Page 1 of 1
why are vdfx double the size than vdf files?
Posted: Thu Jul 22, 2021 8:59 am
by kleemax
When updating to Vensim 8.x the resulting simulatin files (vdfx) are roughly double the size than when I run the same simulation in an older Vensim version, generating a vdf output.
For us this is relevant because due to model size, the old ones were slightly above 0.5GB now they are above 1GB...
What's the main advantage of vdfx anyway? If you have a text that describes this somewhere that you could link to, that would be helpful.
Can Vensim 8+ be switched back to generating vdf files?
Re: why are vdfx double the size than vdf files?
Posted: Thu Jul 22, 2021 9:03 am
by Administrator
VDFX stores double precision numbers, VDF only stores single.
And no, it cannot be switched back to VDF files.
If size is a problem, you can use a savelist to cut down the amount of data you save.
Re: why are vdfx double the size than vdf files?
Posted: Thu Feb 17, 2022 10:51 am
by devM5
Is there a way to have Vensim save simulation results as single precision numbers, even though the DP version of Vensim is installed?
As far as I can see, there has been no SP version released with version 8.2.1 and newer. The simulation size of over 1 GB fills a local hard disk with 250 GB of space quite quickly when doing multiple simulation runs. Therefore, it would be helpful to still have the SP option, in which simulation runs had only 1/2 or 1/3 of the size.
I've noticed a parameter
/DDBLPREC in the CL-command when using compiled simulation, I wonder if there is a way to compute variable results as SP with the external compiler?
Code: Select all
CL /c /I "%2" /DDBLPREC /DWIN32 /DWIN64 /Dx64 /D_x64 /D_WIN32 /D_WIN64 mdl.c
Re: why are vdfx double the size than vdf files?
Posted: Thu Feb 17, 2022 10:54 am
by Administrator
devM5 wrote: ↑Thu Feb 17, 2022 10:51 am
Is there a way to have Vensim save simulation results as single precision numbers, even though the DP version of Vensim is installed?
As far as I can see, there has been no SP version released with version 8.2.1 and newer. The simulation size of over 1 GB fills a local hard disk with 250 GB of space quite quickly when doing multiple simulation runs. Therefore, it would be helpful to still have the SP option, in which simulation runs had only 1/2 or 1/3 of the size.
I've noticed a parameter
/DDBLPREC in the CL-command when using compiled simulation, I wonder if there is a way to compute variable results as SP with the external compiler?
Code: Select all
CL /c /I "%2" /DDBLPREC /DWIN32 /DWIN64 /Dx64 /D_x64 /D_WIN32 /D_WIN64 mdl.c
No, it cannot be done. When we moved to 64 bit, we also had to move to double precision.
You could enable compression on the folder if using NTFS file system on Windows. Or use a save list.
Re: why are vdfx double the size than vdf files?
Posted: Fri Feb 18, 2022 8:27 am
by devM5
OK, thank you!
Re: why are vdfx double the size than vdf files?
Posted: Mon Feb 21, 2022 3:54 pm
by tomfid
Another option would be to use a savelist to reduce the amount of data to be stored.
Just thinking out loud, what you'd really like to have might be an "excludelist" rather than a savelist. Typically you only get to file sizes like this with subscripting, and the majority of the data is going to reside in just a few variables with maximum dimensionality. So what you might want is to exclude x[a,b,c,d] and keep everything else.