PDA

View Full Version : Undervolting G750



Torpedo499
07-06-2014, 11:48 PM
So I'm planning on undervolting my CPU, and I plan on using a guide on these forums, my question is though, where do I stop undervolting? Do I stop when I get a blue screen? or does it act funny? I don't want to damage anything and I really nervous, but I'd love to have a decrease in temperature.

Also, people undervolt by different increments right? I hear that some people can do -50 mV, -70mV, and even -125mV.

Do I just keep going until something happens? Also should I have the G750 plugged in and running at one performance mode?

Thank you, any tips appreciated!

hmscott
07-07-2014, 12:11 AM
So I'm planning on undervolting my CPU, and I plan on using a guide on these forums, my question is though, where do I stop undervolting? Do I stop when I get a blue screen? or does it act funny? I don't want to damage anything and I really nervous, but I'd love to have a decrease in temperature.
Also, people undervolt by different increments right? I hear that some people can do -50 mV, -70mV, and even -125mV.
Do I just keep going until something happens? Also should I have the G750 plugged in and running at one performance mode?
Thank you, any tips appreciated!

Torpedo499, the XTU settings are independent of the Power Plan settings, but you probably want to test under High Performance, CPU max/min 100%/100% just so that the test is under maximum stress. But, often the undervolt failure doesn't happen under load, it happens at idle - and can take hours between events.

The individual CPU ability to undervolt is the limiting factor, so it is hard to say where to start with the undervolt testing. It could fail anywhere between -20mV and -125mV.

You have to decide how fast you want to reach failure, and how long you want to try steps/increments of change. Starting high until failing, halving that value, then increasing again by half, decreasing by half, etc - until you converge on a stable value - under load and idle.

Unlike the GPU tuning failures that often display artifacts around the stable point, so you don't have to actually crash the driver to find the optimal tuning spot, the CPU / CPU_cache undervolt either works, or crashes - blue screen, black screen, power off / restart. That is why you don't want to have any app/file open you don't mind losing - so far I haven't had to reinstall an app or fix a disk, but it is possible.

The good news is you usually only have to crash once or twice to converge on a good undervolt, a few times if you want to be persistent in getting the most undervolt possible.

Plugged in is fine, I haven't done it on battery, but I would assume if you set the Power Plan on battery to High Performance, 100%/100% CPU min/max, it will work on battery too.

My JH is now stable at -40mV, it started out new stable at -20mV, and I have slowly increased the undervolt since 11/2013 by -5mV every few months :)

Let us know what value is stable for you :)

Torpedo499
07-07-2014, 12:31 AM
Just did -70mV and passed the stress test for the Dynamic CPU voltage offset, should I also reduced the processor cache voltage offset to -70mV?

If it is stable, i think I will go a bit more, what do you think is "safe"?

Torpedo499
07-07-2014, 12:41 AM
The good news is you usually only have to crash once or twice to converge on a good undervolt, a few times if you want to be persistent in getting the most undervolt possible.


Let us know what value is stable for you :)

Just undervolted to -90mV, should I be worried that my CPU isn't getting enough power?

hmscott
07-07-2014, 02:44 AM
Just did -70mV and passed the stress test for the Dynamic CPU voltage offset, should I also reduced the processor cache voltage offset to -70mV?

If it is stable, i think I will go a bit more, what do you think is "safe"?


Just undervolted to -90mV, should I be worried that my CPU isn't getting enough power?

Torpedo499, current is the "food" that does the work, voltage is the pressure, and unless you can put your ear to the laptop and hear the CPU wheezing, it will be fine ;)

You want to do both the CPU and cache offsets the same, for now.

I was able to go all the way up to -125mV on my JW, and it was stable, the JX was stable at -70mV, and the JH started at -20mV stable.

Each CPU is different, increase gradually until you experience a problem, then drop back to the previous stable setting.

Run some CPU benchmarks and stress tests. Like Intel XTU stress test - built-in to XTU, prime95, AIDA64, etc - and you don't have to run it long, maybe 15 minutes. Then after you are done working on your G750, let it sit idle for a few hours, and if it doesn't reboot or shutdown unexpectedly you found a stable voltage offset setting for Dynamic CPU Voltage offset and CPU Cache Voltage Offset.

-90mV is a good setting, a couple of people recently that started there ended up at -70mV, just an FYI toward expectations.

Let us know how it works out. :)

Torpedo499
07-07-2014, 03:37 AM
I was able to go all the way up to -125mV on my JW,


-90mV is a good setting, a couple of people recently that started there ended up at -70mV, just an FYI toward expectations.

Let us know how it works out. :)

I am happy to say that I did not crash and have a blue screen at -100mV, anything above that did however cause a blue screen. At least the Window 8.1 blue screen doesn't look as scary as the Windows 7 one! I had 2 blue screens!

I will be doing further testing tomorrow, such as benchmarks, gaming, and general CPU usage. Thank you for your help, you got me through something I was very afraid of doing!

I will post further results later to say how everything goes!

Darnassus
07-07-2014, 04:06 AM
Torpedo499, current is the food, voltage is the pressure, and unless you can put your ear to the laptop and hear the CPU wheezing, it will be fine ;)

You want to do both the CPU and cache offsets the same, for now.

I was able to go all the way up to -125mV on my JW, and it was stable, the JX was stable at -70mV, and the JH started at -20mV stable.

Each CPU is different, increase gradually until you experience a problem, then drop back to the previous stable setting.

Run some CPU benchmarks and stress tests. Like Intel XTU stress test - built-in to XTU, prime95, AIDA64, etc - and you don't have to run it long, maybe 15 minutes. Then after you are done working on your G750, let it sit idle for a few hours, and if it doesn't reboot or shutdown unexpectedly you found a stable voltage offset setting for Dynamic CPU Voltage offset and CPU Cache Voltage Offset.

-90mV is a good setting, a couple of people recently that started there ended up at -70mV, just an FYI toward expectations.

Let us know how it works out. :)

Don't you mean voltage is the food, and current is the pressure? ':x

hmscott
07-07-2014, 04:08 AM
Don't you mean voltage is the food, and current is the pressure? ':x

It depends on what you are hungry for :)

How Electricity Works
http://science.howstuffworks.com/electricity7.htm

"Let's say you have a tank of pressurized water connected to a hose that you're using to water the garden. If you increase the pressure in the tank, more water comes out of the hose, right? The same is true of an electrical system: Increasing the voltage will result in greater current flow."

Voltage - From Wikipedia, the free encyclopedia
http://en.wikipedia.org/wiki/Voltage

"Voltage, electrical potential difference, electric tension or electric pressure (denoted ∆V and measured in units of electric potential: volts, or joules per coulomb) is the electric potential difference between two points, or the difference in electric potential energy of a unit charge transported between two points."

Electricity - Voltage and Current
http://www.reprise.com/host/electricity/voltage.asp

"Who Does the Work?
Current, not Voltage, does the work in electrical circuits. The flow of water through a turbine is what makes the turbine spin. The flow of current through an electrical circuit is what lights the bulb, heats the stove, runs the motor, etc. Routing and controlling the flow of current is the goal of every electrical circuit. "

Torpedo499
07-07-2014, 09:41 PM
Undervolt not stable?

I was in the middle of a game, running fine, then all of a sudden my computer freeze, and I hear a weird "vrrmmm" sounds in my earbuds.

My Undervolt was reset back to default, so I guess that was the problem, I will now try at -90mV :(

hmscott
07-07-2014, 09:58 PM
Undervolt not stable?

I was in the middle of a game, running fine, then all of a sudden my computer freeze, and I hear a weird "vrrmmm" sounds in my earbuds.

My Undervolt was reset back to default, so I guess that was the problem, I will now try at -90mV :(

Torpedo499, yup, that sounds like a symptom :)

Torpedo499
07-07-2014, 10:09 PM
Torpedo499, yup, that sounds like a symptom :)

Do you think it damaged anything? and will 10mV make a difference in temperature?

hmscott
07-08-2014, 01:55 AM
Torpedo499, the only thing that can get damaged for such minor changes, is during the crash if you had open any files they might be corrupted - but Windows is pretty good about not allowing that these days.

If it makes you feel any better you can right click on the volumes you had open files on at the crash time and do a error check. The C drive may ask to do a check at the next boot, say yes, reboot and it will fix any problems - you can do that after every crash if you want to be 100% covered.

I haven't lost any files - for a long time -, and I haven't had anything found to fix in a long time.

The only thing to watch out for is if you have never done an integrity check on that volume, it might find stuff already there - before the OC'ing. I had that happen once a long time ago, it wasn't fun :(

So do error checking on your disks before going any further, I will feel better :)

Yes, -10mV will make a difference, but so will -9mV - you can get finely tuned if you want to spend the time - usually the fall off is only -1mV or -2mV, and that may be why I have been able to get -40mV over time, starting at -20mV stable - going up -1mV at a time...

-90mV is exceptional, most are between -50mV and -70mV :)

ZiggZagg
07-08-2014, 03:47 AM
Hello Mr. hmscott. I've been following this thread because I've been trying to get my GPU temp down... not that it's too high (80c while playing Crysis 3 and Metro for hours both on ULTRA). I remembered my wife had and old aluminum convection cooling stand (no fans) for an old lapper and I dug it out. I installed (taped in place :p) a dual fan setup that is powered from one 3.0 USB port on my JX. Temp dropped 5c to 75c. The only thing I tune-up on the 770m graphics card I have is setting it to the "G" configuration which I suppose is the game safe setting according to the 'GPU Tweak' program. I've had no problems with heat I was just curious to see if I could cool it off you know.

As far as under-volting goes I've had it set to -75.23586.... whatever mili-volts since about the third week I had the computer back in November 13' and it hasn't done one weird thing no matter what the cause could be. I guess I'm lucky as far as that goes.... stretching it down to -75mv huh? The only thing I'm really nervous to do is upgrade to 8.1. I know everything can be straightened out afterward but I figure why screw up a good set up :cool:. Anyway thanks for all the info and tips in this thread. SLFN.

ZZ

hmscott
07-08-2014, 04:00 AM
Hello Mr. hmscott. I've been following this thread because I've been trying to get my GPU temp down... not that it's too high (80c while playing Crysis 3 and Metro for hours). I remembered my wife had and old aluminum convection cooling stand (no fans) for an old lapper and I dug it out. I installed (taped in place :p) a dual fan setup that is powered from one 3.0 USB port on my JX. Temp dropped 5c to 75c. The only thing I tune-up on the 770m graphics card I have is setting it to the "G" configuration which I suppose is the game safe setting according to the 'GPU Tweak' program. I've had no problems with heat I was just curious to see if I could cool it off you know.

As far as under-volting goes I've had it set to -75.23586.... whatever mili-volts since about the third week I had the computer back in November 13' and it hasn't done one weird thing no matter what the cause could be. I guess I'm lucky as far as that goes.... stretching it down to -75mv huh? The only thing I'm really nervous to do is upgrade to 8.1. I know everything can be straightened out afterward but I figure why screw up a good set up :cool:. Anyway thanks for all the info and tips in this thread. SLFN.

ZZ

Back at cha' - been reading your helpful posts over time, tnx :)

Nice idea on the fans / large cool heat sink to preload cool air. It is rare to find a cooling pad that does anything - every laptop venting is different, so I have been thinking about this for a while. Custom fitted adapters for each laptop to effectively route the air and exhaust venting assist. Could really make a difference - adding a refrigerated air assist, that would be sweet :)

-75mV is a good undervolt, and since you have been running it a while, maybe it is time to see if a -80/-85mV will work now :)

I am testing -50mV - running FAH CPU / GPU @ 6300mhz, and has been stable since yesterday!

It is tough to venture past the comfort zone, when everything is working well. That is why I introduce another boot drive for experimenting, and set aside the good config to keep it safe.

I left my RAID0 as is, and did a Backtracker Restore to a 2nd bay SSD (pulling the RAID0 during the restore!) and did all the upgrades and testing on the secondary drive, before deciding I liked the new config and redid it on the primary RAID0.

As long as you know you aren't blowing away the good stuff and it is safe and ready to boot back any time you like, it makes it easier to take chances trying new stuff.

Windows 8.1 drivers are all there now, might as well give it shot - and doing an 8.0 upgrade to 8.1 works a much higher percentage of the time now :)

ZiggZagg
07-08-2014, 02:15 PM
Is there really that much advantage to upgrading to 8.1? As I said 8.0 has worked perfectly since I got the machine. When I installed my ssd a few days after the first power up I got rid of all the bloat at the same time and I (so far.... knock wood) haven't even had one BSOD. I've read where 8.1 is all put together nicely these days but I guess the first impression is hard to shake:confused:. I'll likely do it soon though.
Quick question...
I suppose I should try upping... or rather downing my mili-volts a bit more but will the config in XTU snap back to default if I go too far. That is, will I be able to reboot and fix things if I get a BSOD?
Thanks hmscott.

ZZ

hmscott
07-08-2014, 02:25 PM
Is there really that much advantage to upgrading to 8.1? As I said 8.0 has worked perfectly since I got the machine. When I installed my ssd a few days after the first power up I got rid of all the bloat at the same time and I (so far.... knock wood) haven't even had one BSOD. I've read where 8.1 is all put together nicely these days but I guess the first impression is hard to shake:confused:. I'll likely do it soon though.
Quick question...
I suppose I should try upping... or rather downing my mili-volts a bit more but will the config in XTU snap back to default if I go too far. That is, will I be able to reboot and fix things if I get a BSOD?
Thanks hmscott.
ZZ

ZZ, XTU will not apply it's profile to the CPU if you crashed since last boot, for any reason - even if it isn't due to XTU settings. XTU doesn't know why you crashed, but it assumes the safest thing to do is to reset to defaults at next boot. You need to start XTU and reapply the desired profile manually, or XTU will continue to not set anything at boot time.

8.0 to 8.1 is like any major dot release, it is different. Whether it is better, is relative.

The idea with upgrades, improvements, progress in general - is to make things better.

I can't say if it is better for you, but it is good for me :)

Torpedo499
07-08-2014, 05:17 PM
-100mV failed, and -90 mV failed, the -90mV crash wasn't as bad, but Its making me pretty pissed that it's showing that its stable during testing, but a couple hours into gaming it crashes. What next? Should I do -80mV and see how it goes, and if it goes good slowly increase until I find the crash point?

hmscott
07-08-2014, 08:08 PM
-100mV failed, and -90 mV failed, the -90mV crash wasn't as bad, but Its making me pretty pissed that it's showing that its stable during testing, but a couple hours into gaming it crashes. What next? Should I do -80mV and see how it goes, and if it goes good slowly increase until I find the crash point?

Torpedo499, sometimes it's about the games, and not about the OC'ing, so if I feel the frustration level correctly, try dropping down to -50mV for a few days and get some stability under your belt - not to mention games :)

Then when it occurs to you to return to OC'ing, take it back up to -70mV and try that.

It's all about the fun, keep it going :)

Torpedo499
07-08-2014, 08:55 PM
Then when it occurs to you to return to OC'ing, take it back up to -70mV and try that.



I'm not even OC'ing, I'm just straight up undervolting, I'll give it a try though, thanks!

ZiggZagg
07-08-2014, 11:43 PM
Mr hmscott... I downed the mili-volts to -90.8203125 and increased the active core multipliers to 35x-34x-33x-33x- respectively and the processor cache ratio to 35x. Started running Crysis 3 and played for two and a half hours straight with no problems;). I realized yeas ago that there are going to be subtle differences between chips cut from the same parent wafer but from what I'm reading there seem to be similar model chips that can't 'stretch' (for lack of a more accurate term) as far as others.... parent wafer not withstanding. Everything got good and warm while playing and still nothing snapped. This is nice!!.... maybe I'm doing something wrong. Not that I want my JX to go BSOD but I'm beginning to feel the need to see just how far I can push it:o. Oh just out of curiosity I ran an XTU benchmark at these setting and got 730 marks @ 3.39mhz and 94c. Is that reasonable?

ZZ

hmscott
07-09-2014, 10:10 AM
Mr hmscott... I downed the mili-volts to -90.8203125 and increased the active core multipliers to 35x-34x-33x-33x- respectively and the processor cache ratio to 35x. Started running Crysis 3 and played for two and a half hours straight with no problems;). I realized yeas ago that there are going to be subtle differences between chips cut from the same parent wafer but from what I'm reading there seem to be similar model chips that can't 'stretch' (for lack of a more accurate term) as far as others.... parent wafer not withstanding. Everything got good and warm while playing and still nothing snapped. This is nice!!.... maybe I'm doing something wrong. Not that I want my JX to go BSOD but I'm beginning to feel the need to see just how far I can push it:o. Oh just out of curiosity I ran an XTU benchmark at these setting and got 730 marks @ 3.39mhz and 94c. Is that reasonable?
ZZ

ZiggZagg, it is a good idea to get a baseline reading at default settings - without undervolt and OC changes to CPU - let the CPU / GPU cool (whole laptop) and do a couple of XTU benchmark runs to get a good feeling for what is normal performance / temp.

94c is about right, somewhere in the low 90c's without OC/undervolt. Also, run hwinfo64 and watch for Thermal Throttling during the runs - at defaults and OC/undervolt.

The undervolt can fail at idle almost as often as under load, so be sure and do an idle test when you think you are at the final setting - let it sit for a few hours doing nothing - screen on or off.

Glad to see you are having fun OC'ing :)

ZiggZagg
07-09-2014, 02:04 PM
Gottcha hm.... It's early morning here and I just powered up; that is, I haven't run anything but the internet. I'll run hwinfo64 which I've had and used for for sometime and see where I land. Thanks.

ZZ

ZiggZagg
07-09-2014, 02:20 PM
hm... I ran hwinfo64 extreme and the results are pictured here (if I can get it to upload...

zz
38677

ZiggZagg
07-09-2014, 02:21 PM
This was a 5 minute and 11 second test .Should I go longer?

ZZ

ZiggZagg
07-09-2014, 02:28 PM
Not to belabor the point but have you read hozor's submission "one really pisssssed of dude"? I guess I'm just going to remain at 8.0 and hold for awhile:o.

ZZ

hmscott
07-09-2014, 03:19 PM
Not to belabor the point but have you read hozor's submission "one really pisssssed of dude"? I guess I'm just going to remain at 8.0 and hold for awhile:o.
ZZ

ZiggZagg, it isn't a major problem, and has happened since Windows XP days, the DVD/CD/BD disappears.

The Microsoft Support link I gave hozer should get him back online. It happens more often outside of updates, but an update can reset the registry entry for the device, and that is what needs fixing.

Let us know what you decide, and how the upgrade goes if you do it.

hmscott
07-09-2014, 03:23 PM
hm... I ran hwinfo64 extreme and the results are pictured here (if I can get it to upload...
zz
38677

ZiggZagg, you mixed up the programs, you ran AIDA64 Extreme, and the test you ran was for the GPU, not the CPU :)

But, your max GPU temp of 73c is good :)

You can run AIDA64 Extreme to generate the load, but use hwinfo64 Sensors Mode to do the monitoring for Thermal Throttling.

hwinfo64
http://www.hwinfo.com/download.php

These are the lines you want to look at to see if you are thermal throttling during CPU load tests, highlighted with red lines left/right:

38681

ZiggZagg
07-09-2014, 09:54 PM
hm... I got hwinfo64 up and running and it seems to read only in the 70's (74c or75c) for playing the same Crysis 2 game for another 2 1/2 hours. Is GPU Tweak really that far off? WoW! 10c difference avg. Also it lists 2 GPU temps. Whats up with that? There are 2 separate (and different) sets of temps.... I'm a bit confused. The GPU temps are always pretty close but I was just wondering.... it's not possible the program is reading the Intel HD 4600 graphics set up is it?? Noooo.... thats crazy (right?):eek:

ZZ

Oh, I wanted to ask... how do I get a snapshot like you put up in the previous post?
Thanks again,

ZZ

P.S.- My HWINFO64 is Ver.-4.41-2550

hmscott
07-10-2014, 03:59 AM
hm... I got hwinfo64 up and running and it seems to read only in the 70's (74c or75c) for playing the same Crysis 2 game for another 2 1/2 hours. Is GPU Tweak really that far off? WoW! 10c difference avg. Also it lists 2 GPU temps. Whats up with that? There are 2 separate (and different) sets of temps.... I'm a bit confused. The GPU temps are always pretty close but I was just wondering.... it's not possible the program is reading the Intel HD 4600 graphics set up is it?? Noooo.... thats crazy (right?):eek:
Oh, I wanted to ask... how do I get a snapshot like you put up in the previous post?
P.S.- My HWINFO64 is Ver.-4.41-2550

ZiggZagg, the Intel GPU isn't wired up on the JX :)

The CPU temp is going to read different than the GPU temp - Asus GPU Tweak only shows the GPU temp, not the CPU temp - hwinfo shows the CPU temp as in my snipping tool clip

Snipping Tool
http://windows.microsoft.com/en-us/windows7/products/features/snipping-tool

When you start hwinfo64 a dialog comes up, and you select only the Sensors view box to get the Sensors view like I posted above:

38711

38712

38713

38714

Torpedo499
07-11-2014, 02:32 AM
Alright, -50mV was stable, I'm now testing -70mV, and I hope I can push up to -80mV or -85mV.

Are Intel XTU temperatures for the package? Because Speccy and HWMonitor are giving me lower temperatures.

hmscott
07-11-2014, 02:48 AM
Alright, -50mV was stable, I'm now testing -70mV, and I hope I can push up to -80mV or -85mV.

Are Intel XTU temperatures for the package? Because Speccy and HWMonitor are giving me lower temperatures.

Torpedo499, they are all sampling at different times, and that makes it tough to figure out which is which when they don't match. HWmonitor isn't HWinfo64, you might try that one too - it samples more things, and shows the data in a more compact form in the Sensors only mode.

Also, Intel XTU samples and will display a lot more than in default mode - click the Wrench icon(s) in each display area top right to get more options. In the graph and sensors display areas click both wrenches to display more/less on the graph in meter areas.

38746

38747

Torpedo499
07-14-2014, 11:39 PM
Sorry for posting on this thread, it's pretty old, but I wanted to say that -70mV was stable, and I'm now doing -75mV.

What is considered a good undervolt for the JW?

I'm hoping that everything is stable, and I can hit -85mV without crashing!

hmscott
07-15-2014, 08:27 AM
Sorry for posting on this thread, it's pretty old, but I wanted to say that -70mV was stable, and I'm now doing -75mV.
What is considered a good undervolt for the JW?
I'm hoping that everything is stable, and I can hit -85mV without crashing!

Torpedo499, -70mV is a great undervolt, undervolt range it isn't dependent on the G750 model, all have the same 4700HQ CPU - and each CPU is unique in it's undervolt range. I guess there is a similarity from batch to batch, but I haven't been able to test that theory :)

Keep testing :)

I pushed mine to -60mV from -40mV, and it ran for days before I got the sad face crash... back at -40mV :)