Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 29 Sep 2006 19:11:56 -0400
From:      Randall Stewart <rrs@cisco.com>
To:        John-Mark Gurney <gurney_j@resnet.uoregon.edu>
Cc:        freebsd-net@freebsd.org, Mike Silbersack <silby@silby.com>, freebsd-current@freebsd.org, Andre Oppermann <andre@freebsd.org>, gallatin@cs.duke.edu
Subject:   Re: Much improved sosend_*() functions
Message-ID:  <451DA83C.4050808@cisco.com>
In-Reply-To: <20060929231007.GS80527@funkthat.com>
References:  <451C4850.5030302@freebsd.org> <Pine.BSF.4.58.0609281928020.20971@niwun.pair.com> <451D884F.1030807@cisco.com> <20060929213722.GR80527@funkthat.com> <451D973C.8070004@freebsd.org> <20060929231007.GS80527@funkthat.com>

next in thread | previous in thread | raw e-mail | index | archive | help
John-Mark Gurney wrote:
> Andre Oppermann wrote this message on Fri, Sep 29, 2006 at 23:59 +0200:
> 
>>>w/ 512 byte mbuf and a 2k cluster just to store just 1514 bytes of data,
>>>that's only 60% effeciency wrt to memory usage...  so, we currently
>>>waste 40% of memory allocated to mbufs+clusters...  Even reducing
>>>mbufs back to 128 or 256 would be a big help, though IPSEC I believe
>>>would have issues...
>>
>>mbufs are 256 bytes.
> 
> 
> Hmmm.. I keep getting this confused... maybe because there was discussion
> about increasing this a few years back...  or maybe because NOTES has
> it as 512.. :)
> 
> 
>>>Hmmm.. If we switched clusters to 1536 bytes in size, we'd be able to
>>>fit 8 in 12k (though I guess for 8k page boxes we'd do 16 in 24k)...  The
>>>only issue w/ that would be that a few of the clusters would possibly
>>>split page boundaries...  How much this would effect performance would
>>>be an interesting question to answer...
>>
>>Splitting page boundaries is not an option as it may not be physically
>>contigous.
> 
> 
> unless we do something strange like allocate them contigously...  though
> that introduces another set of issues....
> 
> 
>>Just don't overengineer the stuff.  Mbufs are only used temporarily and
>>a bit theoretical waste is not much a problem (so far at least).
> 
> 
> Well, I beg to differ... most gige cards grab mbuf+cluster for every
> single ring buffer they have.. which is usually 512... so every gige
> interface for the most part consumes 1meg of memory that is not
> reusable...  because if we run out of mbuf+clusters to replace in the
> receive ring, we will not tap into the 1meg of mbuf+clusters available
> to us...  so, if you have a quad gige, that's 4megs wasted, plus w/ the
> fact that we could only use ~65% of that memory, that's a lot of memory
> wasted...
> 
> Yeh, everyone says you have gigs of memory, but do we really want to
> be known as the wasteful OS?
> 

Let me try to find some cycles (somewhere) and play with this :-)

R


-- 
Randall Stewart
NSSTG - Cisco Systems Inc.
803-345-0369 <or> 815-342-5222 (cell)



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?451DA83C.4050808>