Grasshopper vs. batch process performance

hey all,

I have developed a plugin connecting Finite Element Software with Grasshopper and I am struggling with a performance glitch. The basic concept here is that my plugin creates a batch file to run FEM calculation, executes this batch file, freezes the GH thread until the calculation is done and reads back the results. Everything works and up till now I had no problems with it.

The problem started with a new computer. On a new machine, the performance of the batch process started in GH is dramatically lower (10x), than if the same batch file is started outside the GH by hand. Theoretically, there should be no difference if the calculation is started by GH plugin automatically or if by hand - literally the same batch file is executed, starting literally the same calculation. But that's somehow not the case.

Everything is kept on a local disc and the CPU usage during the calculation is low. Where do I start looking? Does Rhino / GH has some performance limits which could slow my calculation down? Generally, help :)

Thanks,

wojt