[Fwd: low-latency patches]

[Fwd: low-latency patches]

[J Sloan]

[-- Attachment #1: Type: text/plain, Size: 1 bytes --]



[-- Attachment #2: Type: message/rfc822, Size: 1997 bytes --]

From: J Sloan <jjs@pobox.com>
To: Bob McElrath <mcelrath+linux@draal.physics.wisc.edu>
Subject: Re: low-latency patches
Date: Fri, 05 Oct 2001 23:51:55 -0700
Message-ID: <3BBEAA0B.C990D2E7@pobox.com>

Bob McElrath wrote:

> It seems there are two low-latency projects out there.  The one by Robert Love:
>     http://tech9.net/rml/linux/
> and the original one:
>     http://www.uow.edu.au/~andrewm/linux/schedlat.html
>
> Correct me if I'm wrong, but the former uses spinlocks to know when it can
> preempt the kernel, and the latter just tries to reduce latency by adding
> (un)conditional_schedule and placing it at key places in the kernel?
>
> My questions are:
> 1) Which of these two projects has better latency performance?  Has anyone
>     benchmarked them against each other?
> 2) Will either of these ever be merged into Linus' kernel (2.5?)
> 3) Is there a possibility that either of these will make it to non-x86
>     platforms?  (for me: alpha)  The second patch looks like it would
>     straightforwardly work on any arch, but the config.in for it is only in
>     arch/i386.  Robert Love's patches would need some arch-specific asm...

In my experience with them, the Andrew Morton patches
provide a "smoother" interactive feel, great for things like
online gaming (quake 3 arena, etc), however the Robert
Love patches are simpler, seem less intrusive, and I've
had better luck with them on smp, highmem boxes.

(just IMHO) I like Andrew's patches on (up) workstations,
and Robert's on (smp) servers, with some grey area of
overlap -

I'm hardly the person to say, but the rml patches would
seem more likely to go in sooner, if at all.  I'd love to see
both remain an option.

cu

jjs

Re: [Fwd: low-latency patches]

[george anzinger]

J Sloan wrote:
> 
>   ------------------------------------------------------------------------
> 
> Subject: Re: low-latency patches
> Date: Fri, 05 Oct 2001 23:51:55 -0700
> From: J Sloan <jjs@pobox.com>
> Organization: J S Concepts
> To: Bob McElrath <mcelrath+linux@draal.physics.wisc.edu>
> References: <20011006010519.A749@draal.physics.wisc.edu>
> 
> Bob McElrath wrote:
> 
> > It seems there are two low-latency projects out there.  The one by Robert Love:
> >     http://tech9.net/rml/linux/
> > and the original one:
> >     http://www.uow.edu.au/~andrewm/linux/schedlat.html
> >
> > Correct me if I'm wrong, but the former uses spinlocks to know when it can
> > preempt the kernel, and the latter just tries to reduce latency by adding
> > (un)conditional_schedule and placing it at key places in the kernel?
> >
> > My questions are:
> > 1) Which of these two projects has better latency performance?  Has anyone
> >     benchmarked them against each other?
> > 2) Will either of these ever be merged into Linus' kernel (2.5?)
> > 3) Is there a possibility that either of these will make it to non-x86
> >     platforms?  (for me: alpha)  The second patch looks like it would
> >     straightforwardly work on any arch, but the config.in for it is only in
> >     arch/i386.  Robert Love's patches would need some arch-specific asm...
> 
> In my experience with them, the Andrew Morton patches
> provide a "smoother" interactive feel, great for things like
> online gaming (quake 3 arena, etc), however the Robert
> Love patches are simpler, seem less intrusive, and I've
> had better luck with them on smp, highmem boxes.
> 
> (just IMHO) I like Andrew's patches on (up) workstations,
> and Robert's on (smp) servers, with some grey area of
> overlap -
> 
> I'm hardly the person to say, but the rml patches would
> seem more likely to go in sooner, if at all.  I'd love to see
> both remain an option.
> 
The two patches are NOT mutually exclusive.  Both can be used and are in
some cases.

MontaVista is actively working on porting the patch you refer to as
Robert Love's patch to most of the other archs.  Which arch are you
interested in?

George

Re: [Fwd: low-latency patches]

[Bob McElrath]

[-- Attachment #1: Type: text/plain, Size: 471 bytes --]

george anzinger [george@mvista.com] wrote:
> The two patches are NOT mutually exclusive.  Both can be used and are in
> some cases.
> 
> MontaVista is actively working on porting the patch you refer to as
> Robert Love's patch to most of the other archs.  Which arch are you
> interested in?

alpha.  ;)  I'd be happy to test early patches.

Cheers,
-- Bob

Bob McElrath (rsmcelrath@students.wisc.edu) 
Univ. of Wisconsin at Madison, Department of Physics

[-- Attachment #2: Type: application/pgp-signature, Size: 240 bytes --]

Re: [Fwd: low-latency patches]

[george anzinger]

Bob McElrath wrote:
> 
> george anzinger [george@mvista.com] wrote:
> > The two patches are NOT mutually exclusive.  Both can be used and are in
> > some cases.
> >
> > MontaVista is actively working on porting the patch you refer to as
> > Robert Love's patch to most of the other archs.  Which arch are you
> > interested in?
> 
> alpha.  ;)  I'd be happy to test early patches.
> 
I am checking.  Meanwhile, do you know alpha asm?  The changes are
relatively simple.

George

Re: [Fwd: low-latency patches]

[Bob McElrath]

[-- Attachment #1: Type: text/plain, Size: 972 bytes --]

george anzinger [george@mvista.com] wrote:
> Bob McElrath wrote:
> > 
> > george anzinger [george@mvista.com] wrote:
> > > The two patches are NOT mutually exclusive.  Both can be used and are in
> > > some cases.
> > >
> > > MontaVista is actively working on porting the patch you refer to as
> > > Robert Love's patch to most of the other archs.  Which arch are you
> > > interested in?
> > 
> > alpha.  ;)  I'd be happy to test early patches.
> > 
> I am checking.  Meanwhile, do you know alpha asm?  The changes are
> relatively simple.

No, but I could probably figure it out (I know x86 asm, so might be able to
convert the routines).  I looked at it briefly and decided it would take me
more than an hour to add alpha asm, so didn't do it immediately.  ;)

If I hack something together (and it works), I'll send it to you.

Cheers,
-- Bob

Bob McElrath (rsmcelrath@students.wisc.edu) 
Univ. of Wisconsin at Madison, Department of Physics

[-- Attachment #2: Type: application/pgp-signature, Size: 240 bytes --]