mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* CONFIG_TULIP_NAPI_HW_MITIGATION latency question.
@ 2004-11-23 17:43 Justin Piszcz
  2004-11-25 16:12 ` Robert Olsson
  0 siblings, 1 reply; 3+ messages in thread
From: Justin Piszcz @ 2004-11-23 17:43 UTC (permalink / raw)
  To: linux-kernel

"at the cost of a small latency"

How much additional latency is added with this option enabled?

5ms? 10ms?

-> Use Interrupt Mitigation
   x CONFIG_TULIP_NAPI_HW_MITIGATION:
   x Use HW to reduce RX interrupts. Not strict necessary since NAPI 
reduces RX interrupts but itself. Although this reduces RX interrupts even at 
low levels traffic at the cost of a small latency. 
x
   x If in doubt, say Y. 
x

^ permalink raw reply	[flat|nested] 3+ messages in thread

* CONFIG_TULIP_NAPI_HW_MITIGATION latency question.
  2004-11-23 17:43 CONFIG_TULIP_NAPI_HW_MITIGATION latency question Justin Piszcz
@ 2004-11-25 16:12 ` Robert Olsson
  2004-11-26 12:07   ` Justin Piszcz
  0 siblings, 1 reply; 3+ messages in thread
From: Robert Olsson @ 2004-11-25 16:12 UTC (permalink / raw)
  To: Justin Piszcz; +Cc: linux-kernel, Robert.Olsson




Justin Piszcz writes:
 > "at the cost of a small latency"
 > 
 > How much additional latency is added with this option enabled?
 > 
 > 5ms? 10ms?
 > 
 > -> Use Interrupt Mitigation
 >    x CONFIG_TULIP_NAPI_HW_MITIGATION:
 >    x Use HW to reduce RX interrupts. Not strict necessary since NAPI 
 > reduces RX interrupts but itself. Although this reduces RX interrupts even at 
 > low levels traffic at the cost of a small latency. 
 > x
 >    x If in doubt, say Y. 

 Hello!

 It's in usec and the tulip only enables this latency when it's needed it
 really tries to keep latency down. 

 It has three steeps of input control.

 Low rate:     No driver latency. No interrupt delay at all.
 Medium rate:  Samll driver latency.
 High rate:    Polling via ->poll

						--ro

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: CONFIG_TULIP_NAPI_HW_MITIGATION latency question.
  2004-11-25 16:12 ` Robert Olsson
@ 2004-11-26 12:07   ` Justin Piszcz
  0 siblings, 0 replies; 3+ messages in thread
From: Justin Piszcz @ 2004-11-26 12:07 UTC (permalink / raw)
  To: Robert Olsson; +Cc: linux-kernel

Thanks for the info.

On Thu, 25 Nov 2004, Robert Olsson wrote:

>
>
>
> Justin Piszcz writes:
> > "at the cost of a small latency"
> >
> > How much additional latency is added with this option enabled?
> >
> > 5ms? 10ms?
> >
> > -> Use Interrupt Mitigation
> >    x CONFIG_TULIP_NAPI_HW_MITIGATION:
> >    x Use HW to reduce RX interrupts. Not strict necessary since NAPI
> > reduces RX interrupts but itself. Although this reduces RX interrupts even at
> > low levels traffic at the cost of a small latency.
> > x
> >    x If in doubt, say Y.
>
> Hello!
>
> It's in usec and the tulip only enables this latency when it's needed it
> really tries to keep latency down.
>
> It has three steeps of input control.
>
> Low rate:     No driver latency. No interrupt delay at all.
> Medium rate:  Samll driver latency.
> High rate:    Polling via ->poll
>
> 						--ro
>

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2004-11-27  4:18 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-11-23 17:43 CONFIG_TULIP_NAPI_HW_MITIGATION latency question Justin Piszcz
2004-11-25 16:12 ` Robert Olsson
2004-11-26 12:07   ` Justin Piszcz

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®