On Fri, 27 Apr 2001, Mike Galbraith wrote:
> virgin pre7 +Rik
> real11m44.088s
> user7m57.720s
> sys 0m36.420s
> None of them make much difference.
Good, then I suppose we can put in the cleanup from my code, since
it makes the balancing a bit more predictable and should keep the
On Thu, 26 Apr 2001, Linus Torvalds wrote:
> Have you looked at "free_pte()"? I don't like that function, and it might
> make a difference. There are several small nits with it:
snip
> I _think_ the logic should be something along the lines of: "freeing the
> page amounts to a implied
On Fri, 27 Apr 2001, Marcelo Tosatti wrote:
> On Fri, 27 Apr 2001, Mike Galbraith wrote:
>
> > I decided to take a break from pondering input and see why the thing
> > ran itself into the ground. Methinks I was sent the wrooong patch :)
>
> Mike,
>
> Please apply this patch on top of Rik's v2
On Fri, 27 Apr 2001, Mike Galbraith wrote:
> On Thu, 26 Apr 2001, Rik van Riel wrote:
>
> > On Thu, 26 Apr 2001, Mike Galbraith wrote:
> >
> > > > > > No. It livelocked on me with almost all active pages exausted.
> > > > > Misspoke.. I didn't try the two mixed. Rik's patch livelocked me.
>
On Thu, 26 Apr 2001, Rik van Riel wrote:
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
>
> > > > > No. It livelocked on me with almost all active pages exausted.
> > > > Misspoke.. I didn't try the two mixed. Rik's patch livelocked me.
> > >
> > > Interesting. The semantics of my patch are
On Thu, 26 Apr 2001, Rik van Riel wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
No. It livelocked on me with almost all active pages exausted.
Misspoke.. I didn't try the two mixed. Rik's patch livelocked me.
Interesting. The semantics of my patch are practically the same
On Fri, 27 Apr 2001, Mike Galbraith wrote:
On Thu, 26 Apr 2001, Rik van Riel wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
No. It livelocked on me with almost all active pages exausted.
Misspoke.. I didn't try the two mixed. Rik's patch livelocked me.
On Thu, 26 Apr 2001, Linus Torvalds wrote:
Have you looked at free_pte()? I don't like that function, and it might
make a difference. There are several small nits with it:
snip
I _think_ the logic should be something along the lines of: freeing the
page amounts to a implied down-aging of
On Fri, 27 Apr 2001, Mike Galbraith wrote:
virgin pre7 +Rik
real11m44.088s
user7m57.720s
sys 0m36.420s
None of them make much difference.
Good, then I suppose we can put in the cleanup from my code, since
it makes the balancing a bit more predictable and should keep the
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> > > > No. It livelocked on me with almost all active pages exausted.
> > > Misspoke.. I didn't try the two mixed. Rik's patch livelocked me.
> >
> > Interesting. The semantics of my patch are practically the same as
> > those of the stock kernel ...
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> On Thu, 26 Apr 2001, Rik van Riel wrote:
>
> > On Thu, 26 Apr 2001, Mike Galbraith wrote:
> >
> > > 1. pagecache is becoming swapcache and must be aged before anything is
> > > done. Meanwhile we're calling refill_inactive_scan() so fast that noone
>
On Thu, 26 Apr 2001, Rik van Riel wrote:
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
> > On Thu, 26 Apr 2001, Mike Galbraith wrote:
> >
> > > > limit the runtime of refill_inactive_scan(). This is similar to Rik's
> > > > reclaim-limit+aging-tuning patch to linux-mm yesterday. could you try
> >
On Thu, 26 Apr 2001, Rik van Riel wrote:
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
>
> > 1. pagecache is becoming swapcache and must be aged before anything is
> > done. Meanwhile we're calling refill_inactive_scan() so fast that noone
> > has a chance to touch a page. Age becomes a simple
On Thu, 26 Apr 2001, Linus Torvalds wrote:
>
> On the other hand, to offset some of these, we actually count the page
> accessed _twice_ sometimes: we count it on lookup, and we count it when we
> see the accessed bit in vmscan.c. Which results in some pages getting aged
> up twice for just one
On Thu, 26 Apr 2001, Linus Torvalds wrote:
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
> >
> > 2.4.4.pre7.virgin
> > real11m33.589s
> > user7m57.790s
> > sys 0m38.730s
> >
> > 2.4.4.pre7.sillyness
> > real9m30.336s
> > user7m55.270s
> > sys 0m38.510s
>
> Well, I actually
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
>
> > > limit the runtime of refill_inactive_scan(). This is similar to Rik's
> > > reclaim-limit+aging-tuning patch to linux-mm yesterday. could you try
> > > Rik's patch with your patch except this jiffies
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> 1. pagecache is becoming swapcache and must be aged before anything is
> done. Meanwhile we're calling refill_inactive_scan() so fast that noone
> has a chance to touch a page. Age becomes a simple counter.. I think.
> When you hit a big surge,
On Thu, 26 Apr 2001, Mike Galbraith wrote:
>
> 2.4.4.pre7.virgin
> real11m33.589s
> user7m57.790s
> sys 0m38.730s
>
> 2.4.4.pre7.sillyness
> real9m30.336s
> user7m55.270s
> sys 0m38.510s
Well, I actually like parts of this. The "always swap out current mm" one
looks
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
>
> > > (i cannot see how this chunk affects the VM, AFAICS this too makes the
> > > zapping of the cache less agressive.)
> >
> > (more folks get snagged on write.. they can't eat cache so fast)
>
> What
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> > (i cannot see how this chunk affects the VM, AFAICS this too makes the
> > zapping of the cache less agressive.)
>
> (more folks get snagged on write.. they can't eat cache so fast)
What about GFP_BUFFER allocations ? :)
I suspect the jiffies
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> > limit the runtime of refill_inactive_scan(). This is similar to Rik's
> > reclaim-limit+aging-tuning patch to linux-mm yesterday. could you try
> > Rik's patch with your patch except this jiffies hack, does it still
> > achieve the same improvement?
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
> Have you tried to tune SWAP_SHIFT and the priority used inside swap_out()
> to see if you can make pte deactivation less aggressive ?
Many many many times.. no dice.
(more agressive is much better for surge regulation.. power brakes!)
On Thu, 26 Apr 2001, Ingo Molnar wrote:
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
>
> > 2.4.4.pre7.virgin
> > real11m33.589s
>
> > 2.4.4.pre7.sillyness
> > real9m30.336s
>
> very interesting. Looks like there are still reserves in the VM, for heavy
> workloads. (and swapping is all
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> 2.4.4.pre7.virgin
> real11m33.589s
> 2.4.4.pre7.sillyness
> real9m30.336s
very interesting. Looks like there are still reserves in the VM, for heavy
workloads. (and swapping is all about heavy workloads.)
it would be interesting to see why
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
>
> > > (I can get it to under 9 with MUCH extremely ugly tinkering. I've done
> > > this enough to know that I _should_ be able to do 8 1/2 minutes ~easily)
> >
> > Which kind of changes you're doing to
On Thu, 26 Apr 2001, Ingo Molnar wrote:
>
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
>
> > More of a question. Neither Ingo's nor your patch makes any
> > difference on my UP box (128mb PIII/500) doing make -j30. [...]
>
> (the patch Marcelo sent is the -B3 patch plus Linus' suggested async
>
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
> > (I can get it to under 9 with MUCH extremely ugly tinkering. I've done
> > this enough to know that I _should_ be able to do 8 1/2 minutes ~easily)
>
> Which kind of changes you're doing to get better performance on this test?
:)
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> More of a question. Neither Ingo's nor your patch makes any
> difference on my UP box (128mb PIII/500) doing make -j30. [...]
(the patch Marcelo sent is the -B3 patch plus Linus' suggested async
interface cleanup, so it should be functionally
On Thu, 26 Apr 2001, Linus Torvalds wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
2.4.4.pre7.virgin
real11m33.589s
user7m57.790s
sys 0m38.730s
2.4.4.pre7.sillyness
real9m30.336s
user7m55.270s
sys 0m38.510s
Well, I actually like parts of this.
On Thu, 26 Apr 2001, Linus Torvalds wrote:
On the other hand, to offset some of these, we actually count the page
accessed _twice_ sometimes: we count it on lookup, and we count it when we
see the accessed bit in vmscan.c. Which results in some pages getting aged
up twice for just one
On Thu, 26 Apr 2001, Rik van Riel wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
1. pagecache is becoming swapcache and must be aged before anything is
done. Meanwhile we're calling refill_inactive_scan() so fast that noone
has a chance to touch a page. Age becomes a simple
On Thu, 26 Apr 2001, Rik van Riel wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
limit the runtime of refill_inactive_scan(). This is similar to Rik's
reclaim-limit+aging-tuning patch to linux-mm yesterday. could you try
Rik's patch
On Thu, 26 Apr 2001, Mike Galbraith wrote:
No. It livelocked on me with almost all active pages exausted.
Misspoke.. I didn't try the two mixed. Rik's patch livelocked me.
Interesting. The semantics of my patch are practically the same as
those of the stock kernel ... can you get
On Thu, 26 Apr 2001, Mike Galbraith wrote:
More of a question. Neither Ingo's nor your patch makes any
difference on my UP box (128mb PIII/500) doing make -j30. [...]
(the patch Marcelo sent is the -B3 patch plus Linus' suggested async
interface cleanup, so it should be functionally
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
(I can get it to under 9 with MUCH extremely ugly tinkering. I've done
this enough to know that I _should_ be able to do 8 1/2 minutes ~easily)
Which kind of changes you're doing to get better performance on this test?
:)
2.4.4.pre7.virgin
On Thu, 26 Apr 2001, Ingo Molnar wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
More of a question. Neither Ingo's nor your patch makes any
difference on my UP box (128mb PIII/500) doing make -j30. [...]
(the patch Marcelo sent is the -B3 patch plus Linus' suggested async
interface
On Thu, 26 Apr 2001, Mike Galbraith wrote:
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
(I can get it to under 9 with MUCH extremely ugly tinkering. I've done
this enough to know that I _should_ be able to do 8 1/2 minutes ~easily)
Which kind of changes you're doing to get better
On Thu, 26 Apr 2001, Mike Galbraith wrote:
2.4.4.pre7.virgin
real11m33.589s
2.4.4.pre7.sillyness
real9m30.336s
very interesting. Looks like there are still reserves in the VM, for heavy
workloads. (and swapping is all about heavy workloads.)
it would be interesting to see why
On Thu, 26 Apr 2001, Ingo Molnar wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
2.4.4.pre7.virgin
real11m33.589s
2.4.4.pre7.sillyness
real9m30.336s
very interesting. Looks like there are still reserves in the VM, for heavy
workloads. (and swapping is all about heavy
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
Have you tried to tune SWAP_SHIFT and the priority used inside swap_out()
to see if you can make pte deactivation less aggressive ?
Many many many times.. no dice.
(more agressive is much better for surge regulation.. power brakes!)
-Mike
On Thu, 26 Apr 2001, Mike Galbraith wrote:
limit the runtime of refill_inactive_scan(). This is similar to Rik's
reclaim-limit+aging-tuning patch to linux-mm yesterday. could you try
Rik's patch with your patch except this jiffies hack, does it still
achieve the same improvement?
No.
On Thu, 26 Apr 2001, Mike Galbraith wrote:
(i cannot see how this chunk affects the VM, AFAICS this too makes the
zapping of the cache less agressive.)
(more folks get snagged on write.. they can't eat cache so fast)
What about GFP_BUFFER allocations ? :)
I suspect the jiffies hack is
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
(i cannot see how this chunk affects the VM, AFAICS this too makes the
zapping of the cache less agressive.)
(more folks get snagged on write.. they can't eat cache so fast)
What about GFP_BUFFER
On Thu, 26 Apr 2001, Mike Galbraith wrote:
2.4.4.pre7.virgin
real11m33.589s
user7m57.790s
sys 0m38.730s
2.4.4.pre7.sillyness
real9m30.336s
user7m55.270s
sys 0m38.510s
Well, I actually like parts of this. The always swap out current mm one
looks rather
On Thu, 26 Apr 2001, Mike Galbraith wrote:
1. pagecache is becoming swapcache and must be aged before anything is
done. Meanwhile we're calling refill_inactive_scan() so fast that noone
has a chance to touch a page. Age becomes a simple counter.. I think.
When you hit a big surge, swap
On Thu, 26 Apr 2001, Mike Galbraith wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
limit the runtime of refill_inactive_scan(). This is similar to Rik's
reclaim-limit+aging-tuning patch to linux-mm yesterday. could you try
Rik's patch with your patch except this jiffies hack, does
On Thu, 26 Apr 2001, Mike Galbraith wrote:
On Thu, 26 Apr 2001, Rik van Riel wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
1. pagecache is becoming swapcache and must be aged before anything is
done. Meanwhile we're calling refill_inactive_scan() so fast that noone
has a
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
> On Thu, 26 Apr 2001, Mike Galbraith wrote:
>
> > > Comments?
> >
> > More of a question. Neither Ingo's nor your patch makes any difference
> > on my UP box (128mb PIII/500) doing make -j30.
>
> Well, my patch incorporates Ingo's patch.
>
> It is
On Thu, 26 Apr 2001, Mike Galbraith wrote:
> > Comments?
>
> More of a question. Neither Ingo's nor your patch makes any difference
> on my UP box (128mb PIII/500) doing make -j30.
Well, my patch incorporates Ingo's patch.
It is now integrated into pre7, btw.
> It is taking me 11 1/2
>
On Wed, 25 Apr 2001, Marcelo Tosatti wrote:
> On Tue, 24 Apr 2001, Linus Torvalds wrote:
>
> > Basically, I don't want to mix synchronous and asynchronous
> > interfaces. Everything should be asynchronous by default, and waiting
> > should be explicit.
>
> The following patch changes all swap IO
;[EMAIL PROTECTED]>, [EMAIL PROTECTED],
Rik van Riel <[EMAIL PROTECTED]>,
Szabolcs Szakacsits <[EMAIL PROTECTED]>
Subject: Re: [patch] swap-speedup-2.4.3-B3
On Tue, 24 Apr 2001, Linus Torvalds wrote:
> Basically, I don't want to mix synchronous and asynchronous
&g
],
Rik van Riel [EMAIL PROTECTED],
Szabolcs Szakacsits [EMAIL PROTECTED]
Subject: Re: [patch] swap-speedup-2.4.3-B3
On Tue, 24 Apr 2001, Linus Torvalds wrote:
Basically, I don't want to mix synchronous and asynchronous
interfaces. Everything should be asynchronous by default
On Wed, 25 Apr 2001, Marcelo Tosatti wrote:
On Tue, 24 Apr 2001, Linus Torvalds wrote:
Basically, I don't want to mix synchronous and asynchronous
interfaces. Everything should be asynchronous by default, and waiting
should be explicit.
The following patch changes all swap IO functions
On Thu, 26 Apr 2001, Mike Galbraith wrote:
Comments?
More of a question. Neither Ingo's nor your patch makes any difference
on my UP box (128mb PIII/500) doing make -j30.
Well, my patch incorporates Ingo's patch.
It is now integrated into pre7, btw.
It is taking me 11 1/2
On Thu, 26 Apr 2001, Marcelo Tosatti wrote:
On Thu, 26 Apr 2001, Mike Galbraith wrote:
Comments?
More of a question. Neither Ingo's nor your patch makes any difference
on my UP box (128mb PIII/500) doing make -j30.
Well, my patch incorporates Ingo's patch.
It is now integrated
On Tue, 24 Apr 2001, Linus Torvalds wrote:
> Basically, I don't want to mix synchronous and asynchronous
> interfaces. Everything should be asynchronous by default, and waiting
> should be explicit.
The following patch changes all swap IO functions to be asynchronous by
default and changes
On Tue, 24 Apr 2001, Ingo Molnar wrote:
>
> the latest swap-speedup patch can be found at:
Please don't add more of those horrible "wait" arguments.
Make two different versions of a function instead. It's going to clean up
and simplify the code, and there really isn't any reason to do what
the latest swap-speedup patch can be found at:
http://people.redhat.com/mingo/swap-speedup/swap-speedup-2.4.3-B3
(the patch is against 2.4.4-pre6 or 2.4.3-ac13.)
-B3 includes Marcelo's patch for another area that blocks unnecesserily on
locked swapcache pages: async swapcache readahead.
the latest swap-speedup patch can be found at:
http://people.redhat.com/mingo/swap-speedup/swap-speedup-2.4.3-B3
(the patch is against 2.4.4-pre6 or 2.4.3-ac13.)
-B3 includes Marcelo's patch for another area that blocks unnecesserily on
locked swapcache pages: async swapcache readahead.
On Tue, 24 Apr 2001, Ingo Molnar wrote:
the latest swap-speedup patch can be found at:
Please don't add more of those horrible wait arguments.
Make two different versions of a function instead. It's going to clean up
and simplify the code, and there really isn't any reason to do what you're
60 matches
Mail list logo