It's well known that finite volume method can be classified into two classes:cell vertex finite volume (CVFV) and cell centered finite volume (CCFV). Some say that the CVFV is better than the CCFV in memory and accuracy, but some say not. I wonder which one is better? Can someone shed light on it?

There are no much difference between them for FV. To implement boundary condition better and to avoid corner sigularity better, CVFV is preferable, at the cost of slight higher CPU time.