-
Notifications
You must be signed in to change notification settings - Fork 45
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
fill the demSeOth with zeroes so that oneRegi tests work #637
Conversation
Hmm. That surprises me. I know that @fbenke-pik was fighting quite hard against this filling of zeros, I think because it massively blows up memory use. What changes did you do to REMIND before you ran |
Ah? I assumed it was ok because of the documentation of readGDX saying:
It's linked to the removal of 32_power realizations, I'm trying to merge this: remindmodel/remind#1769 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I tried to look through your PR but couldn't find any reason that was well-defined before and not now. As noone is around (Michaja, Falk) that is an expert on these memory-related issues and make test
seems to work for you with this change, I approve it.
After you ran make test
, can you please also run Rscript output.R comp=F output=reporting outputdir=output/testOneRegi
and check whether that works with the normal memory configuration?
Would be great if you could check back with @fbenke-pik once he is back from holidays.
@lecfab Oliver is right, restoring zeros should not be the way to solve this (It might still be fine in this case if the data object is not overly large). The sustainable solution is fixing the problem by using the magclass helper Please revert this change in remind2 and provoke the REMIND error again by running "make test". Then fix it in reportSE using matchDim. I am happy to assist, once I can look at a reproducible error. @orichters Thanks for flagging this. |
@fbenke-pik: Sorry, I forgot to mention here that this is exactly what I suggested him later when it came to my mind, see #639 |
@orichters It was never really "well defined". The
Memory overhead for the bunch of zeros is 96 MB.
|
Thanks for the explanation. The memory issue should be fixed anyway with #644 |
I am aware. But this was as good a place as any to document how to figure out the memory use of |
While doing
make test
on Remind, the one-region run gave the following error:It is solved by filling v_demSeOth with zeroes where there is no value.