New CPU - 4 or 6 cores?

[Alex_S]

Which processor will be faster for loading (RAW decode) and editing RAW photos?

Intel Core i7-7700K (4 cores / 8 threads) 4.2 GHZ
Intel Core i7-6800K (6 cores / 12 threads) 3.4 GHZ
AMD Ryzen 7 1700X (8 cores / 16 threads) 3.4 GHZ

Now the speed of loading (decode) RAW photos from Nilon D750 (24MP) on my CPU intel i7 3770 (3.4 GHZ. cores / 8 threads, GPU - geforce 210) - 2.0 - 2.8 sec.

Does anyone have a processor with 6 or 8 cores? How long does the RAW load?

[Glen Barrington]

I suspect the differences the number of cores for processing raw image files doesn't matter THAT much. I personally suspect that the real bottleneck for still photography is still in the type of graphics card you have and the speed of your HDD or SSD. Display speed is pretty much an artifact of the graphics card for still photography.

Now if you intend to get into video, I suspect rendering time (Where the program takes the edited project and turns it into a playable video) would be greatly enhanced by a more powerful processor but even so, a 10 minute video can be rendered in about 10 minutes with a lenovo Desktop with an i7-6700 CPU running at 3.40 Ghz and 12 GB of ram.

My advice is don't skimp on the graphics card or your permanent storage read/write speed. If you have to compromise on something, I would go with the 4 core processors.

[LV_Bill]

Adding to Glen's good advice. In my last PC upgrade, I found that the CPU was important, and the Graphics card was very important. But, changing from HDD to SSD drives was overwhelmingly the most important factor affecting overall performance. Nothing comes close to the performance boost you will get from going to all SSD drives.

[GusPanella]

+1 for SSD Drives... +2 for SSD NVMe u.2 drives.

[Nxdibbles]

My Asus 1070 8gb upgrade made huge improvements on developing and editing running a lot smoother than my old 7870

[Alex_S]

Hi All.
I have a PC system:
Intel core i5-3300 (3 GHz), 4 core / 4 threads
GPU intel hd graphics 2500
SSD Samsung 850 pro
Windows 7 sp1 x64
ACDSee Pro 10.4 x64
Camera RAW photos from Nikon D750 (12 bit).


Please do the test on your PC.

Download my RAW files for the test (106 MB, zip)
https://1drv.ms/u/s!Ap5FENhaDVfcqUOA5I1KUID41YII
or
https://yadi.sk/d/Ion-0seM3JpD2y

First you need to configure the ACDSee Pro:
1) Options -> General -> set "RAW decode"
2) Options -> General -> check or uncheck "Enable OpenCL image processing" (You must do both tests to compare speed)
3) Options -> View Mode -> uncheck "Decode next image in advance"
4) View mode

My results:
Decode time with OpenCL ~ 3.26 sec
Decode time without OpenCL ~2.91 sec


Please. Specify what's your CPU and GPU. I'm interested in the performance gain of 6 or 8 cores CPU. And the influence of the faster video card.

[Emil]

GPU intel hd graphics 2500 . . .

Decoding with OpenCl should be significantly faster than w/o. I'd recommend to install a low price graphics card (E.g. GTX 1050) to replace the very weak Intel hd graphics. Decoding raws and edit mode will benefit, but database management will not.

[GusPanella]

I am not sure what is being timed...

​For instance,
​* maybe the goal is to look at the time it takes to move between images in Develop mode?
* maybe the goal is to look at the time it takes to move between images in View mode?

​System:
4 physical core (or 8 virtual core) ​(i7-6700HQ @ 2.4GHz.); laptop; nVidia 960M card ; NVMe U.2 SSDs connected to the PCIe bus (not SATA)

For any combination of the above, my reflexes aren't fast enough on the stopwatch button to actually measure a time, thus I will only be able to report "under 1 second" for any of the the test combinations.

[Alex_S]

I am not sure what is being timed...
* maybe the goal is to look at the time it takes to move between images in View mode?

Yes, View mode.

I am not sure what is being timed...
For any combination of the above, my reflexes aren't fast enough on the stopwatch button to actually measure a time, thus I will only be able to report "under 1 second" for any of the the test combinations.

Load time displayed on status bar. In the screenshot, you can see this.

[GusPanella]

That makes things easier...

System:
4 physical cores; ​(i7-6700HQ @ 2.4GHz.); laptop; nVidia GTX 960M card ; NVMe U.2 SSDs connected to the PCIe bus (not SATA)

My results:
Decode time with OpenCL ~ 2.5 sec
Decode time without OpenCL ~3.0sec
(​For the record, the embedded JPG preview is ~0.3)

The OpenCL response seemed 20% better than non-OpenCL in my test.

​For non-OpenCL
​Maybe the 3GHz. of the I5 is the reason for the improvement relative to the 2.4 GHz i7

​For OpenCL,
​It would seem the NVidia 960m would have an edge over the embedded Intel chip.

[rbtree]

I just saw this thread... I am running Ultimate 8, and will probably soon go for 10.

I didn't bother downloading the raw files, but tested my system with and without Open CL

System:
i6700K, overclocked to 4.5 Ghz, Samsung 950 Pro nVME, GTX 1070 graphics, 32 gb fast memory

Raw files about 7 mb larger than above test at 27-28 mb

With Open GL 1.28 seconds
Without 1.65 seconds

Later, I'll test two other PC's and an old i7 laptop

Ryzen 5 1600X (6 core 12 thread) overclocked to 4.0 Ghz, SSD but SATA 3, GTX 980 graphics

old X58 based ASUS P6T mb, but upgraded from first gen i7 920 CPU to a Xeon x5670 6 core CPU, with Windows 10 on SSD and GTX 1060 6 gb graphics.
By the way, that CPU cost $30 on ebay..it was ~$1400 when introduced in 2010 or 11