For the Windows version of PhotoLab: does the graphics card you are using make much of a difference in processing speed? I’m upgrading to 4K, need a new graphics card for that, and am trying to decide between the GTX1050 (640 ‘cores’) and the GXT1050Ti (slightly faster processor, more memory, 768 ‘cores’).
The first, slower card is 70% the cost of the second. If PL offloads processing, particularly if there’s parallelism (multiple threads/processes) the second, more expensive card had ought to speed up PL - and should get my $$.
Anybody know if a faster graphics card / more cores does speed PL up? There must be some gamer/photographers out there…
There were the beginnings of a conversation about this on the older forum. as I remember, the key to PL performance was processor/bus/memory/disk speed. Even processor cores didn’t seem to matter much, indicating low parallelism.
Thinking that if I throw money at PL performance, I should get a faster disk system (like a bus mounted SSD). Wrong?
Greetings,
All of those things matter. Yes, I would buy the Ti Card if it was an option, but I wouldn’t stop there. A solid state disk subsystem and faster CPU will all work together to improve system speed and performance. This will be especially noticeable in graphics intensive apps (like PL) or those that offload processing to a dedicated GPU. Additionally, the amount of installed system memory also affects the overall process and experience. Keep in mind however, there is a ceiling to this.
If you are still using plattered disks, you will be amazed at what moving to SSD will do for every aspect of your computer use and experience. I’d go SSD at minimum, m.2 if within your reach. 16-32GB of RAM minimum. My modest build performs extremely well with PL.
Z97X-UD5H (F8) \RM850w, 4790K, 16GB Vengeance, GTX970 FTW SC, 850 Pro’s RAID0, 1TB WD, W10 1803 - Dated now, but boots in 12sec and plenty fast for me.
I’ve started acquiring parts for my next build.
Z370-Gaming7 \850w, 8700K or 8086K, 32GB LPX 3200, GTX 1080Ti, 960 Pro m.2, W10
The difference in performance between the 8700K and 8086K is only 5%. For now the additional cost doesn’t justify the performance increase. But by the time I get around to actually building it, the latter will be under $400 so I’ll probably grab it. I upgrade more frequently than most because of the work I do, but still stay 12-24 months behind the curve for cost savings. I skipped the z170~270 architectures. I felt like both were rushed to market.
You can step into Sky / KabyLake - X series for even greater performance, but this puts you at a much higher price point. I have 7 machines in the house and don’t really need another computer. I game on xBox so having a x299 based board is not as important for me.
What you need to do is determine which option offers the best cost to value ratio. Spending money to upgrade an older platform is not always best. Sure migrating to SSD is always good, but investing in hardware that can be repurposed at least once (given to a family member, donated to a church) is a better investment in the long term.
We can talk about this all day long. If you want to provide your system specs, I’d be happy to make suggestions if you like.
Cheers
Hm, my experience with photo editing software is, that the GPU does not matter much, because it is not very often used in the way, how games use it. As I built my system for example, I bought a i7 7700k + GTX 1070. Because I did not like the ‘sound performance’ of the graphic card under heavy load, I sent it back. Until I bought the next graphics card, I used the CPU integrated graphics. I really could not see a performance difference in Lightroom and Affinity Photo at that time. CPU core count and frequency, enough ram and a fast SSD is what brings the most IMHO.
I will not upgrade to a new Intel CPU until they have done the shrink to 10nm in the second generation and have 8 cores @4.5 GHz in the consumer market. This will not happen before 2020.
i5 4 core / mobile), 32gb memory, 2x 512g SSDs, low-end-ish / fanless Matrox graphics card. I tend to focus on making systems as close to dead silent as I can. Used to w/c for that reason (but then had to deal with pump vibration). Simplest solution was a mobile i5 with big-ass heatpipe cooler.
I went for the ti, will report back if there’s a noticeable improvement. My focus is quiet/fanless. The ti is supposed to not run its fan until 80C, but I’m looking at a heatpipe cooler for it, in case that happens regularly (just using PS/PL)
I ran across an article examining PS CS6 differential performance with graphics cards. Ancient (2012) but germane for me as I’m still using CS6 (stopped upgrading when they went to software-for-rent). Adobe’s leveraging Open CL and Open GL, as is DxO. Hoping to get similar improvement.
And yes, I’ll eventually go i7 or Xeon if I decide I need more cores - and can get there quietly.
Off topic: I find it interesting that MS has responded to the increasing number of available Xeon cores to switching their server licencing to per-core instead of per-CPU.
I’ve installed a Geforce GTX 1050ti graphics card (replacing a card I couldn’t offload work to), enabled OpenCL, and exporting to disc is appreciably faster. I don’t see any change in the delay to display a photo with settings.
That’s all you can really expect (what its designed to do), slight decrease in the amount of time to process. You wouldn’t notice a difference in rendering unless you were displaying at maximum rez. A GPU is not a substitute for a good processor and more memory, but I’ll take it over no card any time.
I’m into quiet myself. Mid size case, big fans that rotate slowly. My systems usually sound like an air purifier set on their lowest settings. Central A/C is louder than my system. If you aren’t within 18 in. of the case, you can’t hear it. My case sits under my desk, so whisper quiet. I use Corsair PSU’s which support Zero RPM fan mode. My 970 FTW SC has a twin fan ATX cooler that is silent unless you are really pushing it, and even then you have to get next to the case.
I wasn’t expecting more than that, just reporting.
I’m into low-noise as well. For four or five years, I was watercooling for quiet, (big honking radiator, slow moving large diameter fan with quiet blades - there’s a lot of noise control in blade shape). But with low voltage processors and heatpipes it’s become unnecessary. The GPU I’m using has a single fan, but it’s not supposed to turn on until the GPU hits 80C. I’ve found a big heatpipe cooler for it (fanless) but that may not be something I need to bother with.
I’m running a low voltage i5, 32GB low voltage memory, lots of SSD space. Next mboard will have SSD on-bus.
I’ve been running the XMeters utility, which gives a good view into how a Windows computer handles workloads, including individual CPU core/thread load, memory use and storage and network performance (https://entropy6.com/xmeters/). At this point, however, it doesn’t monitor GPUs.
PhotoLab takes full advantage of all available cores/threads, which is especially useful when using Prime processing (all available cores/threads peg at 1005 when you do this). So the more cores/threads you have available, the better. It’s less clear how graphics card performance adds in.
As was pointed out above, an SSD can make a huge difference in performance for PhotoLab and other applications.
Hello
I use DxO on PC.
My main significant change was to upgrade system by using:
- SSD 240 Go Intel SATA 6Gb/s as “C:” for windows & system instead
- SSD as D: for Raw/jpg pictures during DxO processing
- 8 to 16 Go RAM
My graphic card is basic one,
I have a quite fast computer with Nvidia graphics card and 6-core Ryzen. Generally PhotoLab 2 works really fast. I can’t complain about performance. It’s important that any sliders work with real time effects without delay. However, I don’t notice any significant difference in the speed of PhotoLab 2 when hardware acceleration is enabled.