Re: [Help-gsl] Memory issues when using GSL ODE solver

[Rhys Ulerich]

Hi Alex,

> I ran the example we discussed using C with valgrind and indeed found no
> leaks.

Excellent.  This was with GSL version 1.16?

> Then I wrote the same thing in Cython and ran the valgrind again with
> --suppressions=/usr/.../python.supp as suggested here,
> http://kratos-wiki.cimne.upc.edu/index.php/Checking_memory_use_with_Valgrind.
> And the difference looks like this:
>
> HEAP SUMMARY:
> ==7329==     in use at exit: 1,013,593 bytes in 245 blocks
> ==7329==   total heap usage: 7,559 allocs, 7,314 frees, 6,631,527 bytes
> allocated
> ==7329==
> LEAK SUMMARY:
> ==7329==    definitely lost: 0 bytes in 0 blocks
> ==7329==    indirectly lost: 0 bytes in 0 blocks
> ==7329==      possibly lost: 6,072 bytes in 17 blocks
> ==7329==    still reachable: 1,007,521 bytes in 228 blocks
> ==7329==         suppressed: 0 bytes in 0 blocks
>
> Does it mean that the problem I am experiencing is because of those 245
> blocks, which were not deallocated? Could you give me a hint on how to fight
> with the 'still reachable' and 'possibly lost' blocks? I guess you have some
> experience with that.

Likely that's the problem, but I don't know Cython well enough to be
sure.  Sometimes scripting environments or shared libraries will "hold
on" to memory until they exit or until they're asked to let go (e.g.
Intel MKL and FFTW both do this).

I'd ask the Cython folks how to proceed from here.  Sorry to not be
able to be more help at this stage.

- Rhys