Tomas Brixi <tomas_brixi / yahoo.com> writes:

> Hello again,
>
> I did some finetuning of the FileHash ruby function from the first
> mail (atached at the end): read method changed to sysread and voila
> :-)
>
>
> mswin ruby jumped from 0.296 MB/s to about 14.250 MB/s !!!! with SHA
> hash cygwin ruby from 12.797 MB/s to about 24 MB/s with SHA hash it
> even outperformed python :-)
>
> A nice step ahead :-)
>
> But what's the problem with  File.read method ? Why it is so slow?

IO is a lot faster with the ruby 1.7 versions, at least under windows.

The sysread function bypasses all the slow parts, which is why you see
the speed increase.  I bet under ruby 1.7 though, the difference
between read and sysread won't be large.

-- 
matt