recommended is 64k due to that being the size of an extent.
An extent is 1 chunck of data that SQL requests. Having your block size and extent size the same means that when SQL tries to read a 64K chunk, it can read 1 block which will be continuous and thus quick to read. If you have a 4K block size, it needs to get 16 of them from disk which may or may not be in physical order on the disk.
Lets say that it is a worst case scenario setup. SQL asks for 1 extent, the first block it gets is the first block on the disk. The second block is the last block on the disk. Third is the second block on the disk and so on until you have all 16 blocks. Presuming this is a spinning platter disk, it would need to jump back and forth across the disk repeatedly to get your 16 blocks. Assuming this is an SSD, it would need to do 16 separate read requests from the disk.
You put that as a 64K block size on the disk, when SQL asks for it, you only need to get 1 block and there is only 1 disk seek. It will perform a lot faster when you have a heavy read operation.
As for finding the block size it was formatted as, I do not know of a "good" way, but you can run chkdsk against the partition and when it is done, it will tell you the number of "bytes in each allocation unit" which is your block size.