GTX 560M hitting 95C on load! Advice please.

Hi, I recently got a new laptop from Scan.co.uk - it's one of their 3XS PowerDAW laptops for audio, basically it's a Clevo P150HM with a custom BIOS for audio work and a bunch of other tweaks. I do however use it for gaming too and have noticed some ridiculous temperatures reported by GPU-Z when on load. Basically the GPU is idling at around 60-65C and after playing Skyrim for a couple of minutes the fan speeds right up (any fan throttling tweaks for silent performance were disabled as I told them it wouldn't be necessary) and hits the 95C mark at it's peak. I'm pretty sure this is insanely hot! Obviously it also impacts gaming performance as once the fan goes on full whack my FPS drops dramatically.

Other info I can think might be useful - The card isn't overclocked and is running at stock speeds, the drivers have been updated to the latest ones, the laptop was on a flat surface with plenty of airflow when testing it, and since it's relatively new I had a check inside and there was very little if any dust in there. The GPU and cooling appear to be correctly seated. The CPU is idling at around 35C which I'm guessing is pretty good.

System specs are -
i7 2820QM @ 2.3Ghz
8GB DDR3 @ 1300Mhz
GTX 560m 1.5GB
500Gb Seagate Momentus hybrid SSD
Windows 7 Professional X64

I'd considered reseating the card with some arctic silver stuff I've got lying around but since I'm not sure if this would mess with my warranty or if it would even have an appreciable effect on a card that's already hitting 90C+ on load I thought I'd ask here first.

On a side note it also drops the core speed right down to 162Mhz when running on battery, obviously this makes gaming on the move impossible (which I thought was rather the point of a gaming laptop). I heard some stuff about Nvidia Powermizer screwing this up but as of yet I've found no way to disable it under the current, or any recent drivers. Any thoughts on that would also be appreciated.

 

First off, gaming on battery is not recommended because the hardware will automatically downclock to relatively low levels in order to enhance the battery life.
Gaming on battery on full performance mode would suck the battery in well under an hour (or something to that effect).

Batteries have been in stagnation for about 2 decades now, so they are under-performing to say the least.
Majority of battery life that you see being advertised as 3 or 4 hours (or more) is when the screen is practically set to its lowest brightness and when you don't do intensive tasks on the laptop.
You can usually get away with Internet browsing and move watching for several hours.
The gpu will unavoidably downclock on battery (I think there's a workaround to prevent that, but it will suck the battery fast as a result).

In regards to your temperatures...
You have to call up and ask the store whether you loose the warranty on opening up the laptop.
Usually, very few manufacturers prohibit users from opening the laptop.

Either way, the temperatures of your gpu are not exactly encouraging.
It is likely the issue with improper and/or low quality paste, and whether the heatsink is touching down on the gpu chip properly.

To that end, once you repasted the gpu, I would advise you to undervolt it.
There are ways to do so inside the OS if I'm not mistaken, or you can flash the gpu's BIOS so that it consumes less voltage on maximum load (this can easily reduce the temperatures by 10 or 15 degrees C - your experience may vary though or be lower).

 

yeah the battery thing tbh, all I need is half an hour to an hour, I just want to play Skyrim or the witcher or somethin on the train ride home from uni. It's no biggie though, as I said I believe it's something nvidia users are stuck with for the time being anyway due to powermizer or whatever they're using now no longer appearing to have editable registry values.

With regards to the temperatures, I've had a look inside and there's no stickers or anything preventing me from easily opening the case - the same panel that requires opening to reseat the GPU is the same as replacing the RAM which I know is normally never a problem. I shall give that a go since I have some decent quality paste here and I can't see any way they'd actually be able to tell if I'd opened it.

Undervolting also sounds like a good idea - does that impact performance particuarly, and are there any good guides on this or should I "just ****ing google it"? Also hopefully it won't be in the system BIOS because it's a custom BIOS with very few settings.

 

I've noticed my GPU getting a bit hotter recently as well. Is it relatively simple to apply paste to to it?

I've built a desktop but never even opened anything on a laptop.

 

good advice from Deks

this is old now (almost 5 years) but guess it still stands for a guide for clevo lappys.

you could also check this guide > http://forum.notebookreview.com/not...idia-ati-cpu-hdd-do-your-temps-worry-you.html

 

Hm interesting, well it looks pretty simple to apply some decent thermal paste, although I've emailed the guy at Scan anyway (as I have a direct email for the guy who built my system anyway due to some previous audio-based issues). I'm definitely interested in undervolting since that seems like the most effective method although I'm having trouble finding any recent guides or anything pertaining to the 560m card.

I've tried Nvidia inspector, which along with GPU-Z reports my GPU voltage constantly at 1.000V, there was only the option to drop this to 0.820V but even experimenting with selecting this, when I hit 'apply' it jumps back up to 1V.. it seems like the card is constantly stuck on max settings and that utility at least refuses to undervolt it.

EDIT, also am I right in thinking these power states are what becomes active on battery and throttles performance? So if I could get it working could I theoretically boost the lower powers states a bit and get some gaming performance while on my daily train journey?

 

Undervolting the GPU will NOT affect performance.
It will merely reduce the amount of power the gpu draws at stock frequencies/clocks (Nvidia and AMD both usually set a certain 'standard' for stability reasons - but its a pretty high margin actually that can be brought down - by how much is dependent on the chip, though majority seems to be able to undervolt to same levels on stock frequencies).
You need to experiment a bit so you can test out what your minimum stable voltage is when the gpu is on maximum load.

If the gpu is hitting highest frequencies on battery, then the undervolt should apply either way and for such a scenario, it should extend the battery life somewhat.
As for the lower states that throttle the gpu performance for extending battery life... I'm not sure if others were able to accomplish that.
You will have to do some research... however... take a look at this thread:
http://forum.notebookreview.com/alienware-m18x/614025-dell-gtx580m-sli-undervolted.html

It should provide some nice insight into minimum undervolts, etc.

Also, when you apply the paste, make sure the heatsink touches the gpu and all other chips properly.

 

It appears my messing about with powermizer to get gaming performance on battery caused some of the issues. I had removed the powermizer settings as they weren't apparently doing anything and the driver reinstall (a clean one might I add) didn't do anything to restore them either, which I'd assumed it would do. Using powermizer manager to fix that and reinstall the default settings has made the power states work again and the idle temps dropped to around 46C and 0.82V - WAY better! Still getting quite high temps on full load while gaming though - improved but around 85C+ still, peaking at 90C when I'm looking down at the entirety of Whiterun on Skyrim with settings on high. Voltage is back up to 1v at this point of course.

So I think undervolting at least a bit will bring them to more acceptable levels where I'm avoiding damage to the card. I'll have a look at that guide, thankyou. I'll leave off the thermal paste for the time being since I'd prefer not to mess with the physical hardware. No point messing with the internals if I can get acceptable results from just undervolting.

 

Here's what I would recommend.
Experiment with the Voltage on stock clocks (full load) and see how far down you can go without stability issues.
Once you find the lowest possible voltage for stock clocks under maximum load, the next step would be to flash that undervolt into the gpu BIOS to keep permanently - or you can just keep the software undervolt if you prefer it like that.
I never touched the lower states voltage personally on my gpu as I was mainly thinking about the maximum load.

Finally, you will have to open the laptop and re-apply the paste (I would do this on the gpu and cpu - as its possible the manufacturer did a shoddy job on both) and make sure you reapply the heatsink properly.

 

My problem at the minute is I can't figure out how to change the voltage. None of the free utilities I've tried (Nvidia inspector, MSI afterburner) allow me to modify the P0 core voltage, even though I've made sure the ability to do so was enabled. I'm starting to wonder if there's a custom vbios since there is a custom system bios installed. GPU-Z is reporting version 70.26.29.00.06 which seems to be one other users have online, the odd thing is most of the posts I found were complaining that they wanted a 1v power state, whereas mine has that as default!

I did find this updated BIOS here from Dell - Drivers and Downloads | Dell [United States]

But I'm obviously a bit apprehensive about installing it because my laptop's a Clevo/ADK - probably not an issue because it's the exact same card, but well you know, better safe than sorry. Basically I've backed up my current BIOS using GPU-Z but wondering whether to replace it, whether using that dell one would be safe to do.

EDIT - Apparently you can't change mobile GPU voltage with software.. so yeah, can anyone even tell me what tool I'm supposed to use to change the core voltage or create a custom BIOS which will? I've been searching the internet all day and found absolutely nothing. I thought I'd found a VBIOS editor that would allow me to create a custom BIOS and flash it on there but it said it didn't support my card.

 

One thing I learned recently (can't remember the user's name) is using a frame limiter will greatly reduce power consumption. DXtory is a shareware program that will regulate fps. Set it to 30 fps and you will see your heat and power draw drop significantly, not only that, I've had things smooth out because fps is pretty much constant.

 

HTWingNut, thanks for the advice. I shall have to try that.

 

That might not solve the underlying issue though.
to the OP... man I just realized I was giving you links to 580M... sheesh.
I can't find anything on undervolting the 560M, but the principles should be the same.

As for software undervolt... if you can't do it like that, then it would be prudent to do it by flashing the gpu bios.
GPUZ can extract your vBIOS and NiBitor (Voltage tab) can allow you to mess around with the voltages.
I think you could set your Extra and/or 3D voltage to less than 1.0 (I think it was extra setting that did it for me on max. load).

But I digress, I also read that another guy had some issue with his DELL laptop because the heatsink was not touching down on his GPU properly.
So, I really recommend you go in there and check the heatsink.

I remember I had a similar issue with my laptop and 9600m gt after repasting.
For some reason, I thought I took care of the heatsink ok, but it turned out that one of the hinges screws wasn't pressed all the way down (which would make the heatsink touch onto the gpu die).

So, repaste, and make sure to get those screws down properly.

 

I do already have Vsync forced on as well as triple buffering, which should reduce my max FPS to 60 (might as well since the display's 60Hz anyway). 30 seems a mite low for my tastes though and in honesty when Skyrim does start chugging with all the bells and whistles turned on particuarly busy areas only hit about 35-40FPS anyway. Plus my other games I'd prefer at a higher framerate anyway. Thanks for the advice though I'll have a look into it anyway.

My only problem right now is that I tried Nibitor 6.04 yesterday and it doesn't support 5XXX mobile cards at all so until they add support it's no good for me. The only program I've found thus far that specifically states it can change the voltage on a 560m is Svet's VBIOS Tuner, which is donationware costing about £12.. seems a reasonable price considering it will likely extend the lifespan of my laptop and avoid overheating so I might invest in that after the weekend when I get a reply back from the guy at Scan. I may also be able to overclock/overvolt the 2nd power state using this tool (moderately) which would theoretically give me at least a bit of gaming performance on battery too.

With regards to thermal paste a bit of research showed that meths is fine to take the existing paste off and I've got some arctic silver 5 somewhere, so I'll give that a go. Shame all the youtube videos on applying thermal paste to laptop GPUs are absolutely hump, but then if those guys didn't break their laptops then I should be fine haha.

 

Frame limiting is different from vsync. Vsync will still churn away but throw away info. Frame limiting will only produce what it needs, at least the way I understand it.

 

It is an ok concept, but the OP needs to take care of his high temperatures either way so that he can game at lower temperatures without resorting to such measures at the very beginning.
95 degrees C for 560M on maximum load fresh out of the store points (at least to me) that something might not be right.
The laptop could simply have faulty cooling or a manufacturing error.

If worst comes to worst, you can always return the laptop and tell them the temperatures are simply too high and unrealistic for that type (or any type) of GPU.
The 580M in some less quality laptops was reaching 80 or 85 degrees C, so I doubt that DELL would allow THIS for 560M (which should have lower temperatures).

 

I know this, but the discussion went on how to reduce power consumption, so I threw in my two cents. I agree it's too hot, repaste is the first thing I would do... but that's just me.

 

Well yeah, I mean it's a custom made job as well. Only thing is this card got put in after first getting it. They had a bunch of models and there were some firewire issues with the first one I got (mega important for pro audio) so I got a new shell and it originally had a 460m in it. Obviously I told them since I'd ordered the latest and the guy put a 560m in there. I don't know yet but I'm starting to suspect he's rushed it now.
I certainly didn't complain at the time as he'd also apparently given me one with a 2820QM which is £120 more expensive than the actual CPU I'd ordered

But also 1V seems too high a voltage for that card, people on other forums were overvolting to that so they could overclock the GPU, others were running cards much higher on a far lower voltage, so I'm thinking somewhere near 0.9v would be more appropriate, still allow me to play current games at a decent resolution and drop the temperatures out of the danger zone. I can't really be sending it back right now because of needing it for university and this being the busiest time of year by far.

 

Odd, my 560M on stock voltage is at .9610v under load. That is if GPU-Z is reading it right.

Edit: It just dropped to .8, so I guess it's a power saving thing.

 

My 9600m GT originally had Extra (or was it 3D) setting (which indicates full load state) set to 1.1 or 1.2 Voltage, which I reduced to 0.82V (the minimum).
This dropped my temperatures by around 15 degrees C (from 85 to 70) and with around 20% overclock (on the lower voltage) it rose to around 75 degrees C.

Anyway... I'm not sure what the standard voltage of 560M is, but some preliminary search indicates a 1.0 Voltage setting on full load (I could be wrong though), and that it 'might' be possible (but don't quote me on this) to set the full load voltage to 0.82 or 0.8V.

Either way, I do recommend, that before you try messing around with the voltage, you open the laptop and do a thorough cleaning from possible dust and re-apply the paste properly to both the cpu and gpu (and possibly the northbridge if needed) and then re-seat the heatsink.
After that, do some tests and see how high the temps go really.

But in the end, if nothing helps, and you exhausted all your options, I would recommend returning the laptop and either making an exchange or something else.
There's no point in keeping it if the temperatures are as excessive as this - which in turn could result in a shorter life-span of the laptop (which is questionable given that both the gpu and cpu have a maximum rated temperature danger zone above 95 degrees C - I think it might be in the 'just over 100 degrees C ballpark' for the cpu and around 125 deg C for the gpu - but again, don't quote me on that).

Clean it and repaste, then report back please.

 

It wasn't remotely dusty as it's pretty new, but I just opened it, cleaned the heatsink with meths, spread some arctic silver 5 really thinly over both the heatsink and entire area of the chip, then stuck a tiny dollop right in the middle for good measure and reseated it firmly.

put Skyrim on, went to Whiterun, looked down at the whole town for a bit to get the GPU good and hot, and flicked out to GPU-Z to see what the max temp got to..

74 degrees max!

Well that was a good investment then! I didn't bother with the CPU as it's temps were well within acceptable range, I'll monitor them with coretemp and do the same if they do get too hot. I may still look into the voltage regulation mind, purely because it might cut that further and svet's tuner looks like it can create a VBIOS that gives me half decent gaming performance on battery, which is another ambition of mine with this thing.

 

DID ANYONE SUGGESTED HIM TO CHECK IF VENTS ARE CLOGGED OR NOT?
sorry for caps

 

I already did that, and he mentioned that it wasn't really dusty (the probability of it was low due the premise the machine was new).
The manufacturer apparently applied low quality thermal paste and/or the heatsink wasn't seated properly (quite possibly a combo of both).

Glad to see your temps are at much lower levels OP.

You can still look into possible undervolting if you want to, so you can get them lower and introduce a slight overclock... that would probably reduce your temperatures down to mid 60-ies.

 

Spoke too soon I'm afraid, played for a while to test it for longer periods and was getting 90+ again.. crap..

I'd still like to try the voltage thing, but I'm waiting on a reply from this guy at Scan who won't be back till tomorrow at least. The thing is I'd rather address the issue at my end than send it back, especially at this important time of year for me where having a laptop to work on is imperative.

It doesn't need to be an uber gaming laptop or overclocked or anything, I just need temps that won't damage the card and can handle a bit of load, especially when I'm working with the Skyrim editor (something to do with my planned 3rd year project).

 

go to bios and set it to defaults. Hied settings may be set wrong.

 

Nah it's a custom BIOS with very few customisation settings, and I'm pretty sure you can't do anything to the VBIOS via your normal setup anyway.

 

oh, i'm tired.
99% of laptops have very few customisations only... do you think that I did not know it? Maybe this is why i said that there are hidden options and you need set bios to defaults?

I don't say that it must be the way to solve your problem but I believe I am more experienced than you and even 1% chance should be covered if it is so easy to cover.

Otherwise I may think: nah, I houldn' even bother to suggest anything

 

The BIOS is made by the guys at ADK and is customised specifically for professional audio work - the only reason I bought this one is because it's been optimised to avoid certain sleep states that cause massive audio problems on the current generation of i5/i7 laptops. Resetting this would not be advisable since if, as you say, there are 'hidden' elements to it then it could reset things that I need in order to use it for audio work.

Besides I don't see how resetting the system BIOS is going to do anything to the VBIOS. Surely the VBIOS is stored on the GPU for discreet graphics cards and would be configured entirely seperately?

 

vbios, bios or else... but if you open some bioses by AMITool or else there are many options which make influence on.. they influence on everything.

Anyway possibly you are right that it is dangerous to set it to defaults... however most likely people who modded bios made that special setting as default one.

 

Yeah I'd considered that, when I emailed the guy I asked him whether the VBIOS was custom too, obviously if not then I can likely flash it without affecting the normal BIOS settings. Still though there's guys running cards at 1V and overclocked too without these heat issues so I do wonder whether it's still a hardware issue. If I can get it low enough to not care and keep it there I'll be happy though, it's not primarily a gaming machine but I will be editing Skyrim for a project so need at least some reliable graphics performance.

 

You mentioned earlier that you hadn't repasted the cpu.
Perhaps you should try that.

As for cards that run on higher voltages and frequencies without these heat issues... well, their cooling could simply be better.
Yours on the other hand might not be able to handle it though (seeing how it IS a rather cutomized laptop).

Either way, it would be advisable to try and affect the voltages on current (and hopefully stock) clocks, and then find a way to flash the gpu bios.

If the CPU repasting doesn't work and you can't find a solution to lower the temperatures, I would definitely advise you to return the laptop, because in the long run, this could cause problems.

 

The GPU and CPU have separate heatsinks and fans so I doubt repasting the CPU would do much for GPU temps.

 

Yeah they're totally seperate with their own heatsinks and fans. Plus I stress tested the CPU with prime95 earlier and it was hitting about 75C, considering it's a quad core CPU in a laptop I would consider that pretty reasonable. Another thing I noted was the heatsink on the CPU side is kicking out fairly luke warm air whereas even on idle the GPU heatsink is pushing out noticeably hotter air. I may try a repaste of the GPU just for good measure mind.. how much is an appropriate amount to use? Maybe I didn't put enough on.. I spread a really thin layer on both surfaces then only put a tiny blob in the middle of the GPU. I keep getting told less is more but there's really no visual guides I could find as to what was appropriate.

 

I just got the temps down to around 70C on full load, also got a few more FPS out of Skyrim too! Basically I reapplied the thermal paste after finding some guides which took the temps down to around 80C on full load. Basically just totally cleaned the card and heatsink and instead of spreading applied one decent dollop on the center, this worked a lot better than some guides saying to spread it first.

The real difference was indeed the voltage, which I'm currently testing at various levels for stability. Thought I might as well post my results since if anyone else is searching like myself they might find this topic helpful, you never know..

I used GPU-Z to backup the VBIOS and renamed the .bin to .rom. I needed Svet's VBIOS Tuner in order to actually modify the voltage for the GTX560m card, then after saving my modified VBIOS used NVflash to update the new VBIOS through windows/command prompt. Actually pretty simple once I got that combo of programs, I was surprised.

Anyway I took the highest power state down to 0.9v and this worked although after leaving Skyrim running for about 15 minutes in a high-load area the display driver crashed completely, so I'm guessing that was too low to be completely stable although it did work fairly well until then. My current VBIOS is instead set to the next setting of 0.93V, this is doing pretty well, I've left the game running for a while now and it seems fine - framerate is actually way better (maybe due to avoiding thermal throttling) and the temps reported by GPU-Z while gaming are a maximum of 73C and an average of 67C. Since that's a reasonably demanding game with super-high res textures and several graphical mods installed I'll consider that a decent result!

On a side note, I'm going to attempt to create a custom VBIOS that allows some limited gaming on battery while in the P(8) power state as it looks like Svet's tool can create this profile. I'll probably slightly throttle the clock speed from the normal 675mhz, but since P(8) drops it all the way to 202mhz I'm thinking of overclocking this state to around 400mhz and seeing if this gives a better compromise between battery life and the ability to game on the move.