I know, the code is not completely stable, but...
1) As i've writed in other thread, there is a tools for gpu with CellML: Myokit.
2) New draft version of OpenCl (2.1) will supports C++ natively (and with SPIR-V the coding is more easy).
3) If you need help you can involve volunteers expert in gpu programming. Poem@Home admins, for example, have the collaboration of Vladimir Tankovich (he is also a Microsoft's employee).

Dear boboviz,
First of all, thank you for your interest and helping us improving the project.
Until now, our code is able to solve 0D (cellular) problems. Myokit uses GPUs to solve 1D(fiber) and 2D(tissue) problems, not the cellular problem. The cellular problem can't be paralelized, the only way to uses GPUs is to solve several problems at the same time (we want to explore this posibility).

We want to improve DENIS to be able to simulate 1D and 2D problems and this kind of problems will run in GPU. These problems are complex and we want to have a complete and stable version for the 0D problem before starting with them.

I have experience programming with CUDA, but, of course, if any volunteer wants to collaborate in programming tasks, our team is open. Now we are few ones but, the more the merrier!

All of you are given us a lot of ideas and we have many more, so there is a lot of work to do!

Thank you very much for everything (time, comments, hosts, ...).

Sincerely,
Jesús.

Thanks for the answer!!!