Punktiarvutusega

Punktiarvutus, known in English as fixed-point arithmetic, is a type of computer arithmetic where all numbers are assumed to have a radix point (or decimal point) in the same position. This contrasts with floating-point arithmetic, where the radix point can "float" to represent a wide range of magnitudes.

In fixed-point arithmetic, the number of digits to the left and right of the radix point is fixed for all numbers. This means that the position of the radix point is implicitly known. For example, a fixed-point number might be represented with 16 bits for the integer part and 16 bits for the fractional part. When performing arithmetic operations, the programmer must ensure that the radix point remains in its designated position or adjust it accordingly.

Fixed-point arithmetic is often employed in embedded systems and hardware accelerators where computational resources are limited. Its advantages include simpler hardware implementation, faster execution for certain operations, and predictable memory usage. However, it is less flexible than floating-point arithmetic, as it has a limited range of representable numbers and can suffer from overflow or underflow if intermediate results exceed the defined precision. Care must be taken during algorithm design to scale values appropriately to maintain accuracy and avoid these issues.