You are not logged in.

#1 2012-01-20 14:13:31

Lockheed
Member
Registered: 2010-03-16
Posts: 1,429

Unvervolting issues

I have been using this script to undevrolt my C2D T8100:
http://openmindedbrain.info/09/05/2010/ … lucid-lts/

All goes well, and all 5 steps go as low as VID 0. Here's an example for the highest frequency:

>Found 2 CPU cores.
Multiple cores found, the script will use the
same voltages for all cores, this means final
setting will be 4 VIDs above the weakest core.

Read default VIDs.
> Success!
Reset position to 0.
Read available frequencies.
> Success!

Default VIDs: 39 34 28 23 19 
Current VIDs: 39 34 28 23 19

Switch to 'userspace' scaling governor.
Password: 
> Success!
Set frequency to 2101000.
> Success!
Run burnMMX.
 PID: 3212
 PID: 3213

-----
Start testing.
Do you want to continue? [Y/n/?] Yes

Default VIDs: 39 34 28 23 19 
Current VIDs: 38 34 28 23 19
Testing VID: 38 (1308 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 37 34 28 23 19
Testing VID: 37 (1292 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 36 34 28 23 19
Testing VID: 36 (1276 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 35 34 28 23 19
Testing VID: 35 (1260 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 34 34 28 23 19
Testing VID: 34 (1244 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 33 34 28 23 19
Testing VID: 33 (1228 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 32 34 28 23 19
Testing VID: 32 (1212 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 31 34 28 23 19
Testing VID: 31 (1196 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 30 34 28 23 19
Testing VID: 30 (1180 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 29 34 28 23 19
Testing VID: 29 (1164 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 28 34 28 23 19
Testing VID: 28 (1148 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 27 34 28 23 19
Testing VID: 27 (1132 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 26 34 28 23 19
Testing VID: 26 (1116 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 25 34 28 23 19
Testing VID: 25 (1100 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 24 34 28 23 19
Testing VID: 24 (1084 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 23 34 28 23 19
Testing VID: 23 (1068 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 22 34 28 23 19
Testing VID: 22 (1052 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 21 34 28 23 19
Testing VID: 21 (1036 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 20 34 28 23 19
Testing VID: 20 (1020 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 19 34 28 23 19
Testing VID: 19 (1004 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 18 34 28 23 19
Testing VID: 18 (988 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 17 34 28 23 19
Testing VID: 17 (972 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 16 34 28 23 19
Testing VID: 16 (956 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 15 34 28 23 19
Testing VID: 15 (940 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 14 34 28 23 19
Testing VID: 14 (924 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 13 34 28 23 19
Testing VID: 13 (908 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 12 34 28 23 19
Testing VID: 12 (892 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 11 34 28 23 19
Testing VID: 11 (876 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 10 34 28 23 19
Testing VID: 10 (860 mV)
...............................
.............................^A..
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 9 34 28 23 19
Testing VID: 9 (844 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 8 34 28 23 19
Testing VID: 8 (828 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 7 34 28 23 19
Testing VID: 7 (812 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 6 34 28 23 19
Testing VID: 6 (796 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 5 34 28 23 19
Testing VID: 5 (780 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 4 34 28 23 19
Testing VID: 4 (764 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 3 34 28 23 19
Testing VID: 3 (748 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 2 34 28 23 19
Testing VID: 2 (732 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 1 34 28 23 19
Testing VID: 1 (716 mV)
...............................
...............................
...............................
Default VIDs: 39 34 28 23 19 
Current VIDs: 0 34 28 23 19
Testing VID: 0 (700 mV)
...............................
...............................
...............................

The lowest acceptable VID is 0.

Recovering CPU.
./intel-phc-undervolt.bash: line 354:  3212 Terminated              burnMMX
./intel-phc-undervolt.bash: line 354:  3213 Terminated              burnMMX

Run this script again to continue the optimization.

However, when I then try to set the tested voltages with

echo "1 0 0 0 0" > /sys/devices/system/cpu/cpu0/cpufreq/phc_vids

the system crashes. The system does not crash if I change the first value to the nominal 39, but will crash with any other value.

What is wrong?


Laptop: ThinkPad W500, C2D P9500, 8GB, Radeon RV635 (HD3650), Arch | Server/fw: Zotac AQ01, A4-5000 Kabini, 4GB, Arch/pfSense VM

Offline

#2 2012-01-21 00:24:29

ngoonee
Forum Fellow
From: Between Thailand and Singapore
Registered: 2009-03-17
Posts: 6,790

Re: Unvervolting issues

Haven't played with that for a while, but aren't you setting the minimum VID (all those 1s and 0s) which is not likely EVER to work?

The last time I played with it I realized that this happened (test supposedly let the VID go all the way down to the minimum) but that would not work in real-life. I'd suggest just reducing the VID by one and using your machine normally (maybe play some games) rather than relying on the script (which never helped for me).

Or just buy a bigger battery smile


Allan-Volunteer on the (topic being discussed) mailn lists. You never get the people who matters attention on the forums.
jasonwryan-Installing Arch is a measure of your literacy. Maintaining Arch is a measure of your diligence. Contributing to Arch is a measure of your competence.
Griemak-Bleeding edge, not bleeding flat. Edge denotes falls will occur from time to time. Bring your own parachute.

Offline

#3 2012-01-21 00:38:13

Lockheed
Member
Registered: 2010-03-16
Posts: 1,429

Re: Unvervolting issues

I lower the voltage to lower heat. I already have bigger battery I never use smile
Setting VID to 0 should just make the CPU run on minimal available voltage, so it should work just fine (and it does seem to work for frequencies other than the highest).

However, it looks like modifying the first VID to any number causes the crash, because I tried something around 1V (normal low C2D voltage) and it crashed, too.


Laptop: ThinkPad W500, C2D P9500, 8GB, Radeon RV635 (HD3650), Arch | Server/fw: Zotac AQ01, A4-5000 Kabini, 4GB, Arch/pfSense VM

Offline

#4 2012-01-21 10:18:04

ngoonee
Forum Fellow
From: Between Thailand and Singapore
Registered: 2009-03-17
Posts: 6,790

Re: Unvervolting issues

Lockheed wrote:

I lower the voltage to lower heat. I already have bigger battery I never use smile
Setting VID to 0 should just make the CPU run on minimal available voltage, so it should work just fine (and it does seem to work for frequencies other than the highest).

However, it looks like modifying the first VID to any number causes the crash, because I tried something around 1V (normal low C2D voltage) and it crashed, too.

Lowering voltage lowers stability as well. Too low and your CPU won't be able to perform its calculations, read up on it.

There MAY also be a bug which explains what you're seeing (you'd need to talk to the phc people, I think). But its entirely possible that your CPU just can't run on a lower voltage than stock.


Allan-Volunteer on the (topic being discussed) mailn lists. You never get the people who matters attention on the forums.
jasonwryan-Installing Arch is a measure of your literacy. Maintaining Arch is a measure of your diligence. Contributing to Arch is a measure of your competence.
Griemak-Bleeding edge, not bleeding flat. Edge denotes falls will occur from time to time. Bring your own parachute.

Offline

#5 2012-01-21 10:28:36

Lockheed
Member
Registered: 2010-03-16
Posts: 1,429

Re: Unvervolting issues

I don't really need to read up on it, I've been doing it since 12 years on windows, and since 2 years on AMD cpus on linux (it's quite a different beast, than Intel undervolting).
Intel CPUs can't run on lower voltages than minimal V for lowest multiplier, but highest multiplier should be able to go down to the lowest multiplier V without issues, if the chip is good. And mine is, because it works just fine on Windows.

So, the issue is not the hardware, but purely linux-related. And talking to PHC people might not give me much, since the project is dead since nearly 3 years.


Laptop: ThinkPad W500, C2D P9500, 8GB, Radeon RV635 (HD3650), Arch | Server/fw: Zotac AQ01, A4-5000 Kabini, 4GB, Arch/pfSense VM

Offline

#6 2012-01-21 15:52:19

EiyuuZack
Member
Registered: 2011-04-11
Posts: 17

Re: Unvervolting issues

Last year I undervolted my laptop using the same script, and I set my VIDs in

/etc/conf.d/phc-intel

and just added the phc-intel daemon

DAEMONS=( ... dbus @phc-intel @laptop-mode ... )

always calling

# /etc/rc.d/phc-intel setup

whenever I update my kernel, which is (as of this writing) a vanilla

$ uname -r
3.2.1-1-ARCH

Regarding your VID problem, my results were 16 0 0 0 0 as my (old) laptop actually crashed on VID 16 during the highest frequency test so my actual VIDs are set as 18 2 2 2 2 (added 2 on the remaining frequencies just to be on the safe side), so I don't think I can help you there as your results were 0 on all frequencies.

Maybe the problem is you echoing the values directly to the file(s)?

Offline

#7 2012-01-22 11:12:19

Lockheed
Member
Registered: 2010-03-16
Posts: 1,429

Re: Unvervolting issues

I did some further testing and I was able to go down to VID 22 (1052mV) on first multiplier.

However, the problem I am still left with is, that my lm_sensors won't show Voltages. Here it is in detail:
https://bbs.archlinux.org/viewtopic.php?id=133948
I would appreciate any suggestions.


Laptop: ThinkPad W500, C2D P9500, 8GB, Radeon RV635 (HD3650), Arch | Server/fw: Zotac AQ01, A4-5000 Kabini, 4GB, Arch/pfSense VM

Offline

#8 2012-01-28 08:39:57

stqn
Member
Registered: 2010-03-19
Posts: 1,189
Website

Re: Unvervolting issues

Thanks for starting this thread Lockheed. For some reason I had forgotten to undervolt my processor after my laptop died and I bought a Core i3 smile.

I'd like to point out that the script doesn't work very well for me: frequency locking didn't work so only the highest frequency voltage was found correctly. When the script checks the lower frequency multipliers, it's actually still testing the highest frequency. So it finds zero for all the lower vids no matter what.

But with my result (a vid of 0 for the highest frequency), I don't think it matters much; I simply set "3 3 3 3 2 2 2 2 1 1 1 1 0 0 0 0" for my vids and it seems to be working fine. (Since it's recommended to add 4 to the crashing vid, so -1 + 4 = 3.)

I'm currently running a linpack stress test to further check that everything's fine. It's a pleasure that even at full load I don't hear my fan any more than at idle before smile.
(linpack only uses two cores out of my four virtual cores though...)

Offline

#9 2012-03-08 22:53:10

mister_playboy
Member
Registered: 2012-01-23
Posts: 17

Re: Unvervolting issues

I have a T2390 C2D (Merom), and it didn't work with the script, either.  Testing by hand (and using a Windows undervolting tool back when I dual-booted), I found VID 19 was the best I could do for all 5 steps.  Anything lower than 19 was actually treated as 19 by the CPU... some sort of failsafe?

Last edited by mister_playboy (2012-03-08 22:57:46)


Teach not that I died in vain; rather, teach that I did not die by demonstrating that I live in you.--- ACIM

Offline

#10 2012-03-09 00:17:01

Lockheed
Member
Registered: 2010-03-16
Posts: 1,429

Re: Unvervolting issues

I cannot comment on your problems with undervolting, as all seem to work well on my side.

I wanted to follow-up on the subject of CPU Voltage reading. Turns out my laptop. as most, does not have dedicated Voltage sensor, so lm-sensors are useless for that purpose. HOWEVER, I discussed it on Gentoo forums and found a solution (similar to the one being used in RMClock under windows). The bottom line is:

____Yes, you can read voltage values of your CPU even if you do not have appropriate sensors.____

Follow here for the details: http://forums.gentoo.org/viewtopic.php? … 96#6975796

Last edited by Lockheed (2012-03-09 00:17:30)


Laptop: ThinkPad W500, C2D P9500, 8GB, Radeon RV635 (HD3650), Arch | Server/fw: Zotac AQ01, A4-5000 Kabini, 4GB, Arch/pfSense VM

Offline

#11 2012-08-05 00:42:30

stqn
Member
Registered: 2010-03-19
Posts: 1,189
Website

Re: Unvervolting issues

If anyone is interested, I wrote another script to find the lowest acceptable vids after finding out that the results returned by the script mentionned in the first post of this thread were wrong and made my netbook *very* unstable.

Offline

Board footer

Powered by FluxBB