Dynamically pull GPU parameters

[notnotnotveg]

Currently, the program has hardcoded max grid and block dimensions.

However, this should be dynamically pulled, to support different GPU architectures.

As an example,

from numba import cuda
device = cuda.get_current_device()
print("name = %s" % device.name)
print("maxThreadsPerBlock = %s" % str(device.MAX_THREADS_PER_BLOCK))
print("maxBlockDimX = %s" % str(device.MAX_BLOCK_DIM_X))
print("maxBlockDimY = %s" % str(device.MAX_BLOCK_DIM_Y))
print("maxBlockDimZ = %s" % str(device.MAX_BLOCK_DIM_Z))
print("maxGridDimX = %s" % str(device.MAX_GRID_DIM_X))
print("maxGridDimY = %s" % str(device.MAX_GRID_DIM_Y))
print("maxGridDimZ = %s" % str(device.MAX_GRID_DIM_Z))