numpy is enormous and i think the only thing we're using it for is to save us the trouble of defining a vector class. defining our own with only the functionality we need might be quicker anyway plus we might want to transfer all geometry processing to the GPU so numpy might just get in the way?
numpy is enormous and i think the only thing we're using it for is to save us the trouble of defining a vector class. defining our own with only the functionality we need might be quicker anyway plus we might want to transfer all geometry processing to the GPU so numpy might just get in the way?