[Guide] 580m GTX throtting fix. No need to modify system Bios.

first off thanks very much to widezu69 for the throttle fix 10/10 for the fix!
i have a question can anyone help? i didnt notice before(the recall) but is it normal for gpu-z or any other program to load up @ 0.9200v then drop to0.8000v.i thought only the overclock 580m92v.rom runs at 0.9200v not the 580m.ROM just so you know i use the 580m.ROM. Am i rite to think that the 580m.ROM should not go to or a higher voltage of 0.87v .Ive only noticed mine go to 0.9200v since they changed my heat sink (recall)
thanks

 

Hi,
What driver are you using?

The beta 3xx drivers did this overvolt and some members posted screenshots of .92v. I saw it briefly (same .rom file as you) but haven't checked out the latest WHQL driver I'm using now.

it seems to be a bug associated with a new boost feature and should not be happening on these cards.

The drop to .8v is normal when the card falls back to low power mode (P8). You would also see this voltage at P1 but the fixed VBios matches P0 and P1 voltages as well as clock speeds.

 

Its like around 10-30 too.

 

Hi micky thanks for the info the driver i use 301.42-notebook-win7-winvista-64bit-international-whql

update ive installed 296.10-notebook-win7-winvista-64bit-international-whql now it is fixed i will wait for new driver to be released and skip 301.42 thanks

 

You're welcome, I just checked and the 301.42 WHQL still pumps up the voltage to .92 briefly. I never see it for more than a few seconds when the card initializes. I read somewhere that it was supposed to be fixed in this release.

I don't think this is a problem for the short time it happens but not nice to just OC the voltage and not tell anyone. I'd try to ask Nvidia but I've never managed to get any sensible answers out of them except 'refer to your manufacturer', and we all know Dell are really rubbish at keeping their 'approved' driver list remotely up-to-date - current R3 =28/07/2011/269.03, nearly a year and supposed to be a gaming machine - oops sorry for the rant

 

It could be a bug in the driver that nvidia is not aware of. I mean this is a Kepler driver so the power management has been completely rewritten and this may affect some Fermi cards.

 

I think that's right, although I've seen enough posts about this before ther WHQL release to think they might have fixed it.

I think what is happening IS for the newer cards where nvidia are going towards a 'boost' feature and they can then publish fast top clock speeds even though 'normal' running clock will be slower - just watching the marketing

 

Yeah I think that some Fermi cards like our 580m's are having adverse reactions to the newer coding of the drivers. For example, on my card since the newer 300 series drivers, I get the 65C throttling even with the fixes I introduced. I am using the 0.92v vbios anyways which is why this is happening but I've fixed it now.

 

Yup, they have done a LOT in these drivers for power management so all the goalposts have moved. I'm actually seeing a lot less of the power throttle since the first 3xx release, only one game kicks it in from the get-go, and it's running a couple of degrees cooler . Your higher power draw probably make it worse.

I doubt this driver will get on Dells' approved list with the 580 already running so close to spec in the MX17

 

I was having issues with HWiNFO yesterday. Sometimes it freezes and I have to restart to get it working again. I messed around with the config to see if anything would help resolve that. As I fired up FFXIV and started playing for a few min, I noticed that my fans weren't running and my GPU temp was at 85C. The strange thing is I didn't get any throttle, and I was running at 50-60 FPS like normal. I reconfigured HWiNFO and the fans kicked in and brought my temps back down to 65-69C while playing.

Just thought it was strange that my laptop didn't shutdown like it has in the past when it went over 78C. I am running the latest drivers 301.42.

 

I had exactly the same problem when I was messing around with the fan speeds. The fans never started and I noticed it when it hit 80+. It appears that HWInfo somehow disables the 78C throttle - well it did for me and others have commented the same. This was on the 2xx driver, I haven't needed to use it since a GPU/heatsink replacement...

 

unplugging and then plugging in your laptop will normally unfreeze Hwinfo also.

 

That sometimes happened to me when using the hwinfo with the lpc ec support checked. Other sorts of problem from that one, like hwinfo can't close properly, freezes the computer during shutting down and had to force restart. So I had to check off lpc ec support, used dell ec and everything went fine. Only had to configure fan speed to the 3 predetermined levels.

 

ive been on the phone to dell and they say its fine and its normal to jump up the voltage but i dont think its normal. i know it just jumps for a few seconds but thats not normal at least i dont think it is lol.he said to use the dell drivers and i would be ok if i didnt like the new drivers. im on 296.10 and i will stay on that untill they do a proper fix thanks for your help and info and also thanks to widezu69 for his mod ( throttle fix)!!!!!

 

Ha Ha, define 'normal'? It happens to everyone so it is normal? Acceptable? Now that is a different thing. It has to be a bug in the coding since there is no corresponding clock jump to qualify a voltage increase over stock.

Maybe their fix was to make it drop out back to stock straight away. I do recall seeing it for minutes at a time in the early beta but I can't remember what tests I was running to see it

 

For HWiNFO if you don't want fan control enabled disable SW-SMI and the EC option above EC-ACPI synchm when you don't see the fan symbol in HWiNFO you did it.
For thise with HWiNFO lockups ensure fan control is set right. You can't use SW-SMI and the option above EC-ACPI synch.
SW-SMI uses Dell EC to adjust fans. The option above EC-ACPI synch uses the Compal EC for fan control. Now by feault both are enabled and require you to set it properly yourself. Don't use both it just won't work right. Or just disable it altogether.
Now SW-SMI is the only one that will work error free on most systems (dell ec).
Compal EC can set any fan speed unlike presets in dell ec but has stability issues. This option is the box above EC-ACPI synch and it combined with the EC-ACPI synch will help bring some stability to the Compal EC for some but issue is it's still causing shutdown hangups, system freezing of the hwinfo app and more for some not all. If you have issues use the SW-SMI or none at all.
If you guys use GPUI2C enabled if you have an Nvidia card you must also enable NVAPI under that to ensure stability on Nvidia based systems.

On fan control if your using it or it's enabled you must ensure you've configured it or by default nothing is set and that means fans will equal 0rpm. If you don't know what your doing disable fan control. One user not at HWiNFO's fault had a faulty fan and he had to get plastics replaced after they melted over his gpu when it ended up running 99c+. Head my warning his might of been a faulty fan but if you don't set HWiNFO the correct way you face the same fate that user did because the fans won't spin up. Take the time to set it the right way. All NB users should see this or something similar as a warning! It's a mistake as dangerous as bios and vbios mods... It will melt!

Make sure if you use fan control and set it automatic you need to set the fan table and the sensor that hwinfo reads from. Intel#0 sensor is a good bet. A GPU temp sensor is a good bet also. You can only read from one sensor so pick one that will read high temps when the system warms up so the fans spin up.
Setting the fan table wrong can cause hwinfo to tell the fans to go to 0rpm so test before gaming a long time to be sure what you set works.
Again if you don't know what your doing disable the fan control as I mentioned how above or major system damage can occur and technically it's not the mfgr's fault it would be user mistake.

Regards RV/MW86

 

Hey toonnut1, I am noticing this voltage jump as well. Could this be causing weird instability or TDR errors? No matter how many times I clean install the 300x verde drivers on my M17x R3 (listed in my sig) I still get these random TDR errors.

Just like you, I'm using the 296.10 drivers until they sort this out. I can't believe they made these an official release.

 

Thanks for the tips. I just adjusted my settings and will monitor it over the day to see if my hangups lessen.

 

Great I hope it brings you stability. feel free to post a pic of how its configured if your still having issues later.

 

Here are my current settings. Most of it has been configured from this thread: http://forum.notebookreview.com/ali...mperatures-game-benchmark-screen-display.html



If you have any other suggestions, please let me know. So far when it has frozen before, I would just unplug the AC adapter. I haven't had HWiNFO on long enough to notice any freezes today.

 

Hi widezu, I appear to be having a similar issue with the .92v for my gtx 580m. it appears to be stuck on the undervolted clock with the latest nvidia drivers. how did you manage a workaround ?

 

looks correct to me. if you run into those issues again uncheck
LPC&EC Support and EC-ACPI Synch and instead check SW SMI. only if you run into the issue again. if you do switch the fan control to the SW-SMI one reconfigure the fan table and control since it may end up being reset when you switch the fan control modes. close and restart HWiNFO to take effect.

 

Hello I'm new here and this is probably a stupid question but I cant seem to find the attatchements for the 580m.ROM file. Please help!!

here is the step below...

To do this, you will need to download:
Download NVFlash 5.100.0.1 for Windows | techPowerUp
and the attachment below. There are 2, one is 580m.ROM which is a normal one with throttle fix, the 580m92v.ROM is an overvolted one.

 

Welcome to NBR!
The file is right before the signature, but here's the link.

Both vBIOS are inside the archive.

 

Thankyou

Hello again, I'm now having trouble with imputing -b backup.rom in the cmd because it keeps telling me '-b' is not recognized as an internal or external operable program or batch file.

Please I need this to work so I can play some Battlefield

 

Did you go to this Location and download the NVFlash utility.

Note the NVFlash utility is an update from the link posted on Page 1 of this thread - please follow the instructions on page 1, they are 100% correct, I have used them

Please be aware that there are 2 different procedures, 1 for 3D machines, the other for Optimus (none 3D), from your question it looks like your none 3D, make sure you down load the DOS version of the NVFlash to be used on your bootable flash drive.

 

My M17x is has a 3D screen, so that makes it 3D right?

 

I had the exact same problem yesterday, and i got to work around it... Here is what is happening... For some reason, the nvflash program drag to the cmd prompt window is just not being successful... For that what you need to do is change your working directory in command prompt to the one containing the nvflash.exe program (use dos command 'cd')... after you have browsed to the directory containing the nvflash.exe, type in "nvflash.exe -b backup.rom" without quotes at the prompt and press enter... Voila! It will make a backup of the current rom in the same directory...

I tried this yesterday coz i too needed to buy and play BF3 Premium; it is awesome!

Oh, and one important note : Make sure you run cmd as administrator (right click on cmd and choose 'run as admin')!

 

So.. I have random slow downs, and I normally can prevent them by just turning down all the graphic settings in whatever game I'm trying to play (right now Diablo III)... but every once in a while nothing seems to make the card come back to full framerates, am I being affected by the throttles you're mentioning here? It seems like I'm stuck in P12 state, because P0 state says the clocks should be at like 620, but even though it says P0 state, My "Current Clock" stays at "25 Mhz".. and I get like 2 fps.



Any ideas? Will this vbios fix this problem, or is this something else?

 

That is just weird, not even hot either!

You're not getting any of the normal throttleing that has been discussed ad-infinitum in threads here. I don't think it's any throttle problem.

The states you would expect to see are P1, which kicks in when power draw on the card is high (also in some games during periods of low usage like in cut scenes). Then P8, which is the 'overheating' throttle at around 78C, and finally P12, the cards final 'Im in trouble but can run' setting.

Lowering the settings in the game might be fixing it just because it resets the card not lowering the power draw.

To see P0 state and 25mhz clock is not something I have heard of in the 6 months or so I have been monitoring (and getting to the bottom of) the throttle problem.

It sounds like the card is faulty but it won't hurt to flash it with the latest Dell one just to rule that out.

There are more knowledgeable people here so maybe one of them has a better idea?

 

Hello guy's!

M17xR3 owner here, I did the blind flash and all was good until I wanted to try a new game (The Secret World), it asked me to update the Nvidia driver, which I did, but when I am in game I get regular drops in framerate, like from 70fps to 18-20, all the time.

I installed Nvidia inspector and the card is not heating much, is my card defective!?

Thanks!!

 

Hi, not sure about the 'blind flash' you mention but if you tried to apply the Vbios fix for the power throttle it did not work. Your screenshot shows P1 at 475. This would be 620, the same as P0.

To see if you are having throttle problems you need to see what is going on in real time when gaming. As soon as you switch out things change and temps drop in seconds.

Check out MSi Afterburner (or EVGA Precision). You can set it up for an On-Screen-Display to show you clock speed, temp and other stuff.

Don't use the overclocking as you can only set P0 but you already have NvidiaInspector for that

 

When you say you did the blind flash, do you mean that you did a conventional flash of the A04 bios, followed by a blind flash of the A08 from the PAR00X64.fd on a USB stick. If not, what did you flash? If the A04-A08 flash is successfully done, the power throttle (63C +/-) should be completely eliminated. Also, when you updated the Nvidia drivers, which version did you install?

How about running GPU-Z, going to the Sensors tab and click both boxes at the lower left (Log to file and Continue refreshing . . .) and then running the game for long enough to experience the slowdown and then recover to normal speed. Then post the log file here to allow us to see exactly what's happening, which should tell us how to fix it.

 

Hello Revelator,

Yes I did the A04-A08 flash and all was fine in the games I was playing but now this new one (The Secret World)has big drop rate in game.

I updated the driver to 301.42 as per the game was asking me as well.

I attached the log file from GPU-Z, did I do it correctly?

I had a drop from 40fps to 8fps in this log.

I am playing in ''Ultra'' settings, no tweaking, just the general Ultra setting.

Am I asking too much from my GPU?

Thank you!!

 

You did the log exactly right, but it does not reflect any GPU throttling at all. The temps look fine as well. The card jumps to .92v as soon as the 3d clocks are triggered and remains there throughout. Have you flashed the vbios to increase voltage? If not, then it is a side effect of the 301.42 drivers with the 580M. The 300 series Nvidia drivers are designed primarily for the 600 series video cards and the Kepler chips and are not completely compatible with the 580M. One of the problems is that it not only increases voltage, but also keeps the clocks at P0 settings even when the load has dropped. In many cases that combo leads to increased temps, but there is no indication that result is happening to you based on the logged sequence. Still, I would uninstall the 301.42 drivers and install the 296.10 set and see whether that helps. If you would repeat the procedure and post the new logs at your convenience, it might shed a bit more light on the subject. You do not want to use the 301.42's in any event. I had to remove them from my R3 soon after installation.

I am not familiar with The Secret World. I suppose it could be too demanding for the 580M at Ultra settings, but seems like it would have received a certain amount of notoriety around here if it was the new graphics killer game.

 

Yes I did flash the vbios to increase voltage : /

Yes, I wasn't sure if I should use the increase voltage or the other one : /

I will install 296.10 and make a new log.



BTW: I thought you were dead Snake

 

Yeah, you and everybody else. I grew a new tail.

 

I installed 296.10 and ran the game, here is the new log in attachement.

I saw drops to 11-13fps but visually it wasn't as harsh as with the 3xx driver, before it could turn into a slide show, now it's bearable.

Should I flash the other file (not the increase voltage)?

Would that help a bit more?

Thanks again for you help, it's already much better than it was

 

Hey, I checked this Secret World out a couple of weeks ago. Also got the warning that I needed a 301 or later driver. There isn't one for my 675 but a beta with an inf from laptopvideo2go sorted that.

Anyway, no card killer, it just uses the new features like adaptive v sync and FXAA from what I could tell. I saw steady 35-40 FPS at max settings

 

Glad there's improvement. No throttling in this log either, but the loads are higher and temps are considerably higher, up to 75-76C from highs of 68-70C on the first run. I would recommend flashing to the .87v vbios. I don't expect that to result in higher fps, but it should reduce temps and avoid throttling at temps reach 78C. Really can't see why you're experiencing even reduced slowdowns at the moment, but reduced voltage can't hurt, and it might help. You don't usually need the .92v unless you're going to overclock the core at or above 775-800, depending on the card. Might just be a matter of early game glitches or driver adjustments that have not yet been identified. Sorry I don't have a better answer.

MickyD, I tried the new version of the Verde 304.48 betas (added .inf files for additional mobile cards) this morning. They're an improvement in that the clocks and voltages adjust as expected and performance was marginally improved (in Mark11), but I ended up uninstalling because I was having some minor issues, mostly in Chrome (and oddly, the Nvidia driver search page). The Chrome issues mostly disappeared (but not the NV driver search page glitches), after reverting to 296.10. Might be something else affecting Chrome; I installed several new extensions recently and may have screwed up the configuratiions or created conflicts. Probably need to learn to leave well enough alone -- again.

 

Thanks, that's the version i have been checking out for the last few days. Same conclusion as you re the card's characteristics. I still see .92V overvolt at initalisation but it drops back in about 10-15 seconds and does not return until after a reboot - go figure.

I've been using a 580 for months so all my experiences are no longer valid. I know this 675 is just a reflash of the 580 but it has run pretty much to spec from the start. No need to worry about the power throttle, I only see P1 when it should like low utilisation during cut-scenes. Even an OC to 720 only causes P1 briefly during high game stress periods. I just use Nvidia Inspector to up it although I know I am getting a voltage drop as well so I avoid the OC if a game plays well.

Never used chrome but the NV drivers page ALWAY hangs detecting for me since I got the 675

I'm wondering if you have any idea about some strangeness I am having though? Since installing the 301 WHQL driver, that needed me to get the inf as it would not install (no detected hardware), I can only use LV2go inf's. If I try to go back to the dell published version that worked fine before, they now give me an error - no suported hardware (or something very similar to that).

I tried a full slash and burn uninstall, even ended up with the machine starting on the generic windows VGA driver, but still the default dell driver refused to run without a modded inf

Not a big issue just annoying that I have to use LV2Go all the time now

Edit: Forgot to add the 304 beta driver is the first that NV state supports my card so I was surprised I needed the inf. Having finger problems, time for bed methinks...

 

Hello guy's,

Me again, I am still having problems with major framerate drops in The Secret World, in the file log attachment I went from 30-35fps to 2-4fps and I get that constantly.

I fired up Battlefield 3 just to see how it would run and in Ultra settings with all the eye candy on it ran flawlessly.

I don't get it, my GPU can't be defective if it runs BF3 like that but in Secret World my god it's bad.

I don't knopw how to fix this

Thanks!

 

This fix won't help with that. You are overheating. Try a repaste and elevate the back of your laptop. You are hitting the 78C threshold.

 

^^ Yep. Surprised it's not throttling harder than it did on that last run.

 

Thanks a lot for this Guide, it really helps a lot.
I'm now wondering just how much can I oc my gtx 580m SLI on M18x.
Currently 1800Mhz memory clock, 1650Mhz Shader clock, 0.92V.

Please tell me your results.
See u !

 

TSW throttles for me after about 30 minutes as well. Already repasted with same results. Not sure how ChrisM can say if it hits 78c within 2 minutes then it's an issue with the heatsink/paste...It's done that since day one. I don't know of any card I have owned that didn't hit 78c in 2 minutes.

 

I have not had any issues with throttling using my method, even if it goes over 78c it won't throttle using my guide:
http://forum.notebookreview.com/ali...e-fixing-m17x-r3-580m-thermal-throttling.html

I was playing the Secret World around launch, I did have to use the method described in the guide to avoid the throttle, but using that method I had no issues.

 

Hey, I've been following your guide and I am at the part where you're supposed to open Nvidia inspector but all the options are totally grayed out. I checked device manager and it only says my Standard VGA Graphics Adapter is working. I tried uninstalling and installing the latest driver update from geforce.com but again it can't find my video card. Any idea what might be wrong? How do I use the backup.rom?

 

Ok, so something truly peculiar has been bothering me as of late.

I successfully flashed your bios some months ago and been happily running an overclocked 580m (725/1650 - regular, .87v bios).

The card ran fine and dandy - no high temperatures (never surpassed the second throttling stage at 73'), no visible artifacts whatsoever; been playing lots of demanding games for the whole time, without a single issue.

Last week or two, I started getting weird BSOD's left and right - google revealed them to be GPU related.


I quickly downclocked the card to stock, but the problem remained; reinstalling (clean install) WHQL drivers didn't help much - I kept getting same BSODs, whenever 580m was in use.
Hell, 5 minutes of Battlefield 3 pretty much guaranteed a BSOD - all at a low temperature (59-65').

Today I've decided to try reflashing the backed up, stock vbios; lo and behold - the issue disapeared completely (or so it seems; no BSOD while benchmarking // playing BF3 and Witcher 2 for hours).

Any idea as to what may have caused it? Why did it start BSODing me after running fine for months?
The only thing I recall has changed, was me getting a brand new, original replacement battery two weeks ago or so (but I always run the notebook connected to its PSU, anyway).

Didn't try going back your bios yet, but I sure as hell want to.
If only to keep the consistent voltage for both P0 and P1.

Any help appreciated.

edit: yikes, sorry bout the flipped image; fixed now.

 

Bump? Anyone?