Two bug fixes in CPU MHD Dedner solver:
Grid_UpdateMHDPrim.C: EOS(eint...)->EOS(eint1...) in DualEnergyFormalsim update
One bug fix in GPU MHD Dedner solver:
Memory checking for cudaMHDMalloc was not correct when there are multiple subgrids. Solution: move GPUMem/GPUMemOffset to a grid member variable by adding them to cuMHDData structure.
Two added features in GPU MHD Dedner solver:
Hi Peng, these changes look great -- thank you for the bug fixes and ability to run with dual energy formalism! I don't see any reason to hold up acceptance by just looking at the diff below. I don't have access to the necessary hardware at the moment to test, so perhaps if someone else gets a chance they could do so. If not, I'd propose merging this in a day or two.