Issue from https://github.com/LoLab-VU/pysb-cupsoda/issues/5 originally by @lh64:
Currently, if the user passes more parameter sets ('c_matrix') than available threads on the GPU, cupSODA will simply exit with an error. It would be nice if PySB could automatically detect situations like this and run cupSODA multiple times in sequence. This will require querying the GPU architecture and having an accurate way to estimate the number of simulations that can fit on the device.
Issue from https://github.com/LoLab-VU/pysb-cupsoda/issues/5 originally by @lh64:
Currently, if the user passes more parameter sets ('c_matrix') than available threads on the GPU, cupSODA will simply exit with an error. It would be nice if PySB could automatically detect situations like this and run cupSODA multiple times in sequence. This will require querying the GPU architecture and having an accurate way to estimate the number of simulations that can fit on the device.