DSDT Change (Compaq Presario 2100)

DSDT Change (Compaq Presario 2100)

[Andre Eisenbach]

[-- Attachment #1: Type: text/plain, Size: 1001 bytes --]

Hey!

Sorry for semi spamming the list tonight - but I'm finally getting places :)

As previously mentioned, my Compaq as well as many other HP model notebooks 
slow down considerably when running on battery. Short off "ACPI=off" there 
seemed to be no workaround.

Well, tonight, while recompiling the DSDT a few times to remove warnings and 
errors, I stumbled across this (in method SMSL

If (0x01)
{
	Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
	Store ("AC remove! Set Throttle Duty Cycle ---: ", Debug)
	Store (TRDC, Debug)
	Store (0x01, TREN)
}

I am assuming that the "If (0x01)" was put there by a developer to disable 
this throttle duty cycle mechanism easily for testing, or maybe for different 
OEMs.

In any case, I changed "If (0x01)" to "If (0x00)" and voila, I am typing this 
email from KDE with the power cord unplugged - no slowdown at all.

Now my questions:

1) Does this sound sane?
2) Should I submit the updated DSDT to acpi.sf.net?

Cheers,
   André

[-- Attachment #2: signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: DSDT Change (Compaq Presario 2100)

[Erik Meitner]

> As previously mentioned, my Compaq as well as many other HP model notebooks 
> slow down considerably when running on battery. Short off "ACPI=off" there 
> seemed to be no workaround.

My HP nx9005 is one of them. It runs terribly slow when on battery, it 
is unusable really.

> Well, tonight, while recompiling the DSDT a few times to remove warnings and 
> errors, I stumbled across this (in method SMSL
> 
> If (0x01)
> {
> 	Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
> 	Store ("AC remove! Set Throttle Duty Cycle ---: ", Debug)
> 	Store (TRDC, Debug)
> 	Store (0x01, TREN)
> }
> 

The DSDT for my laptop has the exact same code.

> In any case, I changed "If (0x01)" to "If (0x00)" and voila, I am typing this 
> email from KDE with the power cord unplugged - no slowdown at all.

I have verified that this works for my machine also.

Am I correct in assuming that TREN is "Throttle Enable"(boolean value) 
and TRDC is "Throttle Duty-cycle"? What other TRDC values might make 
sense other than 0x0 ?


Erik Meitner
emeitner.f2o.org/nx9005




-------------------------------------------------------
This SF.Net email is sponsored by: Oracle 10g
Get certified on the hottest thing ever to hit the market... Oracle 10g. 
Take an Oracle 10g class now, and we'll give you the exam FREE.
http://ads.osdn.com/?ad_id=3149&alloc_id=8166&op=click

Re: Re: DSDT Change (Compaq Presario 2100)

[Andre Eisenbach]

[-- Attachment #1: Type: text/plain, Size: 1831 bytes --]

Am Samstag Mai 29 2004 11:53 schrieb Erik Meitner:
> > Well, tonight, while recompiling the DSDT a few times to remove warnings
> > and errors, I stumbled across this (in method SMSL
> >
> > If (0x01)
> > {
> > 	Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
> > 	Store ("AC remove! Set Throttle Duty Cycle ---: ", Debug)
> > 	Store (TRDC, Debug)
> > 	Store (0x01, TREN)
> > }
>
> The DSDT for my laptop has the exact same code.

That piece of code is the state cange code (i.e. when you unplug the AC cord). 
There is another place you need to change, which initializes the behaviour 
when you boot with the power cable already unplugged.

The method is _REG. Here is the code

                                Store (0x01, \_SB.PCI0.PMU.CRUN)
                                If (0x01)
                                {
                                    Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
                                    Store (0x01, TREN)
                                }

Once again change "If (0x01)" to "If (0x00)".

> > In any case, I changed "If (0x01)" to "If (0x00)" and voila, I am typing
> > this email from KDE with the power cord unplugged - no slowdown at all.
>
> I have verified that this works for my machine also.

Thats great news!

> Am I correct in assuming that TREN is "Throttle Enable"(boolean value)
> and TRDC is "Throttle Duty-cycle"? What other TRDC values might make
> sense other than 0x0 ?

I think you're right about TREN and TRDC. 
That method of throttling the CPU is not a good method apparently. It also 
doesn't save battery (at least for me). Instead, use "PowerNOW" if you have 
an AMD chip. It's built into the 2.6+ kernels (cpufreq) and works like a 
champ for me.  And unlike the DSDT based throtteling, you can control it!

Cheers,
    André

[-- Attachment #2: signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: DSDT Change (Compaq Presario 2100)

[Andrew D. Keyser]

Although, if you update the BIOS, the problem seems to fix itself...
Anywho, I find it ironic that HP touts "AMD with PowerNOW" yet they 
don't actually use the powernow and instead use the crappy DSDT.

Andrew

Andre Eisenbach wrote:
> Am Samstag Mai 29 2004 11:53 schrieb Erik Meitner:
> 
>>>Well, tonight, while recompiling the DSDT a few times to remove warnings
>>>and errors, I stumbled across this (in method SMSL
>>>
>>>If (0x01)
>>>{
>>>	Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
>>>	Store ("AC remove! Set Throttle Duty Cycle ---: ", Debug)
>>>	Store (TRDC, Debug)
>>>	Store (0x01, TREN)
>>>}
>>
>>The DSDT for my laptop has the exact same code.
> 
> 
> That piece of code is the state cange code (i.e. when you unplug the AC cord). 
> There is another place you need to change, which initializes the behaviour 
> when you boot with the power cable already unplugged.
> 
> The method is _REG. Here is the code
> 
>                                 Store (0x01, \_SB.PCI0.PMU.CRUN)
>                                 If (0x01)
>                                 {
>                                     Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
>                                     Store (0x01, TREN)
>                                 }
> 
> Once again change "If (0x01)" to "If (0x00)".
> 
> 
>>>In any case, I changed "If (0x01)" to "If (0x00)" and voila, I am typing
>>>this email from KDE with the power cord unplugged - no slowdown at all.
>>
>>I have verified that this works for my machine also.
> 
> 
> Thats great news!
> 
> 
>>Am I correct in assuming that TREN is "Throttle Enable"(boolean value)
>>and TRDC is "Throttle Duty-cycle"? What other TRDC values might make
>>sense other than 0x0 ?
> 
> 
> I think you're right about TREN and TRDC. 
> That method of throttling the CPU is not a good method apparently. It also 
> doesn't save battery (at least for me). Instead, use "PowerNOW" if you have 
> an AMD chip. It's built into the 2.6+ kernels (cpufreq) and works like a 
> champ for me.  And unlike the DSDT based throtteling, you can control it!
> 
> Cheers,
>     André



-------------------------------------------------------
This SF.Net email is sponsored by the new InstallShield X.
>From Windows to Linux, servers to mobile, InstallShield X is the one
installation-authoring solution that does it all. Learn more and
evaluate today! http://www.installshield.com/Dev2Dev/0504

Re: Re: DSDT Change (Compaq Presario 2100)

[Andre Eisenbach]

Andrew D. Keyser wrote:

> Although, if you update the BIOS, the problem seems to fix itself...
> Anywho, I find it ironic that HP touts "AMD with PowerNOW" yet they 
> don't actually use the powernow and instead use the crappy DSDT.


Yes, some people tell me that updating the BIOS helps. Which BIOS 
revision and which notebook is it however?
There is no BIOS update (beyond KEM 1.58) for my notebook as of now and 
that one still has the DSDT throtteling enabled.

Cheers,
   André


-------------------------------------------------------
This SF.Net email is sponsored by the new InstallShield X.
>From Windows to Linux, servers to mobile, InstallShield X is the one
installation-authoring solution that does it all. Learn more and
evaluate today! http://www.installshield.com/Dev2Dev/0504

Re: Re: DSDT Change (Compaq Presario 2100)

[Francesco P. Lovergine]

BTW, I had no fortune in booting that laptop with 2.6.5+ and 2.4.25+ 
without either acpi=off or nolapic. I saw an old patch around for
that, but apparently it's (no more?) integrated in recent kernels.

-- 
Francesco P. Lovergine


-------------------------------------------------------
This SF.Net email is sponsored by the new InstallShield X.
>From Windows to Linux, servers to mobile, InstallShield X is the one
installation-authoring solution that does it all. Learn more and
evaluate today! http://www.installshield.com/Dev2Dev/0504

Re: Re: DSDT Change (Compaq Presario 2100)

[Pavel Machek]

Hi!

> > > and errors, I stumbled across this (in method SMSL
> > >
> > > If (0x01)
> > > {
> > > 	Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
> > > 	Store ("AC remove! Set Throttle Duty Cycle ---: ", Debug)
> > > 	Store (TRDC, Debug)
> > > 	Store (0x01, TREN)
> > > }
> >
> > The DSDT for my laptop has the exact same code.
> 
> That piece of code is the state cange code (i.e. when you unplug the AC cord). 
> There is another place you need to change, which initializes the behaviour 
> when you boot with the power cable already unplugged.
> 
> The method is _REG. Here is the code
> 
>                                 Store (0x01, \_SB.PCI0.PMU.CRUN)
>                                 If (0x01)
>                                 {
>                                     Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
>                                     Store (0x01, TREN)
>                                 }
> 
> Once again change "If (0x01)" to "If (0x00)".

Try your machine on battery, with full load. Monitor battery
temperature. If it gets hot, your battery was not really designed for
this, and forget this...
								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!


-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com

Re: DSDT Change (Compaq Presario 2100)

[Erik Meitner]

Pavel Machek wrote:
> Hi!
> 
> 
>>>>and errors, I stumbled across this (in method SMSL
>>>>
>>>>If (0x01)
>>>>{
>>>>	Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
>>>>	Store ("AC remove! Set Throttle Duty Cycle ---: ", Debug)
>>>>	Store (TRDC, Debug)
>>>>	Store (0x01, TREN)
>>>>}
>>>
>>>The DSDT for my laptop has the exact same code.
>>
>>That piece of code is the state cange code (i.e. when you unplug the AC cord). 
>>There is another place you need to change, which initializes the behaviour 
>>when you boot with the power cable already unplugged.
>>
>>The method is _REG. Here is the code
>>
>>                                Store (0x01, \_SB.PCI0.PMU.CRUN)
>>                                If (0x01)
>>                                {
>>                                    Store (\_SB.PCI0.ISA.EC0.TTDC, TRDC)
>>                                    Store (0x01, TREN)
>>                                }
>>
>>Once again change "If (0x01)" to "If (0x00)".
> 
> 
> Try your machine on battery, with full load. Monitor battery
> temperature. If it gets hot, your battery was not really designed for
> this, and forget this...
> 								Pavel

Is the concern here that this allows the system to draw a greater 
current from the battery than it was designed for? If so, too bad that 
there is no way for the system to monitor current draw and throttle back 
when it gets to great.

Erik



-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com

Re: Re: DSDT Change (Compaq Presario 2100)

[Pavel Machek]

Hi!

> >Try your machine on battery, with full load. Monitor battery
> >temperature. If it gets hot, your battery was not really designed for
> >this, and forget this...
> >								Pavel
> 
> Is the concern here that this allows the system to draw a greater 
> current from the battery than it was designed for? If so, too bad that 
> there is no way for the system to monitor current draw and throttle back 
> when it gets to great.

Yes, I was told that can be the issue.

OTOH monitoring battery temperature should be ok.
								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!


-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com

Re: Re: DSDT Change (Compaq Presario 2100)

[Andre Eisenbach]

Pavel Machek wrote:

>>>Try your machine on battery, with full load. Monitor battery
>>>temperature. If it gets hot, your battery was not really designed for
>>>this, and forget this...
>>>								Pavel
>>>      
>>>

Thanks for the heads up on this, Pavel!

I did unplug my batter, started an updatedb and compiled firefox, 
coreutils and a new kernel. The battery got warm, but not hot. 
Definitely not more then when charging it.

A curious (unrelated) observation:
The ACPI thermal zone temperature indicated about 68 C at most  on 
battery doing this. However, when the battery was drained and I plugged 
the AC back in, the temperature rose and stayed at about 75 C and the 
fans worked audibly harder. This continued even after I removed the 
battery, so it's not the additional heat from the charging.
I wonder if something else is throttled when on battery. It's not the 
CPU though as benchmarks dont slow down on battery anymore after my DSDT 
change.

As far as the DSDT change goes. Pavel, do you think that the kind of 
throtteling they do is an effective way to limit battery current draw?
I am getting more battery life now - even with powernow at "performance".

Cheers.
   Andre


-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com

Re: Re: DSDT Change (Compaq Presario 2100)

[Pavel Machek]

Hi!

> Thanks for the heads up on this, Pavel!
> 
> I did unplug my batter, started an updatedb and compiled firefox, 
> coreutils and a new kernel. The battery got warm, but not hot. 
> Definitely not more then when charging it.
> 
> A curious (unrelated) observation:
> The ACPI thermal zone temperature indicated about 68 C at most  on 
> battery doing this. However, when the battery was drained and I plugged 
> the AC back in, the temperature rose and stayed at about 75 C and the 
> fans worked audibly harder. This continued even after I removed the 
> battery, so it's not the additional heat from the charging.
> I wonder if something else is throttled when on battery. It's not the 
> CPU though as benchmarks dont slow down on battery anymore after my DSDT 
> change.

Strange... It could be backlight, but that does not produce enough
heat...


> As far as the DSDT change goes. Pavel, do you think that the kind of 
> throtteling they do is an effective way to limit battery current draw?
> I am getting more battery life now - even with powernow at "performance".

If you are getting more battery life now... well it looks like bios
vendor did something very wrong.
								Pavel

-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!


-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com

Re: Re: DSDT Change (Compaq Presario 2100)

[Michael Frank]

On Fri, 2 Jul 2004 21:21:52 +0200, Pavel Machek <pavel-+ZI9xUNit7I@public.gmane.org> wrote:

> Hi!
>
>> Thanks for the heads up on this, Pavel!
>>
>> I did unplug my batter, started an updatedb and compiled firefox,
>> coreutils and a new kernel. The battery got warm, but not hot.
>> Definitely not more then when charging it.
>>
>> A curious (unrelated) observation:
>> The ACPI thermal zone temperature indicated about 68 C at most  on
>> battery doing this. However, when the battery was drained and I plugged
>> the AC back in, the temperature rose and stayed at about 75 C and the
>> fans worked audibly harder. This continued even after I removed the
>> battery, so it's not the additional heat from the charging.
>> I wonder if something else is throttled when on battery. It's not the
>> CPU though as benchmarks dont slow down on battery anymore after my DSDT
>> change.
>
> Strange... It could be backlight, but that does not produce enough
> heat...
>

The AC adapter produces a voltage which gets stepped down in
the box to power the  bus to which battery and rest of system connect.

The additional temperature rise can be explained by the additional
heat generated in this circutry or other than that, perhaps CPU core
voltage is increased when on AC accounting for the increase in
power consumption. Note that in this case your DSDT change
my run the CPU "too fast" for it's Vcore on battery...

The only way to find out is to measure Vcore - is there a means in the BIOS?

	





-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 -
digital self defense, top technical experts, no vendor pitches,
unmatched networking opportunities. Visit www.blackhat.com

Re: Re: DSDT Change (Compaq Presario 2100)

[Pavel Machek]

Hi!

> The AC adapter produces a voltage which gets stepped down in
> the box to power the  bus to which battery and rest of system connect.
> 
> The additional temperature rise can be explained by the additional
> heat generated in this circutry or other than that, perhaps CPU core
> voltage is increased when on AC accounting for the increase in
> power consumption. Note that in this case your DSDT change
> my run the CPU "too fast" for it's Vcore on battery...

Increasing Vcore on AC when it can run on lower Vcore and same clock
on battery would be pretty strange, no? 
								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!


-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com

Re: Re: DSDT Change (Compaq Presario 2100)

[Michael Frank]

On Sat, 3 Jul 2004 02:37:13 +0200, Pavel Machek <pavel-+ZI9xUNit7I@public.gmane.org> wrote:

> Hi!
>
>> The AC adapter produces a voltage which gets stepped down in
>> the box to power the  bus to which battery and rest of system connect.
>>
>> The additional temperature rise can be explained by the additional
>> heat generated in this circutry or other than that, perhaps CPU core
>> voltage is increased when on AC accounting for the increase in
>> power consumption. Note that in this case your DSDT change
>> my run the CPU "too fast" for it's Vcore on battery...
>
> Increasing Vcore on AC when it can run on lower Vcore and same clock
> on battery would be pretty strange, no?

Considering the original system behavior, it is not strange at all.
IMO they have a simple implemententation of 2 Vcores and 2
frequencies and switch to the lower one one battery.

As it t would not make sense to reduce clock frequency without
reducing Vcore, my concern is that the "design change" increasing
the  frequency may have  unexpected side effects.

Thus right now the system seems "overclocked" on battery and
I suggest again to find a way to measure Vcore to be sure...






-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 -
digital self defense, top technical experts, no vendor pitches,
unmatched networking opportunities. Visit www.blackhat.com

Re: Re: DSDT Change (Compaq Presario 2100)

[Pavel Machek]

Hi!

> >>The AC adapter produces a voltage which gets stepped down in
> >>the box to power the  bus to which battery and rest of system connect.
> >>
> >>The additional temperature rise can be explained by the additional
> >>heat generated in this circutry or other than that, perhaps CPU core
> >>voltage is increased when on AC accounting for the increase in
> >>power consumption. Note that in this case your DSDT change
> >>my run the CPU "too fast" for it's Vcore on battery...
> >
> >Increasing Vcore on AC when it can run on lower Vcore and same clock
> >on battery would be pretty strange, no?
> 
> Considering the original system behavior, it is not strange at all.
> IMO they have a simple implemententation of 2 Vcores and 2
> frequencies and switch to the lower one one battery.
> 
> As it t would not make sense to reduce clock frequency without
> reducing Vcore, my concern is that the "design change" increasing
> the  frequency may have  unexpected side effects.
> 
> Thus right now the system seems "overclocked" on battery and
> I suggest again to find a way to measure Vcore to be sure...

Well, I'd expect it to crash it faster on fast clock but low
voltage... He did some stress testing and it worked well...

Undervolting cpu should not really be dangerous for the hardware (but
is bad idea if you care about your data), right?
								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!


-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com

Re: Re: DSDT Change (Compaq Presario 2100)

[Sérgio Monteiro Basto]

On Fri, 2004-07-02 at 18:33, Andre Eisenbach wrote:
> Pavel Machek wrote:
> 
> >>>Try your machine on battery, with full load. Monitor battery
> >>>temperature. If it gets hot, your battery was not really designed for
> >>>this, and forget this...
> >>>								Pavel
> >>>      
> >>>
> 
> Thanks for the heads up on this, Pavel!
> 
> I did unplug my batter, started an updatedb and compiled firefox, 
> coreutils and a new kernel. The battery got warm, but not hot. 
> Definitely not more then when charging it.
> 
> A curious (unrelated) observation:
> The ACPI thermal zone temperature indicated about 68 C at most  on 
> battery doing this. However, when the battery was drained and I plugged 
> the AC back in, the temperature rose and stayed at about 75 C and the 
> fans worked audibly harder. This continued even after I removed the 
> battery, so it's not the additional heat from the charging.
> I wonder if something else is throttled when on battery. It's not the 
> CPU though as benchmarks dont slow down on battery anymore after my DSDT 
> change.
> 
> As far as the DSDT change goes. Pavel, do you think that the kind of 
> throtteling they do is an effective way to limit battery current draw?
> I am getting more battery life now - even with powernow at "performance".
> 
> Cheers.
>    Andre
Hi 
with cpufreq on kernel 2.4.2x
you can get even better "performance" (I guess)
http://ftp.linux.org.uk/pub/linux/cpufreq/

About nolapic problem. I understand that you, also, have to put
"nolapic" on boot kernel to boot. Well is an old problem on some
presarios like my 700. 
We have one patch that can resolve boot problem, I am testing it and the
only problem that I found by now, is that hangs when trying change video
output (Fn+F3).
the last propose patch on
http://bugme.osdl.org/show_bug.cgi?id=1269

ah! btw my dir /proc/acpi/thermal_zone/ is empty and laptop lives well
with that.

cheers,
-- 
Sérgio M. B.



-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 -
digital self defense, top technical experts, no vendor pitches,
unmatched networking opportunities. Visit www.blackhat.com