The "Undervolting" Guide

Yes error in orthos means BSOD or system freeze is next.

Did you tick the 13x index in the sub-profile page aswell?

 

I've got a t9300 as well, and recently went through this.

You need to tick the IDA 13x box under p-state transitions, and ensure that it is enable on the profile page. You don't need to have IDA enagaged, and it won't be used regardless. The 13x multiplier needs to be enabled however in order for RMClock to allow your CPU to hit full speed.

You will not lose any performance whatsoever as a result of undervolting. Undervolting does not cause anything to run slower.

Error messages in Orthos mean your processor is not stable at that voltage - raise it up a bit, let it run again. Other potential things that can happen are crashes, freezes, spontaneous reboots, etc if the voltage is too low. This won't cause any harm, so no need to worry, just raise the voltage a bit and try again.

 

Yeah haha that fixed it. Mind going over my question in the previous post that I edited? The voltages do seem... rather low.

 

Check on the CPU info page.

You'll likely find that your processor is actually running at a higher voltage. In my case, my processor won't go below 1v. I can set it far below in RMClock, but it never actually goes below 1v.

This is probably - though not necessarily - the case for you too. At 10x less than 1v would be pretty low.

Still, if it works at a given multiple/voltage combination, it'll work at that voltage for all lower multipliers. You never need a higher voltage for a lower multiplier.

 

Well, during Orthos testing, the current VID is indeed at 0.9250V.

 

The voltage cap might only be for your system wintersdark

Yes, its should be. I suggest testing the other multipliers anyway if you have the spare time, just to be 100% sure. Maybe just for 15-20mins.

Also you should put the a 1-2 step voltage margin for the 13x multiplier to definately ensure that your system is rock stable at full load.

You are one of the very few who are lucky enough to use the lowest voltage for all multipliers.

 

Well here are the final VIDs:

8x - 0.9250V
9x - 0.9250V
10x - 0.9250V
11x - 0.9375V
12x - 0.9750V
13x - 1.0000V

Do you recommend raising the final 3 multiplers up a notch? I'll be doing an extended stability testing tomorrow regardless.

 

Yeah, I know voltage caps and overall limits differ from system to system and chip to chip. Just thought I'd point that out as it's actually pretty easy to think you're running at a lower voltage than you actually are, due to the voltage cap, as the VID editing screen will allow you to set them lower than they'll actually run.

Damn, that's fantastic! Good luck with the extended test =)

 

Out of curiosity, is there any real advantage to a 3-4 hour test on each voltage to test stability? I could stick to my current test results and see if daily usages causes system hang/crash.

 

Yeah you could do that if you like, but theres no 100% guarantee. Just do a 15min test on all the other multipliers, since these multipliers dont get used much and even when they do ,its only for a short time.

Testing it in real time use is also a good way to test. If it doesnt crash after 1 weeks use then the voltages must be stable. Doing the stability test just ensures all the multipliers are tested and ensured.

The 2 mainly used multipliers is 6x (idle) and 13x (max). So you only really need to test the 13x since idle will always be stable. You only need to do a 3 hour test for this multiplier, not all.

 

Yeah, it's just that from reading all these threads, it seems if a BSOD were to occur, it would have done so in the first 10 or so minutes of testing already. I have yet to see anyone get an error 40 minutes into his or her test.

On the contrary, I faced a rather strange error today; I'm not sure if it resulted from undervolting, but I don't think so. Anyway, when I booted up the first time, I waited for Vista's welcome screen to load -- it never did. It shut off instead. This was odd, I thought. I pressed the power button again, and his time the computer told me there was no hard disk detected and it did not have a place to boot from. It suggested that I run a Dell diagnostics test, which I did, that also confirmed there was no hard drive intact. I aborted the test and on the next restart, the hard disk was once again detected and I was given the option to boot Windows in safe mode/normally etc etc. Are the cables loose in my laptop or what?

 

My battery life actually decreased when I undervolted. I'm on 0.9500V on a 6x multiplier using a T7200 (2.0Ghz). I went from having around 1h30m of battery life down to around 56 minutes somehow. I put the OS on power saver. I'm going to try to figure this out on my own but if I don't a response would be sweet

Edit: Nevermind it seems like my whole battery is dying. Weird.

 

I would never have seen 2 hours with max screen brightness if it were not for Undervolting

 

I was going to try and undervolt my x9000 cpu, but orthos wont even run for more than 3 seconds and I get this error.
>.> I have not had any bsod or blue screens at all with this laptop and I haven't changed anything yet. This was just the first run with orthos to see what my temps where.

Type: Blend - stress CPU and RAM Min: 8 Max: 4096 InPlace: No Mem: 1791 Time: 15
CPU: 3191MHz FSB: 0MHz [0MHz x 0.0 est.]
CPU: 3191MHz FSB: 0MHz [0MHz x 0.0 est.]
6/20/2008 1:45 AM
Launching 2 threads...
Using CPU #0
Beginning a continuous self-test to check your computer.
Press Stop to end this test.
Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 0 minutes 4 seconds - 1 errors, 0 warnings.
Execution halted.

 

How much voltage were you trying to drop at the first test? Can you post up the default voltage settings and the settings you tried?

 

As I said this test was to check temps, I haven't changed anything yet. I figured it out though, the test wont run in overclocked mode for my cpu. Changing it back to stock in the bios makes the test run fine. =/ guese that means no undervolting for me then if it wont run even at the stock voltages.

 

I see, now that you mentioned about overclocked.

Yeah not every case is possible with overclock/undervolt combined.

 

but thats just it, I didn't adjust anything and orthos is the first program that wouldn't run, it actually stopped itself from running. Everything else runs perfectly fine, so I think the bios settings are already configured to run at the lowest voltages possible.

 

Try a different stress program like intel TAT. ORTHOS might not be recognising the OC mutipliers, causing an error.

We have had several x9000's undervolted sucessfully. Most of them were Gateways.

Heres one http://www.imagebam.com/image/d36d757680515

 

Penryns are very good with UVing and temps . As Flip said , try TAT, Orthos is good, but not the best yet.I`ve had problems with my T2300 on it, it would overheat it in a matter of seconds,which no other stress tool managed.

 

does anyone experience this?

when i leave my notebook on IDLE and on high performance mode, my cpu clock and throttle activity is erratic and the multiplier constantly jumps from 6x to IDA.

when i'm on battery mode or office settings, the cpu clock and throttle activity stays consistently low and only jumps when the cpu load increases.

is this normal?

 

go to the Advanced CPU options and untick "Engage IDA" > Apply

 

thanks i'll try it out

 

i think i figured out what was causing the random cpu spikes but not exactly sure what the real reason is.

the power4gear high performance profile for some reason kept causing my cpu to randomly spike and appear unstable. i checked the settings for that profile and the cpu is enabled to 100% and set for high cpu performance.

if i use the power4gear office profile and change the cpu settings to make them identical to the high performance profile i DO NOT get the random cpu spikes.

also if i use the vista high performance profile no random cpu spikes occur.

 

It is a great guide! Thanks!
I really enjoyed the proccess and managed to cool cpu down from 86C to 74C degrees under stress. And to lower voltage from 1.250 to 1.012 for the highest multiplier(10x). It's a big temp drop for my laptop which felt like frying pan before...
The biggest question now is how I can make the RMClock to keep the right voltages after I reboot?
Every time I reboot the fields in Profile become blank and gray, so I have to do the "ticking" over and over...
Ï'd appreciate any help on this....

 

In 'Advanced CPU settings' make sure 'Apply this settings on start up' is ticked. In 'Profiles' select both 'Current' and 'StartUp' profiles. In 'Settings' tick both 'Start Minimized to system tray' and 'Run at Windows StartUp' (Via StartUp RegKey). Don't forget to hit 'Apply' after every change(or when you done it all).

 

Ok, got it now.
The problem was in me not selecting ''performance on demand" for Startup profile...Everything seems fine now.
Thanks a lot for a promt answer!

 

I've found I get inconsistent results if I don't click apply after every change. Not every screenful, but each individual change, particularly in regards to VID settings. It was a huge problem for me for a few days, actually, because settings wouldn't "stick" properly unless I did that. I'm not sure why it's like that, and it doesn't seem to apply to everyone, but it's something to keep in mind.

Also, I find right clicking the system tray's RMClock icon and selecting Restart RMClock ensures settings get saved properly helps a lot. Without doing that, sometimes settings don't seem to actually get saved.

YMMV of course, but at least for me rmclock has been a very twitchy program. Extremely useful, but tempermental.

I'm hoping they come out with a version with better inherent support for the T9300 and it's .5x multipliers. Should such a day arrive, I'd buy it in a heartbeat.

 

so who has the coolest running T9300 penryn now? reading through some of the earlier pages, i saw people seeing low to mid 60's C before undervolting....i never saw 60s even without undervolting...barely saw mid 50's.

 

quick question...i have slfm and IDA enabled for this stress test so i have full load for cpu/os, but when hovering the mouse over rmclock in the system tray, it still says the multiplier is 12x, but with the voltage i set for the IDA (which should be 13x) any ideas on how to correct this?

 

You have a 12.5x multiplier, thats why its using 13x voltage . It will only display 13x when IDA actually engages

 

Will this guide work equally well if I set the profile to "power saving" rather than "performance on demand"? I'm planning to have my notebook go to power saving mode on battery. Will I still have good results?

Also, if I'm running RM clock and set Vista to power saving mode, will RMclock automatically kick in and undervolt or will I have to activate power saving mode in RMclock as well?

 

sweet. thats what i was hoping was up with it.

 

Yes the profiles wont matter. Power saving profile will just underclock the CPU to 1 multiplier (usually set to lowest one or which ever speed you choose). Just google around with it and see which gives you better results.

RMclock will take over Vistas 'CPU' power management regardless of profile. You can also setup RMclock to take over the whole power management.

 

its interesting to me that while running orthos, either with or without IDA off, rmclock doesnt report anything higher than 2493 mhz. i thought enabling the IDA multiplier but unchecking the "engage IDA" box would enable the 12.5x multiplier...

 

Your 12.5x multiplier is 2.5ghz

200mhz FSB x 12.5 = 2500mhz

If you untick the 13x IDA multiplier your max speed will only be 2.4ghz

 

*whistles casually* then whats up with this?

 

wow, thats new discovery. Can you restart RMclock just to make sure? Also open CPU-z

Other users reported that the 12x will not use the 12.5x multiplier. Thats why i recommended to use the 13x with IDA disengaged.

 

let me finish the stress test and if you can link me to a valid cpu-z 1.41 download....

 

Heres all the CPU-z versions. http://www.filehippo.com/download_cpuz/

It probably wont matter. Its only the voltage thats innacurate on the latest version. Try the latest version first for penryn support

 

interesting...on the left side is cpu-z 1.41, and cpu-z 1.45 on the right.

 

Nothing seems to be consistent.

Bummer.

 

well...thats great news, i suppose. if i wanted to be limited to 2.4 ghz i would have bought a T8300...guess its time to check the IDA box and run the stress test again. *rolls eyes*

you think maybe RMclock reports the rated speed of the processor when using its highest multiplier even though its technically not running at that speed?

 

RMclock always shows the active core clock down to the .1hz

Run PCmark cpu test, only way to surely find out.

 

pcmark 05 shows the processor as being a T9300 2.5ghz @ 2494 mhz with or without IDA checked

 

ahh...screw it. ill deal with this IDA stuff some other time. as of now though...really really nice idle temps. 17C and 16C for the processor. a max of 53C using orthos.

 

Hmm these temps are most likely reported incorrectly..even penryns cannot go this low...

 

slfm + laptop cooler = those temps.

 

Still I don't think it's possible...but this is just my opinion..

I used to get even lower temps(10-12ºC) but it was a BIOS issue.

 

Im a little sceptical aswell but it is possible. Your thermal zone temps are consistent.

The lowest i can achieve is 28c, and this is a fully modified notebook. Massive cooling fan, AS5, undervolted, gpu copper mod, etc)

Try using core temp and speedfan. See if they show the same temps aswell.