I took the libery of posting an issue together with the commit that introduced the bug here: https://github.com/JuliaLang/julia/issues/11698
Thanks for finding this, I also use cholfact for FEM analysis but was not aware of this. On Friday, June 12, 2015 at 8:28:26 PM UTC+2, Mauro wrote: > > This is a bug. Also note that 0.3.9 works just fine. If there is no > issue on this could you file one? > > On Fri, 2015-06-12 at 19:50, Felix Henneke <oim...@googlemail.com > <javascript:>> wrote: > > hi, > > > > i have a problem with some finite element code, where i have to solve > > several sparse systems of equations. i noticed that solving the > equations > > consumes a lot of memory. for the problems i consider i quickly run out > of > > memory. interestingly the memory allocation is not captured by the @time > > command. but i can see the memory consumption in htop for example. i > > reduced the problem to the following simple code which causes the > > allocation problem. > > > > function f(n) > > A = speye(n) > > fact = cholfact(A) > > for i=1:1000 > > fact = cholfact(A) > > end > > end > > > > for n=10_000, i see that around 1GB of memory is allocated, but > according > > to @time it should only use around 20MB. the memory is not freed until i > > quit the julia session. > > is this a bug or am i missing something here? > > > > julia> versioninfo() > > Julia Version 0.4.0-dev+5317 > > Commit b3a7342* (2015-06-11 15:18 UTC) > > Platform Info: > > System: Linux (x86_64-linux-gnu) > > CPU: Intel(R) Core(TM) i5-4200U CPU @ 1.60GHz > > WORD_SIZE: 64 > > BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Haswell) > > LAPACK: libopenblas > > LIBM: libopenlibm > > LLVM: libLLVM-3.3 > >
