The Backend
Memory Management
For some calculations it might be a smart idea to process the requested angles
in chunks of a certain size. First, because it might result in fewer cache
misses and second one might be running out of RAM to run the calculations. To
this end, you might set the EADF_LOWMEM environment variable and use it
together with the blockSize property and the
optimizeBlockSize function.
Datatypes
We support single and double precision calculations. This can be set with the
EADF_DTYPE environment variable. Most of the time single precision allows
approximately twice the computation speed. This is reflected in the
dtype environment variable.
Environment Variables
A list of all available shell environment variables.
- EADF_LOWMEM: yay or nay?
- EADF_DTYPE: single or double?
-
eadf.backend.rDtype
alias of numpy.float64
-
eadf.backend.cDtype
alias of numpy.complex128