yade-users team mailing list archive

[Janek Kozicki <question695081@xxxxxxxxxxxxxxxxxxxxx>]

Question #695081 on Yade changed:
https://answers.launchpad.net/yade/+question/695081

Janek Kozicki posted a new comment:
>> Did you do anything specific to workaround this [MPFR] compilation error
> No. But after I followed you suggestion at #9, setting CHUNKSIZE=1 , this error disappeared.

Thanks for the answer, this means that we have more CHUNKSIZE related
bugs.

> May I mark your post at #7 as the solution?

sure no problem :)

If you feel adventurous, maybe you would  like to try building higher
precision versions of yade: long double, float128, maybe even MPFR150 ?

http://yade-dem.org/doc/HighPrecisionReal.html

you only need to pass this cmake parameter for each of the builds:

REAL_DECIMAL_PLACES=18
REAL_DECIMAL_PLACES=33
REAL_DECIMAL_PLACES=150

each of these builds generate a different yade binary with different default precision of calculations.
You might want to use cmake SUFFIX=.. parameter to generate yade binaries (and .so files) with corresponding different names.

long double calculations are 1.4 times slower, float128 are 4.9 times
slower, mpfr150 are 19.8 times slower.

-- 
You received this question notification because your team yade-users is
an answer contact for Yade.