0,010 secs (0,024 CPU) processed `Raw-Schwarz-/WeiÃpunkt'
on CPU, blended on CPU [export]
[dev_pixelpipe] took 0,042 secs (0,088 CPU) processed `WeiÃabgleich' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,008 secs (0,020 CPU) processed
`Spitzlicht-Rekonstruktion' on CPU, blended on CPU [export]
[dev_pixelpipe] took 0,763 secs (2,536 CPU) processed `Entrastern' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,105 secs (0,120 CPU) processed `Basiskurve' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,058 secs (0,188 CPU) processed `Eingabefarbprofil' on
CPU, blended on CPU [export]
[dev_pixelpipe] took 0,160 secs (0,448 CPU) processed `Schärfen' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,110 secs (0,372 CPU) processed `Ausgabefarbprofil' on
CPU, blended on CPU [export]
[dev_pixelpipe] took 0,034 secs (0,096 CPU) processed `Gamma' on CPU, blended
on CPU [export]
[dev_process_export] pixel pipeline processing took 1,289 secs (3,892 CPU)
[export_job] exported to `/home/fritz/Bilder/Test DSLR/DSCF6768no-opencl -
m1.jpg'
no-opencl - m3
[export] creating pixelpipe took 0,109 secs (0,152 CPU)
[pixelpipe_process] [export] using device -1
[dev_pixelpipe] took 0,000 secs (-0,000 CPU) initing base buffer [export]
[dev_pixelpipe] took 0,010 secs (0,020 CPU) processed `Raw-Schwarz-/WeiÃpunkt'
on CPU, blended on CPU [export]
[dev_pixelpipe] took 0,058 secs (0,060 CPU) processed `WeiÃabgleich' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,008 secs (0,024 CPU) processed
`Spitzlicht-Rekonstruktion' on CPU, blended on CPU [export]
[dev_pixelpipe] took 1,714 secs (6,180 CPU) processed `Entrastern' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,115 secs (0,144 CPU) processed `Basiskurve' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,061 secs (0,188 CPU) processed `Eingabefarbprofil' on
CPU, blended on CPU [export]
[dev_pixelpipe] took 0,184 secs (0,444 CPU) processed `Schärfen' on CPU,
blended on CPU [export]
[dev_pixelpipe] took 0,109 secs (0,372 CPU) processed `Ausgabefarbprofil' on
CPU, blended on CPU [export]
[dev_pixelpipe] took 0,030 secs (0,096 CPU) processed `Gamma' on CPU, blended
on CPU [export]
[dev_process_export] pixel pipeline processing took 2,289 secs (7,528 CPU)
[export_job] exported to `/home/fritz/Bilder/Test DSLR/DSCF6768no-opencl -
m3.jpg'
