If you have a 1000 files that are less than 8K in size and you use 8K clusters, you use 1000 clusters, and 8000K of hard disk space.
If you are using 32K clusters, you are using 32000K of hard disk space for the same files!!!
In short, use large clusters on NTFS ONLY for big files. The advantages of big clusters for large files are:
* IO are faster
* the MFT is smaller (less clusters to track)
And put small files somewhere else...
Creating a specifically sized pagefile on a dedicated volume on a dedicated volume is a good idea _except_ if it is a very old slow drive... The file system really does not matter at this point except maybe that NTFS wastes space for the MFT when we really don't care about protecting files enties. After all there is only one: pagefile.sys