Yes I agree and interestingly if you stick to basic algorithms with integers it can compute better. It’s almost like it doesn’t have true floating point capabilities but has somehow built itself integer operations correctly.