Need a super-fast filestream hashing algorithm

[cgallery]

I guess that doesn't entirely surprise me. Calls for checking a filesize can be slow. I still thought they'd be way faster than reading the entire file. Sounds like the MD5 (not familiar with) doesn't read the entire file?

[Alex Franke]

I guess that doesn't entirely surprise me. Calls for checking a filesize can be slow. I still thought they'd be way faster than reading the entire file. Sounds like the MD5 (not familiar with) doesn't read the entire file?

I guess that's what I don't get -- I thought there was an entry for file size in the file allocation table. It seems like of overkill to read in the whole file just to get the file size if the size is already up to date in the allocation table.

It's as if the MD5 algorithm is written in ASM -- maybe there's a set of processor-dependent implementations in the WinAPI or something. I could only get reasonably close using managed code.