Undervolted my 7811

Well, I 'think' I did this all correctly after reading the 'undervolting' guide posted elsewhere in this forum, but so far it seems to be ok.

If anyone has any experience doing this, maybe you can comment on whether or not this is worth it and/or whether I did everything correctly.

I followed the guide and recorded the high temps before (59) and then started stepping down the voltage for the max multiplier (8x).

It started out at 1.175 and I ended up with the lowest voltage available of 1.0125, which also matches the lowest multiplier of 6x which I understand is the idle mode.

I then ran Orthos for an hour and then again for another 45 minutes and all worked well - no hiccups. The highest temp recorded was 50.

If I understood everything correctly, I'm running maximum performance at minimum voltage. I haven't gotten an idea of battery life yet as I just started with that.

I know the guide says it helps extend battery and component life, but this doesn't seem much of a difference does it? And is this all worth it?

 

9 degrees C is a huge difference. Plus you'll save a good amount of battery life. Run ORTHOS for 4 hours though minimum, I've had systems crash after 3 hours 50 minutes before!

 

i ended up with a minimum of 0.900v don't u have this option?..temp only dropped to 53

 

The Right Mark utility did not have an option lower than that. This also made me wonder if I could go lower since it was running ok at that voltage, but since there was no option I just figured that was that.

I'll do the longer stress test this evening as suggested just to make sure it is really stable.

 

If you uncheck the "automatically manage intermediate P-states" or whatever and restart rightmark, you should be able to lower it even more. Pretty sure core 2 duos have a minimum of .95 volts...

 

I believe if you check the mobile box instead of the desktop you will find the lower voltage.

This unit runs fairly cool at idle so you will see more benefit under stress than at idle.

 

Thanks all, will work on that tonight. Good source of knowledge here!

 

Seen that you can go as low as 0.925 with P-7811, lowered mine to 0.95 and has been running fine since then. Temp drop I got is approx. 10 C.

 

really? can we see that in cpuz?
says the lowest it can go is this
http://ark.intel.com/cpu.aspx?groupID=35569

just curious is all.

 

and as this picture shows...it can only undervolt to 1.013
cpuz reports your true voltage, while rightmark is what you can shoot for....

 

there was discussion before about cpuz vid readings, if i remember since a certain version(1.41) they do not detect mobile cpu vid correctly. anyway try core temp 0.99.3, it will give same reading as rmclock while everest will give same reading as cpuz(1.42+). pick your side

 

Hmm, seems like the CPU-Z shows the 1.013V also for me even though its set in RMClock for 0.95V. Oh well, there is definitive temp drop neverteless, so I'm happy with it as it is...

 

Any version higher than v1.41 will show incorrect voltages for Merom/Penryn cpu's. It shows desktop equivallent temperatures instead.

For Yonah cpus and older, any version of CPU-z will work consistent.

 

I suppose you mean "desktop equivalent voltages?"

 

Prolly. Mine's stable to the listed value in my sig. I love my T9300. :3

 

all very interesting... i just checked my t9300 and the lowest it will go is 1.063, even though rightmark says it will go lower. i checked it using a few programs and they all agree...it will not undevolt to anything lower than that.

so can we see a cpuz picture please....

edit: and when running orthos..it needs to be set to small fft's to super stress your cpu.

 

You should list all of your final voltage settings after full stability tests. And yes, super stress it with Small FFT's.

 

This is indeed correct and the #1 reason I'm still running 1.41.

 

well now that explains everything!

my apologies to steatharcher

i just downloaded hwinfo32 and this better explains just what your system is suppose to be able to do. and the information seems to be a bit more accurate.

it would seem that an x9000 can drop as low as .0925

 

John, hwinfo32 is a very detailed program. Thanks. I'd never heard of it before. Amazing what all it reads, even video bios version.

 

your welcome... but.....nah, thank dook. he be making me dig deeper for answers and reasoning. and i think someone also made mentioning of this software as well....

and im checking on some new programs...we need to have a handle on just what is "suppose to" be the real temps & voltages.

 

Shoot, no need to thank me. I've learned more than my fair share from you! If I claim something, you're right there making sure I'm in check and spewing the right info! We kinda compliment each other.

 

tis is true.

and on the flip side.

im just a nosey "B" when it comes to technology. *LOL*

edit: 1.42 works and it shows the correct revision code as well.

 

Any major difference from 1.41?

 

nope. just the revision shows on the opening screen, instead of having to see it on the 3rd tab

 

i couldnt install rmclock, it said that it couldnt install or load the driver because my version of windows requires the drivers to be signed, what a load of ****

 

What version of Windows are you using? I'm pretty sure RMClock is not Vista X64 compatible. Try booting with F8 options and turn off the requirement for signed drivers, I forget which option it is specifically.

 

FACK!!!!!!!!!!!


im using vista ultimate x64.

 

I just checked - it's called "Disable Driver Signature Enforcement"
Choose that option and also run the installer as administrator. Good luck!

 

kk did, thanks. now will that stay disabled or do i have to do that every boot?

 

It's a one time thing, but you just need that to install drivers. Once the unsigned drivers are installed, Windows will continue to use them.

 

oh can someone give me a copy of 1.42 i have 1.47 and i dont wanna be left out

 

has all the cpuz versions
http://www.filehippo.com/download_cpuz/3486/

 

when i restarted it said it couldnt load them, same thing again except it took "install" out.





cpuz version 1.42 says i have a T8300 but 1.47 says T8400

 

you may have to try a few versions in there. mine reads my extreme chip and gives the correct revision where as the another one didn't give the revision code at all.

i tried them all and 1.42 seems to be right for me...

 

Did the software work prior to install? There may be a way to permanently disable it, I will have to check.

 

im running orthos right now, btu i notice that in Rightmark, my cpu usage never goes above 94.12% it went to 100% before the undervolt, whats stopping it from using all 100% now?

 

try restarting ORTHOs... you might be running at a lower power state now, and ORTHOs still has the old reading. So even though you're maxed out for your power state, ORTHOs is expecting the power state above. Have you disabled any multipliers in RMClock yet? Either way - restart ORTHOs and if that fails exit RMClock.

 

check your throttle settings....

 

RMClock will not allow your P8400 CPU to use the top .5 multiple so you will be slightly underclocked. This is not the end of the world and the temp advantage may outweigh the clock disadvantage.

You may want to read the Undervolting Mother Load found here:
http://forum.notebookreview.com/showthread.php?t=235824

 

this look right?

 

Looks beautiful.

 

fixed, the attachment didnt work for some reason. photobuckets a nice backup

 

Yea. Photobucket acts a fool sometimes.

I agree with Syn(as usual), don't worry about the .5 multiplier. You're not gonna notice a difference.

 

I personally would use RMclock for battery to reduce heat and improve battery life, and switch it off when on AC to get the extra performance. You wouldn't be using the max power state on battery anyway, so the .5 doesn't make a difference.

 

did anyone find a fix for when i load up windows that it doesnt allow the driver for RMclock to load. its rather annoying to have to press F8 everytime i restart my comp

 

Yes, here are the steps:

1. Open an elevated command prompt
2. type “bcdedit /set loadoptions DDISABLE_INTEGRITY_CHECKS” without the quotes (and no the DD is not a typo).
3. Reboot and enjoy being able to use unsigned drivers in Vista x64.

Credits to source: http://chris123nt.com/2007/04/28/driver-signing-in-x64/

 

question....have you tried enabling all of rmclocks functions?
i see you have dffs enabled. was wondering about ida...

mine doesn't have either of them settings for a x9000, but did for my t9500

 

"bcedit" is not recognized




EDIT: read it wrong sorry, says operation completed successfully. lets hope it works, will do a reboot ina few