Undervolting the AMD Turion 64 ML-34 processor (L2000)

Newer revisions did not allow the lowest multiplier voltage to be changed in many processors(both Intel & AMD). I think this might be the case as well with your A64.

 

Humm, I guess this is how AMD produce different different CPU model from same production line. They made same die, and lock those chips that cannot pass the low voltage test, sell it as desktop version. Same thing happened in Athlon XP.

 

I stumble upon this page when looking for ways to lower my voltage and man am I happy I did. After reading through this thread, I decided to test my luck.

I am able to get the ml-30 down to 0.9v @ 4x and 1.05v @ 8x. I ran the prime95 test for over 8 hours and it was rock solid. When running the test, I turned on the fan to "always on" and the temperature never got higher then 48c. Before undervolting, it was consistenly around 63c when at full load.

Thanks a bunch guys. Now, I have a much more efficient laptop top.

 

Does anybody know how to make undervolt available in Linux?
I just read couple of post talking about how to do it using Intel Centr CPU. Would there be any different?

 

Can you give us a link to those Linux undervolting posts?

 

Check this.
http://www.thinkwiki.org/wiki/Pentium_M_undervolting_and_underclocking

I am sure there have others, but I don't see any on Turion.

 

hi all!

At first, thanks for this topic and informations..

I've an acer travelmate 4402, with turion ml-30 (SH-E5),
and i can underclock to:

4x 0.9V
8x 1.1V

Prime95 running for 2 hours now, i'm writing a dvd now, and write this post
The cpu's temp is 61C, before underclock it was about 72-73C.

:turion:

 

Excellent thread. I saw an interesting note regarding 4400 vs 4000mAh laptop batteries. How can you tell if you have a "crippled" one?

Also, I've been trying to figure out how to use RM Clock 2.0.5. The interface has changed since the begining of the thread. I also saw a note by Chinna_n,
but I found the instructions a little lacking, it would be nice if they where a little more details

Regards,

Mike

 

You can check the battery by taking it out and reading the lable

 

Thanks .

 

What program that tells the temp of my cpu would work for my HP dv8210? See sig. for specs.

 

I have a dv5000z. I just tried to load the AMD Dashboard and to my surprise no temperature reading! I'm running Windows Media Center. Could this be why the termperature isn't showing?

The CPU fan is comming on, so I'm guessing the thermostat must be working.

Mike

 

No, best guess is that HP didn't implement ACPI correctly in their BIOS. Most temperature monitors (like MobileMeter) try to read ACPI.

 

Is this the case for all dv5000z? Strange since it worked fine on the
Compaq V2000z.

Mike

 

What are the vaild steps for the ML-32? Can you just select any multiplier you want? I only see 4x and 9x in the report screen.

Mike

 

Yeah for the ML 34/32, the multiplier ranges are from 4x to 9x, so you can use anyone inbetween. Avoid using fractions. The stock Powernow settings should be 4x, 8x and 9x(I think). It is good to start off with the stock settings and slowly undervolt and then select various other multipliers as and when you get to know the minimum voltages for these 3 multipliers.

 

Hello everyone,

here's my quick processor specs:

AMD Turion(tm) 64 Mobile Technology ML-30 (1.6Ghz)
MMX, SSE, SSE2, SSE3, MMX+, 3DNow!, 3DNow!+, NX/XD, PowerNow!/Cool'n'Quiet and AMD64 Enabled
___________________________________________

Voltage settings BEFORE tweaking:

Idle: 1.0V (Max Battery Setting)
Idle Temp: 43 C
CPU Speed: 50%
CPU Frequency: 800 Mhz
Voltage: 0.900 V

Idle: 1.5V (Home/Office Desk Setting)
Idle Temp: 54 C
CPU Speed: 100%
CPU Frequency: 1,600 Mhz
Voltage: 1.350 V


After following this thread of unvolting techniques/methods, I thought I had successfully got this working...but whenever I switch in and out of power schemes, the Voltage on the Power Monitor and CrystalCPUID keeps reverting back to the original settings.

I was able to undervolt -.150 on Home/Office Desk setting and then ran a test in Hot CPU tester pro 4 and my laptop was still running with no crashes. All Module testing ran through once and finished without error.

What could I be doing wrong? I had a nice report i was gonna type out of my experiments until I double checked just now and notice it was doing this-_-

 

Your Power Scheme is set to Always On? So long as you leave it as that and let CrystalCPUID manage your speed/voltage you should be ok.

 

Oh it has to be on "Always On" Setting? Isn't Presentation, Always On, and Max Battery the same thing/settings? Ok i'll give that a try then.

What if I want to try it on Batteries, should I change it to Max Battery?

 

No. CrystalCPUID is managing everything for you. "Always On" disables the regular power management that CrystalCPUID replaces.

 

Ohhh I see. Well I was able to reduce it by another .050 so I think i'm good for now. Thanks^^

Oh I notice you're in Michigan I used to live there, ever been to Wizzywigs there?

 

Wizzywigs... hmm, nope. There's a few anime series that I like (who doesn't like GiTS and Cowboy Bebop?) but I'm not seriously into it. I bet the UofM provides them with a steady stream of customers.

 

Here are my stats:

Undervolting on the Turion ML-37 (2.0GHz). CPU temps are unknown due to unsupported or not integrated sensors, but I can tell you there is a noticable difference.

Before:
4.0x @ 1.000V
10.0x @ 1.500V

After:
4.0x @ 0.875V
10.0x @ 1.275V

I use RM Clock 2.05.

 

Is there ANY programs out there that will read the cpu temp on the dv8000 series notebook?
What would be the best fan speed program that will work with the dv8000?

 

That is pretty good at 4x. I can not go below 0.925v at 4x, but at 10x mine min is 2.225( but set at 2.25 for margin).

So, I guess you should be able to go down a lot at 10x.(like 1.2v at 10x). And make sure you add some intermediate steppings ( I have 4x, 6x, 8x,9x,10x).

 

What are your voltages for the intermediate steppings? Or did you set them as auto?

 

I tested each stepping individually. But Auto works pretty good too.
Mine are

[email protected]
[email protected]
[email protected]
[email protected]
[email protected]

These are lowest stable voltage + 0.025v. As I mentioned, I added 0.025v for all steppings over lowest stable voltage for a margin.

 

Howdy, everyone.

First time poster, but long, long time leecher. First of all, thank you all for this awesome thread, and i mean it.

I bought a Compaq V2410US while on holidays at Hawaii six months ago and I hadn't use it yet (well, i checked everything worked fine during a couple of days right after i bought it, obviously). The reason i had never used it was because surprisingly, my old notebook, a mitac 8640 powered by a P4 eek: ) still runs. And that's pretty amazing, since the DVD-Drive and built-in Memory Card Reader haven't properly worked for over a year (in fact, they no longer work at all), the fan makes so much noise it seems it's gonna burst or something even worse, battery is 100% useless... I could go on and on, but i guess you get the point. However, i can still use it for basic things, even light gaming, as a weak desktop replacement.

But all things come to an end, and the poor creature is about to die. Anyone can see it, and I have started to use the Compaq. I've carefully read this over 20 pages thread, and i think i understand the advantages of undervolting. It should improve battery life and reduce heat, which is perfect. The battery issue does not concern me, at least right now, as unfortunately mobile meter says my original battery has reached a 35% wear level. I live in Tahiti and well, i guess tropical weather, with high temperature and humidity, is atrocious for electronic devices. I will get a new one once i leave this island.

However, heat generation does concern me. A lot. It's always hot, here, and i assume that won't help at all, but i haven't noticed the slightest improvement in this area after the undervolting.

System:
Turion 64 ML-30 (1.6Ghz) - Six months ago it wasn't that bad
Radeon Xpress 200M (128Mb All shared)
1Gb Kingston DDR PC2700 333 (2x512)
80Gb HDD (it's the original one for this model, don't know the manufacturer)
BIOS F.11

Specs before undervolting.
FID 4x VID 1.000 800Mhz
FID 8x VID 1.450 1600Mhz

Specs after undervolting.
FID 4x VID 0.925 800Mhz
FID 6x VID 1.025 1200Mhz
FID 8x VID 1.100 1600Mhz

I decided to play safe and pushed both vids .025 over limit values. System crashed at 0.875 and 1.050, hence the results. At minimum power, it run Prime95 for over four hours and on maximum performance over six. In both cases system was stable as a rock. No errors, no warnings. The intermediate value was automatically scaled by RMClock.

But, as i told you before, i see no improvement in the heat issue. Mobile meter always reports temperature between 50ºC and 55ºC, both before and after undervolting. Those are fan operating values, as far as i know. When gaming, it reached 60ºC before undervolting. I haven't checked that out yet.

So, am I doing something wrong? Is my Turion hottie? I did expect a drop in temperature values around 5-10ºC (as others claim). Was that overly optimistic, maybe?

Well, any answers will be more than welmcome. Thanks for your reading and sorry for the long post (it's definitely longer than i expected).

Cheers, all!

 

Dihan, the temperature of the laptop depends primarily on the ambient temp. If the ambient temperature is above 27-28 degrees (here this is the benchmark for the summer ), there is no way this thin notebook will run at 40 degrees. Actually, the greater concern with my L2000 was the HDD and its temperature. The thin design of the machine does not allow great heat dissipation and my Toshiba ran between 48-52C. This makes the whole notebook much warmer and so I bought a cooler which instantly broght down the temp to 35-39C.
What you can do about the CPU is to update your BIOS. Newer BIOSes have a higher temp threshold for the fan. The fan kicks off at 58 and stops at 50; thus it really runs less often (once every 20 min for me without the cooler with just Internet browsing).
Other than that, you can try testing the CPU at 0.9v for some time with Prime95 and see if it works stable. I removed my 0,25v cushion after testing the CPU for some hours and have not had problems so far (except for the brief priod when I used RMclock 2.05; went back to RMclock 1.8)

 

Ops, then... if the temperature of the laptop depends primarily on the ambient temp, i'm screwed

We're heading into winter here, although we have pretty much the same temperatures all year around. Right now, i'd say 22-28ºC all day. The problem is high humidity makes it feel like it is much hotter, up to 35ºC. Anyway, right now is not that bad. December and January are just inhuman. However, inside my tiny appartment it's always around 4-5ºC higher than outside. So, basically, there's nothing i can do about heat.

I do have a coolpad. I didn't turn it on during the undervolting tests, but it's usually on and even with it, temperatures are stable at 50-55ºC. So basically, it's not doing any good. It has two fans at 1800rpm. I guess i'll have to upgrade to something more powerfull or at least with the fans properly placed. In my particular model, they're too close to the center.

Two more questions, if you don't mind. Is it worth steping .025 down to reach 0.900 at 4x? I mean, will it clearly help my system run cooler? I don't want to risk stability if it's not really worth it.

And second, if i upgrade to F.24A, the fan will kick off at 58ºC and stop at 50º. Fan will work less, but overall temperature will increase, won't it?

Thanks a lot for your answers.

Cheers, all!

 

About the cooler. When I chose mine, I was very careful to find one which blows air into the lapton and not one which takes air out of it. Many coolers act like de-heaters, that is they take the warm air and blow it away. This won't do much good with a laptop that does not have a constantly running fan (like v2000z, l2000....). The cooler I bought is a little bit unusual - only one big fan in the centre. The cooler is like a hat turned upside down; concave and the fan is at the bottom; I really don't know how to explain it but the fan is some 5 cm below the laptop and blows air into it. Actually, on the package of the cooler it says "Patent Pending"... it does a pretty good job.
Going down 0.25v will not really bring you a considerable heat reduction; if any. Yet, I chose to do it. The difference in temp from the new BIOS is negligible. The fan brings it down to the same temp so, it is just a matter of how annoying you find the fan noise.

 

I finally decided to undervolt the ML-32 processor in my V2000, and here are my results:

FID @ 4.0x, VID - 0.850V
FID @ 8.0x, VID - 1.050V
FID @ 9.0x, VID - 1.125V
(I already added 0.025V buffer to the stable values.)

My only question is what exactly defines "stable"? I've been running Hot CPU Test for 6 hours at each frequency, and it hasn't discovered any problems at all.

 

I have the results for your first ever MT-37 undervolt! :turion: !

I have an Evesham Notebook (based off an MSI MS-1029 chassis) with 1024MB of DDR RAM and a 128MB Radeon X700.

Odd thing here is AMD Power Monitor always reports 0.1V less than RMClock/Dashboard Demo, and I don't know which is right, so I'll just say what I set in RMClock. I test with the Prime95 In-Place Max Heat Torture Test.

So here goes:

Stock Voltages:
4x 1.000V
10x 1.300V

Results:
4x
0.875V FAIL (System hangs)
0.900V PASS (Ran Prime95 for 10 minutes with no issues)
0.925V PASS + USING (For stability)

10x
1.200V FAIL (Prime95 reports errors as soon as torture test started)
1.225V UNSTABLE (Prime95 reported errors after 10 minutes, probably as this is borderline voltage)
1.250V PASS AND USING (For stability)

I have the latest version of RMClock set to lock at 10x when plugged in, lock at 4x when running from battery. Yes I know I could use Performance on Demand but I do notice the microsecond differences from clock changes

Now the thing about this notebook is it's a hot little thing. At stock voltages it would overheat very quickly and use emergency throttling back to 4x all the time to stop from reaching critical temperature of 95C. When I forced that off it would often shut down from reaching 95C. Bad for gaming.

When I ran the Prime95 Torture Test on full load at my undervolt this morning the temperature barely goes above 82-83C, which is a massive difference! Now I don't have to worry about overheating! I think it's because I've finally dropped below the threshold where Heat Produced < Heat Notebook Cooling Can Physically Get Rid Of When the processor is idle even at 10x after load it quickly drops to 54C or so, at 4x after load it will kick back to idle at 48C. It's a hot summer's day now (unusually REALLY HOT for the UK!) so when I get home later I'll do more extensive temperature testing now it's warmed up.

I do need to run more extensive stability testing. Is it OK to leave Prime95 running overnight, or could that be unsafe? I mean I'd just wake up in the morning and see errors or no errors right?

I'm really glad I found this thread, it's basically solved the biggest problem I was having with my notebook

 

Thanks for the report. It seems to me that the introduction of MT and ML versions was inadequate of AMD's part. I have ML-34 and use 0.925 at 4x and 1,2 at 9x. Many people reported better results, so the MT class really loses its advantage.

There is no problem leaving Prime95 overnight. It is even recommended in order to be sure about the stability of the new voltage.

 

Well the 1029 ONLY supports MTs (I think I'm actually using the fastest processor MSI officially support with this chassis), so I'm guessing the cooling isn't good enough to support MLs, which would also explain my heat issues. But all is good now, and thanks for your advice on Prime95!

 

The MSI MS-1029 also has a decent GPU in it, but those ATI GPUs do run rather hot. Still, I have to wonder if whoever assembled the notebook didn't mount the heatsinks and apply the thermal compound correctly. So long as the cooling fan is spinning there's no way a MT series CPU should be getting anywhere near that hot.

 

It's a possibility, it used to be even worse, I sent it back twice.

But it's insanely hard to prove.

 

Hello,
I have a BenQ Joybook with AMD Turion X2 TL-50 + Ati integrated 1100.
I tried undervolting using RM clock version 2.1 but there are only 3 VID available for it (1.125 , 1.100 , and 0.075) despite of the fact that my CPU has following default voltage:
@FID 4x = 0.800 (min)
@FID 8x = 1.125 (MAX)
Am I missing something ???(pls bear with me being a complete newbie in this kind of adjustment)

On the contrary using crystal CPUID, I managed to bring down VID to 0.900 at FID 8x ( still torturing with prime 95 right now).
But CPUID cannot lower my VID at FID 4x being 0.800 the minimum voltage available.

How to make CPUID to automatically load at start-up? (very detail explanation would be really appreciated as I am really computer noob)

 

Add a shortcut to CrystalCPUID to your Startup folder (right-click and drag the CrystalCPUID program file from Windows Explorer to Start | All Programs | Startup, release the mouse button, and select Create Shortcut Here from the context menu). Do Properties on that shortcut and append /CQ /HIDE to the Target entry.

I get the impression that TX2's don't undervolt very well but not enough people have them yet to know for certain. I didn't even know about the BenQ Joybook you have.

 

MSI-1029 is known to have heating problem. That's why they use MT instead of ML. My V2000Z ML-37 not even goes above 60C after undervolting. Fan starts to tick at 58C and cool down in 2 minutes --- at full speed. In 4x mode, I don't even notice the fan start or not.

Really a good try. The 4x--0.8v is possibly fixed by AMD. That is bad. But 0.9V at 8x is really a great news. Fans never get below 0.96v at 1.6G when they tried on Intel. This means AMD is still wining their fans.

RMClock seems having some problems with v2.1. I install it on my desktop, and it mess up my old setting. Not even showing POD profile. I have to manually set it up.

 

Thank you for all the great tips.
So far I only used prime 95 for stress test & according to amd dashboard meter, only one core is on full load

I wonder if my undervolting result is solid. at 0.9VID, it did pass prime for over an hour. I also tried at 0.875 for about 10 minutes w/o any problem but when I went down to 0.85, windows immediately froze and no button was working including power button. I had to unplug the power to shut down.

Crystal cpuid correctly identify my cpu whereas RMClock not



BenQ Joybook P41 (turion x2 + radeon xpress 1100) and Joybook S73 (Intel core duo + radeon 1600) are based on Mitac ODM

 

Is there any other working temperature monitoring program?
As you can see from my above post, AMD dashboard did not show any.
Going down from 1.125 to .0900 VID, I did noticed the fan spinned slower (always spinning)

 

One hour test is the minimum of stable testing. Usually I ran 4~6 hrs just for one FID/VID pair. It is very time consuming, but it worthy.

Your result is really sound. 0.875V means you are working on 18.7 W TPD at full speed. Wow.

It a little bit strange to me that your RMClock recognize your Turion X2 as Opteron.

The Prim95 is single thread test like other benchmarks. Your second CPU load is your general OS load. Prim95 not even let you start two threads/tasks. If you really want to test two cores, use Linux version.

 

That is probably a bug. When I tested my ML-37, the temp was there.

If you are in full speed, it could be spinning 2 mins and stop 2 mins.

 

Hey Lonthong, I think it is necessary to start a new thread about dual core undervolt discussion. You are the first one with the result. What do you think?

 

I thought I just documented my experience in the hope that in case I made a wrong step, you guys can correct it and hopefully this can be helpfull to people with little to no in depth knowledge of undervolting technique

Download underdervolting: crystalcpuid or RM Clock
For my case, I use crystal cpuid since the latter does not seem to work to my expectation (or maybe I do not know how to make it work)
stress testing program: prime95

Install prime95
Extract crystal cpuid to a location of your choice (no installation is required)

Go to start - control panel - power options.
On power schemes tab, change power schemes to home/office desk, click apply and OK

run prime 95, you will see red icon on yr system tray (which means program is running)
go to location where you extracted crystal cpuid and double click crystal cpuid.exe (chip icon)
you will see this:

click on function and AMD K6/K7/K8 multiplier, you will see this

Check if yr multiplier is at max ( my max is 8.0 x)
Click on enable change voltage and start yr undervolting experience by decreasing new voltage. stop immediately if you noticed the prime95 icon changed color to yellow or you experience windows freezing/lock-up
Remember the last stable voltage

Restart yr computer and double click crystalcpuid again
Redo above procedures for min multiplier. But in my case - min multiplier = 4.0 x is already at the very minimum voltage (=0.800) so I can't reduce any further.
On crystalcpuid windows, Click on file - multiplier management setting

click on enable voltage & put yr new stable undervolting value on its corresponding position (I got 0.875 stable for about 15 minutes and to be and the safe side, I put 0.900).
I am too lazy to find out the undervolting value for middle FID, so I just interpolated the value of new max and min.
For my case
FID = 8.0x VID = 0.900
FID = 6.0x VID = 0.850
FID = 4.0x VID = 0.800
click on apply and then OK

Add a shortcut to CrystalCPUID to your Startup folder (right-click and drag the CrystalCPUID program file from Windows Explorer to Start | All Programs | Startup, release the mouse button, and select Create Shortcut Here from the context menu). Do Properties on that shortcut and append /CQ /HIDE to the Target entry.

 

That would be great.
In which thread should we put it???

 

Try Core Temp, it worked quite accurately for me...
http://www.thecoolest.zerobrains.com/forums/viewtopic.php?t=137

You can create a new thread right here in the HP/Compaq forums or in the Software section.

 

Core Temp works on my dv8000z too, showing the fan turning on at 55C and shutting down below 49C. Thanks for the link miner!

 

For those Linux user, here is how you can undervolt your AMD turion.