Denis Demidov

And with 6d4db1aeab360c591f70809bbe77d30832287677 you don't have to explicitly define `AMGCL_NO_BOOST`; the definition is included into `amgcl::amgcl` target when Boost is not found.

This error usually means what it says: the system matrix A has zero somewhere on the diagonal. The ILU implementation in amgcl does not have pivoting, so it can not...

How are you using amgcl then? If you set up ILU0 for a matrix `P` that is approximating your system matrix `A`, then your `P` is the "system matrix" for...

Ok, I see (sorry for not getting it from the first time). So it looks like your preconditioner matrix (`std::tie(mat_size, preconditioner_ia, preconditioner_ja, preconditioner_a) `) has zero somewhere on a diagonal.

The ILUt is somewhat tricky, as it often requires manual tuning of its `tau` and `p` parameters for a specific problem (and sometimes it is hard to find the working...

409a505b41fad1e724c983a27630b30525282d3b ports scipy version of minres into amgcl. It does seem to work, but it looks like the computed residual is too optimistic. I am computing true residual after the...

Looks like a synchronization bug somewhere. Unfortunately, these are usually hard to pinpoint. A few things I would try: * You are using 4 GPUs with a single MPI process....

> This solved the problem. How did the runtimes change after this? The "solution" is not very effective. Can you also enable the VEXCL_AMD_SI_WORKAROUND preprocessor macro and see if it...

I would check if adding something like ```cpp for(auto &q : x.queue_list()) q.finish(); ``` at the end of the loop [here](https://github.com/ddemidov/amgcl/blob/affc72ec009393134688e17110a8d38661d2a99a/amgcl/solver/bicgstab.hpp#L217) would still work, but be more efficient. This is...

I would check if the example is compiled with full optimization turned on (release mode in visual studio), and that no other processes compete for the CPU resources at the...