-
Notifications
You must be signed in to change notification settings - Fork 47
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Feature request: Improved memory management for root files #628
Comments
Like using heap space for this? I think this is the default, no? @tbaudier is this handled in python or C++ ?
Awfully difficult, we need several root files, one for each thread, with exclusive access to it, then combine every x minutes and at the end of the simulation combine all the root files. I think it is better to keep root things in heap space and store in disk only temporary results with non memory intense operations like stats or images. Anyway, my 2c |
Hi ! I dont think data are kept to memory and only written at the end. ROOT should take care of disk/memory tradeoff automatically. Are you sure ? |
Hi, looking at the output of the currently running simulation, it very much looks like it. The root files are created on disk but have about 400 byte(!), most likely, that is just the header. |
That's not RAM but virtual memory (Heap). In any case, a utility to merge root files from memory and write to disk is needed. This shall be made no more than once per hour as it is really intensive and will slow down the simulation for some minutes!! The only advantage is that you will have temporary results. This would be useful only for very long simulations. |
Hi,
I tried to run a simulation storing phase space data. The data stored can be extensive.
Currently, all data is stored in memory until the end of the simulation, then it is written to the hard drive.
Depending on the size of the phase space, this could be a few Gigabytes of data.
Do you think it would be feasible to implement a kind of direct storage? E.g. store the data to disk every x MB?
All the best,
Hermann
The text was updated successfully, but these errors were encountered: