Hi MTEX community,
I would like to submit this proposal of feature request, in fact it is more an optimization request, which might interest people using large 2D EBSD data and now 3D EBSD data and who don't have access to HPC with large amount or RAM.
I think it could be relevant to adapt Data Type of variables used in the MTEX source code to single precision whitout sacrificing too much in precision.
MATLAB uses by default double precision float which costs a lot in RAM memory and calculation effort.
Typically using double float precision by default use twice as much memory than single float precision and same for integer.
What are developers feedback / suggestion regarding this request?
I don't know how much effort this requires and if it is relevant or feasible regarding inherent MATLAB constrained.
Cheers !
François
Hi MTEX community,
I would like to submit this proposal of feature request, in fact it is more an optimization request, which might interest people using large 2D EBSD data and now 3D EBSD data and who don't have access to HPC with large amount or RAM.
I think it could be relevant to adapt Data Type of variables used in the MTEX source code to single precision whitout sacrificing too much in precision.
MATLAB uses by default double precision float which costs a lot in RAM memory and calculation effort.
Typically using double float precision by default use twice as much memory than single float precision and same for integer.
What are developers feedback / suggestion regarding this request?
I don't know how much effort this requires and if it is relevant or feasible regarding inherent MATLAB constrained.
Cheers !
François