RE: [patch v3 6/7] crc32: add-slicing-by-8.diff

[Joakim Tjernlund]

"Bob Pearson" <rpearson@xxxxxxxxxxxxxxxxxxxxx> wrote on 2011/08/09 22:52:34:

> From: "Bob Pearson" <rpearson@xxxxxxxxxxxxxxxxxxxxx>
> To: "'Joakim Tjernlund'" <joakim.tjernlund@xxxxxxxxxxxx>
> Cc: <akpm@xxxxxxxxxxxxxxxxxxxx>, <fzago@xxxxxxxxxxxxxxxxxxxxx>, <linux@xxxxxxxxxxx>, <linux-kernel@xxxxxxxxxxxxxxx>
> Date: 2011/08/09 22:52
> Subject: RE: [patch v3 6/7] crc32: add-slicing-by-8.diff
>
> > >    - extend largest BITS size from 32 to 64
> > >    - extend table from tab[4][256] to tab[8][256]
> > >    - change algorithm to align on 8 byte boundary
> > >      (it was noted that all that is really required
> > >      is that the inner loop must comsume 8 bytes.
> > >      As written it can start on even or odd 4 byte
> > >      boundary.)
> > So why not do that in the code too?
> >
>
> I did the experiment with the random test set and it is a couple of % faster
> so I made the change. I had thought that it wouldn't make a measurable
> difference.

That could depend on how the test data table is aligned. What if you change
the alignment of the table?

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/