Create a Post
(This consumption is only during sync.)
(System) 8 core, 32GB RAM, GPU RAM 4GB dedicated
"Photo Organizer" at one point used over 4 million memory (working set), and topped out the CPU at 100 percent for over two hours. This morning I had to restrain memory and restrict "Photo Organizer" to use only 4 cores to reduce temps and increase responsiveness.
Currently into 26 straight hours in the sync process and only about two thirds complete. Keep in mind I had not restrained the program till this morning.
Now the way I have set it up will control the situation, but suggest keeping this in thoughts for improving since I'm sure some others will encounter and not know how to restrict such usage, and could cause overheating and system shutdowns.
I do understand that I have a very large number of images (34K+) to process and this most likely will only affect others with such high numbers.
(Disclaimer / Advice to Others)
Photo Organizer is a great program and is taking great strides to enhance and improve. So don't let this report discourage using and help improving by giving feedback.
Thank you very much for your report.
I am also experiencing issues with my private photo set using a similar machine.
In response, we reduced the number of threads used in the last update, but this might not be sufficient. We may need to consider another throttling method.
To prevent running out of memory, the app monitors memory usage and decides whether to launch more parallel threads. I believe we need to apply the same approach to CPU usage.
The free version should be fine; we have not encountered these issues there.
The problem arises with face detection, which is part of the pro subscription for focus assessment. Face detection is performed locally using the PyTorch RetinaFace ML model, which consumes a lot of resources. I need to research more to find a better way to manage this.
@stefan
So, after you mentioned it, I looked at what stage of sync it was in, and you're absolutely right. In "Face Detection."
Although it is useful, you might consider that as an option to use in settings.
@stefan
I believe it may be a bigger problem such as a major memory leak. Doing some testing I made sure sync was off, then I changed the CPU priority to an "idle" state, set memory priority to "low", and minimized Photo Organizer to the task bar. So, it should have used very little memory, yet watching the program with ProLasso the memory working set still went up to 6 million kb in less than an hour. Looks like a leak, not saying it is for sure. As much as I hate to say this, Photo Organizer is not a usable option at this time. The program also gained 2GB in size.
Thank you for sharing your findings. We will investigate this further.
Please note that synchronization starts regularly every 15 minutes. Was the synchronization started during this time?
Perhaps we should provide a control level for this option.
Who is the maker of the ProLasso tool? Do you have a link for me?
@software @stefan
I checked every 10 minutes to see if it had started.
Pro Lasso https://bitsum.com (Very trusted software)
confirmed with task manager.
Just started with this program, how long will it take to sync the folders, what is it doing when you say sync does it produce a new file? I thought it would read the current file information and put into database at this rate the program may be unusable
@alantrinder
Hi,
Thank you for reaching out.
The total duration of the course depends on the number of photos you need to scan. The more photos you have, the longer it will take.
The sync process does exactly what you expect: it reads file metadata, calculates a visual hash for duplicate detection, detects faces in the photos, performs a focus assessment on each face, and stores all the extracted and calculated data in a local database file.
Face detection and focus assessment occur on your device and are resource-intensive processes. For now you may want to wait until these are complete before working with the app.
We call it "sync" because, after the initial data gathering, we only look for changed files to keep the database up to date with the sources.
I hope this clarifies things. If you have any further questions, I’m happy to answer them.
Kind regards,
Stefan