| 摘要 |
The equivalent computational precision in an associative memory is increased by determining the difference between the bit precision that is required in order to represent a given number in the memory and the bit precision that can be represented in a memory element of the memory, which is dictated by the inherent characteristics of the memory; determining, on the basis of the difference, the number of memory elements of the memory required in order to represent the given number with the required bit precision; and dividing the given number over the number of memory element of the memory, determining a base value to be loaded into the number of memory element and a remainder which indicates a subset of the number of memory element of the memory over which the remainder is to be divided.
|
