bus PCI limited?

[Robert Morris]

Alessandro,

Have a look at the graphs at the end of this paper:

  http://www.pdos.lcs.mit.edu/papers/click:asplos02.pdf

They suggest that a 32bit/33mHz PCI bus would limit you to just over
400,000 packets per second. They also suggest that an 800 mHz Pentium
III CPU limits you to just under 400,000 packets/second. I suspect
that a 1.6 gHz P4 is about the same speed, so you're likely to need
both a faster CPU and a faster bus if you want to forward faster.

Robert

> From: "Alessandro" <bujia at libero.it>
> To: <click at amsterdam.lcs.mit.edu>
> Subject: bus PCI limited?
> Date: Sat, 15 Mar 2003 10:56:53 +0100
> 
> Messaggio in formato MIME composto da più parti.
> 
> ------=_NextPart_000_001B_01C2EAE1.96D248E0
> Content-Type: text/plain;
> 	charset="iso-8859-1"
> Content-Transfer-Encoding: quoted-printable
> 
> Hi, can help me???
> 
> I've done test with Click in a Pentium 4 1600MHz with 32bit/33Mhz bus PCI, =
> and 2 INTEL PRO1000 XT server adapter. max forwarding rate result 370000 64=
>  byte packet per second (without optimization, I made some test and don't c=
> hange this rate...). It's a good forwarding rate? What's the bottleneck? Is=
>  PCI bus?
> 
> Thanks,
> 
> Alessandro
> 
> 
> ------=_NextPart_000_001B_01C2EAE1.96D248E0
> Content-type: text/plain; charset="iso-8859-1"
> Content-Transfer-Encoding: quoted-printable
> 
> Hi, can help me???
> =A0
> I've done test with Click in a Pentium 4 1600MHz with 32bit/33Mhz bus PCI, =
> and 2 INTEL PRO1000 XT server adapter. max forwarding rate result 370000 64=
>  byte packet per second (without optimization, I made some test and don't c=
> hange this rate...). It's a good forwarding rate? What's the bottleneck? Is=
>  PCI bus?
> =A0
> Thanks,
> =A0
> Alessandro
> =A0
> 
> ------=_NextPart_000_001B_01C2EAE1.96D248E0--
>