netdev
[Top] [All Lists]

Re: [Patch 3/3 2.4] e1000: Update Documentation/networking/e1000.txt

To: Jason Lunz <lunz@xxxxxxxxxxxx>
Subject: Re: [Patch 3/3 2.4] e1000: Update Documentation/networking/e1000.txt
From: akepner@xxxxxxx
Date: Fri, 15 Oct 2004 12:20:48 -0700 (PDT)
Cc: <netdev@xxxxxxxxxxx>
In-reply-to: <ckp6tr$nrg$1@xxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
On Fri, 15 Oct 2004, Jason Lunz wrote:

> ganesh.venkatesan@xxxxxxxxx said:
> > +          This sets the InterruptThrottleRate to 3000 interrupts/sec for 
> > the 
> > +          first, second, and third instances of the driver. The range of 
> > 2000 to 
> > +          3000 interrupts per second works on a majority of systems and is 
> > a 
> > +          good starting point, but the optimal value will be 
> > platform-specific. 
> > +          If CPU utilization is not a concern, use RX_POLLING (NAPI) and 
> > default 
> > +          driver settings.
> 
> I don't understand this bit. Isn't the point of NAPI to disable rx
> interrupts entirely, and only gather as many frames from the driver as
> CPU time permits? That's supposed to limit network-rx CPU usage, not
> increase it as the above paragraph implies.

In my experience (on Altix systems, so that experience may be 
somewhat atypical), NAPI results in few (2-7) packets processed 
per interrupt, so the interrupt rate (and CPU utilization) can be 
very high. Interrupt coalescence is much better in both of these 
respects.

-- 
Arthur




<Prev in Thread] Current Thread [Next in Thread>