Hi Marek,
thanks for a great answer.
Is it not possible to allow full precision for user provided configuration values? I realize that calculating an optimal ratio to achieve a target output frequency can be computationally intensive, but that is not an issue for user provided values. Maybe perform the frequency calculation with limited precision as is done at present, but allow full precision user provided values?
If it is not practical to allow full precision user provided values it would be nice if the clock tool could message the problem in some way, perhaps a hint in the "Computation failed for element" error? Maybe "Computation failed: allowed PLL resolution exceeded" along with discoverable help to explain the issue.
Our use case is phase locking timers and various other clock based I/O to an external time reference. We would most likely set DENOM to a fixed large value and adjust NUM to fine tune the output clock frequency. We are not setting an initial high precision clock frequency. However we do need to use the full 30 bit precision when the system is running to dynamically control the clock frequency in a firmware PLL.