Re: dentry bloat.

[Jens Axboe <axboe@suse.de>]

On Fri, May 14 2004, Paul Jackson wrote:
> > so I guess you do.
> 
> Sorry - I'm being thick.
> 
> Is the new hashing good or bad?
> 
> (Usually, performance is thought of as something 'good', so when you say
> it is 'brought down', that sounds 'bad', but since it's ms/iteration,
> I'm guessing that you mean to say that the ms/iteration is lower, which
> would I guess improves performance, so I'm guessing that bringing
> performance down is 'good' in this case, which is not idiomatic to the
> particular version of English I happen to speak ...  So please favor
> this poor old brain of mine and state outright whether the new hash is
> good or bad.  Does the new hash makes performance better or worse?)

:-)

I can only say the way I read the numbers, the new hashing scores higher
ms/iteration which is a bad thing. So when it is stated that
'performance is brought down' I completely agree that it describes the
situation, performance is worse than before.

First table shows 2.6.6 (with old hash) doing better than 2.6.6-BK with
new hash. It then shows 2.6.6-Bk with old hash doing worse than 2.6.6
still, so it's not just the hash that has slowed things down.
2.6.6-new_hash does worse than 2.6.6-stock.

-- 
Jens Axboe