Managing Pentium-M Speed/Voltage with software (temp sticky)

Message

CoolDragon · #91 Post by **CoolDragon** » Mon Jun 27, 2005 7:20 pm

I saw on the CHC homepage, it list "fan speed control" as a feature. So anybody tried this out?

Aristotle11 · #92 Post by **Aristotle11** » Tue Jun 28, 2005 9:16 am

CoolDragon wrote:I saw on the CHC homepage, it list "fan speed control" as a feature. So anybody tried this out?

Unfortunately, it doesn't work on thinkpads. I WOULD SELL MY OWN MOTHER FOR A T42 FAN CONTROL UTILITY!

(Ok, maybe I'd just rent her, so long as they were following OSHA regulations, but still.)

sugo · #93 Post by **sugo** » Tue Jun 28, 2005 7:29 pm

The author says he's working on a fan control solution via ACPI.

CoolDragon · #94 Post by **CoolDragon** » Tue Jun 28, 2005 11:45 pm

I think it is just that IBM/Lenovo doesn't want to correct this issue. Controlling the fan speed based on the temperature is so easy.

JHaislet · #95 Post by **JHaislet** » Tue Jun 28, 2005 11:47 pm

nirvana0001 · #96 Post by **nirvana0001** » Wed Jun 29, 2005 2:02 am

JHaislet wrote:Here we go: http://www.cenatek.com/

Gigabyte might be the first out with a PCI Express slot, but Cenatek has had 32 & 64bit PCI-X RocketDrives that run at 33,66,100, & 133mhz PCI-X bus speeds for almost 5yrs now.

wrong reply??

JHaislet · #97 Post by **JHaislet** » Wed Jun 29, 2005 2:24 am

Ops! Sorry about that! That's what I get for having multiple windows of thinkpads.com open LOL!

DJHingapus · #98 Post by **DJHingapus** » Thu Jul 07, 2005 3:03 am

I'm not so good at this computer stuff, so I want to know if turning down the voltage with CHC effect the CPU performance or is it the same performance (ex. If I run at 600mhz with voltage 0.998 or 0.700 is there any performance difference)

I have currently said CHC to dynamic switching and the voltage on 600mhz to 0.700 the other CPU speeds are set to normal voltage. I use this setting so when I want to take advance of the CPU it runs default. But it is here I want to know if I again can turn down the voltage on the other speeds an still have same performance??

Another question I can't find an answer on is what 'advanced power management' means in CHC, what does it do??

Best Regards
Rasmus

jhonyl · #99 Post by **jhonyl** » Thu Jul 07, 2005 7:25 am

Hi DJHingapus,

The performance can be hit if the processor can't operate well in the voltage you set for it i.e. You have set a too low voltage and therefor the processor does mistakes. To avoid it you should run test programs such as Prime95 which will check the accuracy of the computations that the CPU does.

Lower voltage means lower heat. Lower MHZ means less computations. If you lower the voltage for the same MHZ the performance should be the same, as long as Prime95 test checks out OK.

Because you want less heat (=less physical power=more battery time) you need to find the lowest voltage that checks out OK on Prime95.

EatMorGlue · #100 Post by **EatMorGlue** » Sun Jul 10, 2005 9:52 pm

How long are you folks testing for? Is two hours of Prime95 testing sufficient to consider the processor stable at a given speed and voltage?

Or put another way, have any of you run into failures more than two hours into testing?

JHaislet · #101 Post by **JHaislet** » Sun Jul 10, 2005 10:13 pm

I ran Prime95 for about an hour I think, after setting my voltage lower. I have also run SiSoft's Sandra benchmark burn in program a couple times which stresses the ram, cpu & cache individually. Also, I've run the 15min CHC burn in a couple times. All without error. This is good enough for me, as my T43 is mainly used for non-cpu intensive applications like internet, word, and e-mail. I've yet to experience any lockups or crashes, and will keep my cpu at 15x @ 1.084v until I do.

Doltan · #102 Post by **Doltan** » Mon Jul 11, 2005 9:41 am

Run mine for 1 day to test the different settings, then run for 3 days to confirm the final setting.

Strange thing is, Prime95 may not show error at times, but when I run other applications at the same time, it wasn't so stable. So I increased the voltage a little bit more to create some buffer in the end.

EatMorGlue · #103 Post by **EatMorGlue** » Mon Jul 11, 2005 8:23 pm

Doltan wrote:Strange thing is, Prime95 may not show error at times, but when I run other applications at the same time, it wasn't so stable. So I increased the voltage a little bit more to create some buffer in the end.

Interesting... what other programs were you running? The reason I ask is that while testing my Dothan 725 at 8x, I would test error-free for a couple hours at different voltages before bumping it down a step, waiting to get an error and establish the lower limit for voltage at 8x. But it never happened, and it finally got to where I tested 8x @ 0.700V for 8.5 hours error free. All the while though, I'm not actively using other programs.

The CHC homepage seems to suggest 0.700V is pretty low for 6x, operating without a safety margin per se, so it seems unbelievably low for 8x.

Like JHaislet, I'm also not normally running CPU intensive programs on my machine, so this isn't an urgent problem for me or anything. But I'm just finding it curious that I've not been able to establish a failure point using Prime95, even when the voltage seems extreme. I guess I'll just try running some other programs to see if this will help me establish my limits.

Doltan · #104 Post by **Doltan** » Tue Jul 12, 2005 8:40 am

One was that, when I rebooted, while windows is still starting, an error popped up asking me whether or not to send the error report to Microsoft. Suspect some of the startup processes did not startup properly.

Another was, while still running Prime95, I started Firefox, and it gave me an error too.

Third time, I can't remember starting which other application.

All these happened when I adjusted for 16x. 8x at 0.7V no issue all the time. In the latter 2 cases, Prime95 was still running, and I have never seen any errors in all the times that I ran. In the end, I just bumped up the voltage 2 notches (since its good to leave some buffer too) and run for 3 days without error. No issue so far.

Aristotle11 · #105 Post by **Aristotle11** » Tue Jul 12, 2005 1:07 pm

1. Prime95 is the best single program for testing stability at a certain speed and voltage, but it is not the only program. Just because it is Prime95 stable doesn't mean it is stable for all other programs (although often it is a very good indicator).

2. I'd raise the voltage a bit and see if there are any more crashes. My wife's X31 was prime95 stable for hours at 600MHz, 0.7 volts, but it crashed at the same setting a few times and so I unloaded CHC entirely.

3. If your processor is 100% stable at 600 MHz 0.7 volts (as my T42 is), it is just as fast as it is at 600MHz at the default 0.945 volts (or whatever the default is). Higher voltage by itself does not increase performance. Higher voltage just enables higher MHz, or more stability at the same MHz (both of which can mean more performance). Lower voltage usually means less heat, less fan noise, and more battery life (and maybe a slight increase in processor life from 50 years to 51 years or something).

Best,
Aris

zane · #106 Post by **zane** » Sat Jul 16, 2005 2:05 pm

CHC is really a good software. At the default voltage, my t43 is about 92C at 100% loading. After the voltage is set to 1.002V (1866MHz, 14x), the CPU temperature is about 65C at 100% loading.

Does anybody know the similar software in linux?

Tan Mann · #107 Post by **Tan Mann** » Sat Jul 16, 2005 10:51 pm

zane wrote:CHC is really a good software. At the default voltage, my t43 is about 92C at 100% loading. After the voltage is set to 1.002V (1866MHz, 14x), the CPU temperature is about 65C at 100% loading.

Does anybody know the similar software in linux?

92C all full speed!?!?!?! Are you kidding me!?!?!? That is near boiling point of water...MY Dothan 2Ghz in my 2372WUB does only 66C at full speed at 100% load...and 53C at 1.054V at 2Ghz Full Load.

This is very strange

bhtooefr · #108 Post by **bhtooefr** » Sat Jul 16, 2005 11:17 pm

Call IBM about the heat if you're still in warranty.

If you're not, go in, and replace whatever thermal transfer material that they're using (goop or a pad) with some Arctic Silver 5.

I'm going to do that to my X21 - I've been as high as 96 C on a hot day in the UltraBase at 100% CPU (granted, I have no AC).

Navck · #109 Post by **Navck** » Sat Jul 16, 2005 11:54 pm

96C?! Near the point of BOILING WATER!

bhtooefr · #110 Post by **bhtooefr** » Sun Jul 17, 2005 12:03 am

Let's just say that the ACPI shutdown point is 100 C - 212 F - the temperature of boiling water.

I was THAT close to the laptop performing an emergency turnoff.

I immediately pulled it out of the UltraBase, and put it in battery mode on SpeedStep (possible, even when on AC power).

Doltan · #111 Post by **Doltan** » Sun Jul 17, 2005 12:56 am

I think mine reached around 85-87 degC max, before undervolting. Not really sure what it reached after undervolting though.

roast · #112 Post by **roast** » Mon Jul 18, 2005 12:06 pm

zane wrote:CHC is really a good software. At the default voltage, my t43 is about 92C at 100% loading. After the voltage is set to 1.002V (1866MHz, 14x), the CPU temperature is about 65C at 100% loading.

Does anybody know the similar software in linux?

No, but I believe you can encode the lower voltages that you determined into the arch/i386/.../cpufreq.c for your specific processor and recompile. The included governors should help you achieve the same thing as CHC does.

BoostedA4 · #113 Post by **BoostedA4** » Mon Jul 18, 2005 2:11 pm

any update on fan control software for the T4x ???

Henry · #114 Post by **Henry** » Wed Jul 20, 2005 2:32 pm

My Pentium M 735 is running at 67C with distributed computing (100% cpu usage). Is this bad for the computer, or should I undervolt? I mean, obviously the lifespan of the cpu will decrease, but should it still be alive by the time it's outdated?

nirvana0001 · #115 Post by **nirvana0001** » Sat Jul 23, 2005 10:52 am

anybody know the new CHC 1.9 was released?

Henry · #116 Post by **Henry** » Mon Jul 25, 2005 1:28 pm

Yeah, CHC 1.9 Beta is released. But, for some reason, it would not run when I tried it. Maybe it's just me, or maybe some bug with the .net framework or something.

nirvana0001 · #117 Post by **nirvana0001** » Mon Jul 25, 2005 7:33 pm

Henry wrote:Yeah, CHC 1.9 Beta is released. But, for some reason, it would not run when I tried it. Maybe it's just me, or maybe some bug with the .net framework or something.

it works cool with me

Doltan · #118 Post by **Doltan** » Tue Jul 26, 2005 6:51 am

Yup, no prob here too.

Henry · #119 Post by **Henry** » Wed Jul 27, 2005 2:21 pm

nirvana0001 wrote:
Henry wrote:Yeah, CHC 1.9 Beta is released. But, for some reason, it would not run when I tried it. Maybe it's just me, or maybe some bug with the .net framework or something.
it works cool with me

Hmm, that's weird. Does it have to do with the directory in which it's saved then? Where did you guys download it to?

Celestar · #120 Post by **Celestar** » Sun Jul 31, 2005 5:53 am

Somehow it seems I am only able to find voltages for the Banjas CPUs in the linux kernel, has anyone undervolted a Dothan @ linux?

Celestar

Managing Pentium-M Speed/Voltage with software (temp sticky)

Who is online