Are you asking why the tools took different amounts of time to do the synthesis, or are you asking why the synthesis result was a faster digital design?
Either way, I would think this is a hard question to answer without having some pretty deep knowledge about how each of the synthesis tools are implemented and some more specific info about your input RTL. Were the outputs from the different tools significantly different?
Some ideas? One tool was simply lucky, in that how it was implemented it stumbled onto a solution that satisfied the input constraints faster that the other one, or was just lucky to create a faster solution. Perhaps one tool had a specific capability to recognize and synthesize a particular digital structure that the other tool did not. Maybe one tool was simply coded more efficiently than the other, or has heuristics to help it skip unneeded steps when it can.