Two aspects of difference are important:
Memory-wise:
[n,m] - Saved as a long bunch of memory, as if it was [n*m]
[n][m] - Saved as a simple array of size n, where each element is a pointer to an array of size m.
Access-wise:
[n,m] - To access cell i,j what really happens is that it takes the pointer of the [n*m] array and offsets it by n*width+m, and then accesses the value.
[n][m] - To access cell i,j you just access the sub-array pointer at index n (offset #1), and then access the sub-array at index m (offset #1).
[][] Is better in both aspects. More efficient in access and more flexible in memory.
Also, you can access a single row just once and process it even more efficiently, since you don't perform full multidimensional access for each cell in that row.
There is, however, an advantage of [,]: You always know that the width of the matrix is fixed. Using [][], each sub-array can have different lengths, and can even be null. This can be considered as an advantage, although the opposite can be useful as well, at times.
