That is a good question. I know from experience with my 8800GTS 512, you simply get better overall results with higher frame rates if you leave them locked.
Yes, it has to do with synchronisity and multiplicity. That is to say, it has to do with how long the data on one side must wait to link up with the data on the other.
so what happens if theyre set far apart? what happens if the shader is set too high while the core is left normal?
All I know is that the simpler the ratio between the two the better it works...but that the faster either side operates the better it works...so that it would be ideal if one side reached its limit as an even multiple of the other
so what happens if theyre set far apart? what happens if the shader is set too high while the core is left normal?
It might be possible that the higher-clocked side always seems to finish first, then it must wait until it could "sync-up" or reach a certain frequency at a certain point in time until they could pass data around. Faster clocks is still better though.