Not much.
“New in DxO PhotoLab 9.3.2 Bug fixes • Minor bug fixes and improvements”
Nothing more - nothing less.
Had some failures using premade AI masks but noticed that the VRAM stayed in the middle range (16 GB or so) rather than climbing 4-5 GB each image processed until hit 18GB then crashing.
Using Surface 10 Pro for Business with 32 GB, Intel Ultra core 7, Intel Arc GPU with 18 GB shared.
Ron
Hi Ron - I finally have hopefully found someone using PL9 with a similar PC set up to my own!
I have an Azus Zenbook with an Intel Ultra 9 285H CPU, 32Gb Ram, Intel ARC 140T iGPU with 18gb shared and Intel NPU AI Boost for Core™ Ultra, which I think is very similar to your own setup?
I tested out PL9 on the machine when it launched in September, and had failures when trying to access the Loupe magnifer and the full screen previews when using Deep Prime XD images as well as all of the initial problems with exporting and AI masking that have been well reported with the initial release of PL9.
My trial expired after 30 days and I am currently not able to test my setup with the latest revisions to PL9 - I was wondering if you would be able to share your experiences with using PL 9 and the updates/improvements that DxO have introduced when using the Intel Arc iGPU ?
I am currently continuing to use PL8 on my machine and it has worked flawlessly until the release of PLV8.10 when accessing the Loupe magnifier caused a failure with a black screen requiring a re-start of PL and a reboot of windows. This was the exact failure that I experienced withmy trial of PL9.
The only way that I could stop this version crashing was to select “CPU only” in the drop down accelaration menu - but this resulted in the loupe taking an age to focus and exports of deep prime images taking 1 minute 40 seconds rather than the 15 seconds when using PL8.9.
With the release of the latest version of PL8 - PLv8.11.1 build 695 I now find that despite all of the options on the accelation drop down menu being there (Auto Selection, CPU Only and Intel ARC 140T GPU) auto selection and Intel ARC 140T cannot be selected and the setting always defaults to “CPU Only” with the Loupe focus and Export problems remaining the same. A couple of other forum members seem to have commented on this same probelm.
My suspicions are that this could be due to some coding from PL9 being introduced to PL8 and the latest “CPU Only” default being introduced by DxO as a workaround rather than a fix. Only a guess tho’ in reality
Have you experience anything like this with PL9?
I have an as yet unanswered support ticket raised with DXO about this along witha request to allow me to download another trial version of the latest version of PL9 to test this out on my machine.
Any learning or advice will be much appreciated -
Cheers DaveH
Seems a good idea to check the GPU driver version.
Hello Andras
Good to meet you here. I follow the WOMBAT as close as I can to learn a few things about not only how to get the most out of PL9.x.x software but also grafix computing in general.
And as you say the GPU drivers get changed.
Dave that is the first thing I check after a Windows update and I have had a few lately including one labeled “repairs.”
Since the Surface is a Microsoft product I often get regression of the GPU drivers on updates sometimes multi drivers back. I am slow responding because I have a tower in the radio room and PL 9.3.1 installed on it. I checked it and as far as the GPU selection goes I have the same options as you except I can select them. Actually, I have one more included I think beginning with 9.0 and that is Intel’s NPU (Neural Processor Unit- Intel (R) AI Boost). I think it requires a separate download and installation. It is as slow as the “Only CPU” option
I also discovered the driver was changed to one I had never seen before. Tracking it down on the Intel Driver page I think it was for a GPU Card but I’m not sure. I also tried to update using the Device Manager and it said Microsoft said I had the best driver installed.
This is the current driver on my (on board) system is 32.0.101.8331 You’ll have to do a search for Intel GPU drivers.
I’ll check back and see if this helped you.
Ron
Is your processor the same as mine? I have the Intel Core 7 Ultra 265K. It’s the first neural processing unit (NPU) in a desktop CPU. I have an Arc 570 with 10G RAM for a GPU and 32G DDR5 6000 RAM. The system runs pretty well as it is, I haven’t selected any special options. I’m currently running 9.2 because I’ve been afraid to update with all the problems.
No not a desktop for primary use. My system is a Microsoft Surface with Intel core 7 165U 2.10 GHz with 12 cores and 14 threads. It isn’t a desktop and not really a laptop. It isn’t totally sealed but close to that. I can replace the C: drive card with the system and some storage on it. I have a 1 TB card installed. It has 32 GB LPDDR5 RAM. The GPU is Intel (R) Graphics Integrated, 2000 MHz, 18 GB shared VRAM.
I am able to use PL 9.3.1 and PL 9.3.2 without problems as long as no AI predefined masks are used. I have limited use of about 1, sometimes 2, predefined masks and DeepPrime 3. XD2s is iffy with predefined masks but useable with limited manual masks. I limit the manual masks when I use them to just what I feel I need, up to 3.
I have noticed that prior to PL 9.3.1 and maybe even 9.3 that every file processed took up about 4 GB of available VRAM and never released it when it got to 18 GB it crashed. for me that was about 5 images processed with minimum AI involvement. I don’t have 9.3 installed now but I now notice that VRAM use goes up to about 16 GB in the first image processed then levels off with each succeeding image processed around that level when using DeepPrime 3 noise reduction and manual masks. I don’t know how many manual masks it would take to cause a crash. But this is a new behavior from PL9.3.1 and could have started with 9.3.
I have Dell tower in the Radio room that has an NVIDA RDX 5070 (plain) 12 GB dedicated VRAM and an Intel (R) GPU. and 64GB DDR5 RAM. Both show 36.2 GB Shared Video RAM. It has all the power it seems to process whatever comes at it.
It was in stock, on sale and needed to replace a Win 10 system. Of course, I selected it as an upgrade to my old system because of the increased use of AI in a lot of applications.
I prefer using the Surface because of the portability and small footprint.
My use case is processing Wildlife photos for my personal enjoyment with limited distribution to friends.
I have gone overboard with my reply and I’m sorry about that but…
Ron
Hi Ron, thanks for getting back to me 
I had to look up your “I follow the WOMBAT as close as I can” quote! 
I am running the same GPU driver as you are - 32.0.101.8331. Like yourself windows update will install the MS drivers over the Intel. but I loose the shared memory override funtion with the MS Drivers and allways re-instal the latest Intel drivers.
I’m a little confused over your comment that you checked that the acceleration options seem to work for you with PL9.3.1 on your tower machine rather than your surface pro machine?
Do you have the PL9.3.2 version on your surface pro?? and if you do, are you able to select all of the options without it defaulting to the “CPU only” setting as I am finding with PLV8.11? ( better still, are you by chance still running PL8 on your surface pro?)
I can confirm that when I was running the trial version of PL9 on my Asus the option to run the Intel NPU as one of the choices was displayed and like yourself I found that this ran as slowly at selecting the “CPU only” setting. Incidently, I receive the NPU driver updates from Intel® Driver & Support Assistant (https://www.intel.com/content/www/us/en/support/intel-driver-support-assistant.html), and weirdly ajust as I am typing this an update for the NPU has just arrived on my laptop! Scary Stuff!
I can recall treading somewhere when I was trialing PL9 that DxO at that time did not use the capabilities of the Intel NPU but a quick google two minutes ago produced this google AI result( if anyone can trust this?)
“Yes, DxO PhotoLab 9 (Elite edition) leverages Intel’s NPU for specific AI-powered features, provided you have a compatible processor with the Intel AI Boost NPU.
The software uses the NPU to accelerate demanding AI tasks such as:
DeepPRIME 3 and DeepPRIME XD3 X-Trans noise processing.
AI Masking.
These features can still run without a compatible NPU, but performance will be slower as the processing falls back to the CPU or a compatible GPU. For the best performance, DxO recommends a dedicated GPU or a modern Intel Core Ultra processor with the integrated NPU. “
Cheers and that you for your assitance ( and Andras!)
Dave
Just realised I forgot to ask you if you are using “Open CL” in performance setting!
Dave
@dh57 Trying to remember your questions, I’m over 40 and getting forgetful.
Ron
I expect you noticed @Andras icon/symbol “street WOMBAT”, my reference.
Cheers Ron - getting forgetful at 40??? add 28 years and you’ll be like me with a notebook and pen!
Sounds like I’ll continue to use PLV8.9 as it works for me and does everthing that I need from it and wait to see if I get a response from DxO on the “CPU Only” issue in PLV8.11.
If DxO do grant me an additional trial of PL9 I’ll probably try it out again to see how it performs with the latest refinements. The AI masking did look good and something that I could use when I did get it to work on the trial, but with the continued problems that you report with it I doubt that I would be willing to part with +£100 for something that doesn’t yet work as it is supposed to. I’ll maybe miss our PL9 and wait until PL10 arrives in sept 26 when hopefully the AI mask problems have been eliminated - I have had a trial of ON1 26 and the AI masking there seemed to work perfectly on my machine - but I found it very difficult to use and I dont think the end results are anywhere near those when using PhotoLab.
I’ll keep an eye on your posts to see how you are getting on with PL9 and I I get a positive response from my support ticket I’ll keep you posted
Merry Christmas, and thanks for your help,
Dave
@dh57 Hi Dave and before I forget "Happy Christmas and have prosperous New Year.
I have been using the “over 40” bit for several years now. As for the notebook and pen, I do that but forget where I lay them down and the housewide search is on again.
When you have time after the holidays and you haven’t received a reply from support, you may try to download *.11 again and use this method. I hardly ever delete things I download and going back in the download directory I downloaded PL8.9.200355 662 on Sep 30 2025. I got the specific file name by hovering my pointing arrow over the file in the directory is just says PL 8 (x) otherwise. If you have the same habit check and see if you have the same file, PL 8, very close to that date and see if it is the PL8.9 by hovering over it. File size is 348,096 KB.
If you have the file then go to your DXO account and download the PL8.11 from the directory. Once it is downloaded go to Control Panel, Programs, Uninstall and uninstall PhotoLab 8. If you have two or three entries for PhotoLab 8 and if both 8.9 and 8.11 are installed, then uninstall both. Exit the Control Panel.
Install the PL 8.11 file again then test and see if it works for you.
Now for the disclaimer: If you are uncomfortable with this action or don’t have the PL8.9 install file then do nothing until you get a response from DXO support.
Just my take here
“part with 100 pounds for something that doesn’t yet work as it is supposed to.”
“part with 100 pounds for something that doesn’t work as I expect it to.”
I have dropped Topaz this year. I had been with them since it was advertised as the last program you will ever own. All updates both major and minor for life. It was expensive but I bought it. A year or so more later AI/Machine Learning happened and they had to go from hiring only programmers to also hiring engineers. One day all owners got an email saying from then until one year from that date they would receive all updates. After that they would have to pay for the next update. I skipped a lot of the updates but last year I bought Topaz during the Black Friday sale. When I ran it is said my computer wasn’t powerful enough to use the program but did me give me an offer to upload to their system and it would process the files for a charge. I decide that I wasn’t in a position to update equipment or pay to upload files for processing. I only found Gigapixel and JPG Noise Removal to be something I would use from time to time. I’ll go back in my files from former computers to see if I can find a copy of Gigapixel for my Sony FD 91/92 and CD 400 files to upscale from 640 X 480. I shoot RAW now so jpg NR isn’t needed.
I have found that PL 9 has improved a lot from 9.0 up to now and expect it to continue but it probably will be a while before meeting all expectations. Kind of like cameras, I am near the end of the chase.
The price increases might be going from paying programmer salaries to paying AI Engineer salaries and before I went from Windows 10 in the radio room to Win 11, more memory and a more powerful GPU I expected that PL 8 or PL 9 would be my last processing program and that may still be a fact.
Again if you don’t have a PL8.9 installation file then stay with what you have now. It is working.
Finally, I think AI has a way to go, I have pictures of Screech Owls taken during a night owl prowl in the Amazon basin. One had a limb across its chest. and every effort to remove the limb resulted in 3 eyes, 4 eyes, no beak or a two tone color scheme using Adobe Photoshop Elements Removal in Photoshop Elements 2025. A case of if you aren’t willing to pay the price you will end up paying the price.
Again Happy Holidays to all. WOW! I remembered.
Ron
Good morning from a very overcast and bleak north-east England Ron!
Sound like we are kindred spirits with regards memory going missing - but I’m not talking VRAM here!!
I bought my Asus laptop in September and downloaded exactly the same version of PL9 from my account that you mention whilst setting it up so I have a copy of it.
I have also used windows control panel to delete the PL9.10 update and the full download of PL 8.11 that I downloaded from my DxO account and then reinstalled V8.9 that works for me with the ability to select my Intel 140T IGPU rather than CPU only.
I may try again with V8.11 as you suggest , but I’m not really sure what upgrading from V8.9 to V8.11 may give me? - I’ll wait to see what DxO come up with on the support ticket … not really holding my breath there tho!
I note that they are now trying to sell priority support! Given my experiences when contacting DxO to date I really hope that this won’t merely mean an even poorer customer experience for those of us not willing to flash the cash towards DxO!!
Happy Holidays and all the best for the new year!
Dave
@dh57 Good Morning, your early afternoon, from Similarly overcast South Central Texas. The clouds are forecast to break up to mostly cloudy later in the day.
Sounds like you are back in business with PL8.9 at least until you feel PL9 works to the level you want/need.
I am happy with PhotoLab overall and especially after mid PL7 when they added ‘create color profile.’ Before that if I tried to improve the color I would usually be met with ‘Nice processing but the bird isn’t purple.’ or similar. I have severe color vision deficiency. Now, unless I forget (again), my first shot is of the calibration card I bought. I know I should take more shots as the lighting changes, but I am so caught up in getting the shot I don’t. I have found that comparing the histogram of the calibrated included come close to the DxO camera and lens files provided. There is very little change in the histogram.
I am off to our daily 3 mile walk. The best to you and your family for the holidays.
Ron