M17xr2 with 7970m and playing crysis 3 on ultra = shutdown at 85 degrees.. Any help?

Ok, so i have a 17xr2 with amd 7970m card on it. As I play far cry 3 my temp goes from high 70s to mid 80s, and at 85 i got a shutdown ( which is expected with the r2 and 7970m combo). So my questions is... Is this normal for far cry 3? Or did i do my pads and paste wrong? Or am i suposed to reclean and repaste every 3 months? I just dont want to reopen and do all that stuff again, if this happend to everyone with an r2 with 7970m. But then, if my temps are too high in comparison, i might need to do it again.

Sorry for title mistake, it was far cry 3" not crysis 3

 

Try cleaning the fans.

Sent from my GT-N7100 using Tapatalk 2

 

Does it happen every time you play far cry 3 or any other game? 85c should be well in the limit of your gpu's. What paste did you use when you repasted? You shouldn't have to repaste again for at least a year depending on what you used and if you did it correctly.

 

varies from game to game, but farcry is the on with the highest temps. 85 is a safe temp for the card, but unfortunatly on the r2, a temp above 85 will cause a system shutdown from the bios

 

OK, i reopened the damn tbhing, blew compressed ait in all fans, also onthe heat syncs, and tighten a little the screws on the x bracket (2 felt a bit loose). Anyways, it worked great, was able to runa complete session that did nto go beyond 80 (thats at least 6 degrees less). Thanks again.

 

The M17x R2 will not shutdown on GPU temp, only on CPU temp, the GPU can trigger throttle event not a shutdown event.

 

Unphoto..are you sure about that? I am a noob, so i am the less likely to question anyones assertions, but i have had ,y processor temp below 80 on a constant basis ( and i believe that is well within the acceptable temps), and after cleaning the fans on the gpu... Not only did i not get any shut downs, and temps were under by about 7 degrees, but I actually finished far cry on a long playing session. Seems to,me that gpu temps d,actually cause some,sort. Shutdown after 85 degrees, and i have read here that others have had the same issue with the r2 and 7970 combination.
Just saying

 

From what I've read its the Dell 7970M's that have this problem. The Clevo cards do not do this.

 

I am not a 100% sure but i from benchmarking (what i do ALLOT ) i have pushes my Dell 680m beyond 93 degrees celcius and i think it's Svl7's mod that initiates thermal throtteling as appossed the thermal shutdown.

My cpu is a different story i have reached temps @103 degrees celcius and that when thermal throttle starts.
So in short, your hardware can stand allot of heat (be it for a short time) and shutdown with a windows error message? (Is that what you had) usually indicates lack of power from the PSU the anything else.

Either way, nice that you can play the games the way they are supposed to haha

 

The shut down i experienced was basically the computer powering down. like pressing and holding the power button (no windows message or BSOD)
Masybe i should have ordered a 680m isntead of the 7970... but then again, if i can keep my temps below 85 degrees, i am good

 

Perhaps you can mod the bios into going higher.
Svl7 also has mods for the 7970m.

The 7970m is not really worse then the 680m.
The 680m has more potential and better driver support but is also costs allot more then the 7970m soooo...it always comes down to what team you like best and how much chow you have in ye'r pocket

 

Since the Dell cards aren't able to ramp up the GPU fans according to load, the BIOS may detect that the fans are still in "low rpm mode" even though HWinfo is overriding the fan control? So the BIOS detects temp at 85c and fans still in "low mode" and shuts down as a safety precaution?

I'm just thinking out loud here, maybe the creator of HWinfo can shed more light on how it controls the fans for more clarity and if the changes to fan RPM would be visible to the BIOS/vBIOS.

 

I highly doubt Hwinfo will determine bios/hardware behavior...it's possible but to have that easy access to the bios seems unlikely.
All it does it respond to temperature readings

 

No, I didn't mean access to the BIOS. It has access to the controller that sends the pulse to the fans. It overrides the command that the controller is given by default to increase the pulse intervals to speed the fan up according to the levels you set in its(HWinfo's) settings.

I was just wondering if the BIOS sends the shutdown command because it see's the temp at 85c and according to its "fan speed tables" the fan is still spinning on low RPM and it initiates shut down as a precaution (It isn't aware that HWinfo has ramped up the fans to drop temps)

 

As far as i understand, no, the GPU cannot send command:"Thermal shutdown", only:"Thermal throttle"

 

The GPU is the cause of the shutdown. My M15X with a 7970M today started shutting down too. My 7970M reaches 79-80C then shuts off. Monitoring my CPU temp it only gets up to 71C max on one core so it definitely has to be the GPU. Same goes for the OP. I'm going to try and repaste, redo the thermal pads and tighten the screws a bit more. My laptop has been working with the 7970M fine for the past five months, no shutdowns. I'm guessing it's the vram overheating since the GPU should shut down at 85C not 79-80C.

On the M15X/M17X-R2 with a 7970M it is possible to have shut downs due to higher GPU temp.

 

I can confirm this behavior in my m17xR2 as well. at 85 degrees the GPU will send the kill command to the system and the whole computer will shut down just as if pulling the plug. the only way to bypass this is to have a real good paste job and have a HWInfo Gamer profile to kick the fans in high gear starting at 50 degrees (about 3500 revs).
It could also be the fault of using a combination of a XM chip with these high end GPUs. Other users who dont use XM chips get allot more forgiving behaviors from their custom GPUs than we do. Some people actually get the 7970ms in Crossfire! Those 10 extra watts mean something...

 

well, as i see it, this AMD thermal shutdown is a nasty thing.
Strange that it does not throttle the card down to keep it cool instead of cutting off the power....weird behaviour.

All the more reason to buy a 680m

 

Its only on the Dell cards afaik. The Clevo cards don't exhibit this behaviour. So its not an AMD issue per se. Funnily enough, the Clevo's are the one's with auto fan control and I think that may be the key to resolving the issue using HWinfo. But that might not be possible as it may involve making a value available to the vBIOS which I don't think HWinfo is able to do.

I would prefer the 680M as well but with the 7970M we have an option of adding another down the line since AMD's dual-GPU config doesn't require certification for the cards in the BIOS. Still... its amazing that this beauty of a machine is still giving us upgrade options

 

certification??? what are you talking about...Dell cards, clevo cards....if you have an R2 with A10 BIOS...just cram 2 680m in em and go SLI....nothing holding you back except for the 240W power limit.

 

I wish it were that simple

SLI requires certification in the BIOS for nVidia cards. The Hardware ID's of those cards supported for SLI are stored within the BIOS. We have certification up to 285M's if I remember correctly. There was a work around using the hyperSLI patch to get the 580M's to work but I don't think anyone was able to get the 680M's to SLI.

 

I think about a handful of users with an M17x R2 have a 680m SLI setup here.
also, almost 50% of the users with a GPU BIOS/Volt mods their card so i don't think that the custom bios is a real problem.

Drivers are also modded, so you got to explain this to me

 

Unless I missed something I don't think any M17x R2 users have 680M SLI, maybe you were looking at the M18x guys?

No, it doesn't have anything to do with the GPU BIOS, its the motherboards BIOS that has the certification that determines what cards you can in SLI. We might be going off a bit topic here so I'll PM you some useful links to explain in more detail once I get the time.

 

Ah ok, i see what you mean

 

What are you talking about?

The 6970M in my R2 hits around 87*C all the time and never shuts off.

Something else is straining in your PC methinks...

 

When I had 5870Ms in my R2, they could go to >100°C without causing a shutdown. From temperature monitoring with my 7970Ms, they never hit 80°C, and only cause a shutdown when overdrawing power in Crossfire (still haven't found a solution to that!).

Have you tried forcing the GPU fan to constant max speed with HWiNFO?

 

Sorry this is a bit off topic but are you running them with the 240W PSU? Clevo or Dell cards? Do you have an XM CPU? Have to tried dropping the voltage/clocks on the vBIOS to reduce power consumption?

 

Like i said before, a GPU cannot initiate thermal shutdown, only thermal throttle.
Instant shutoff is usualy caused by insufficient power draw

 

It's with a 940XM. I've flashed them with an undervolted BIOS and run the CPU without Turbo enabled: If Crossfire is enabled and providing a performance gain (i.e. I get 45 FPS on Ultra in Crysis 2 with one card, and it increases to 60 with two), it will shut off. Even with the cards underclocked, in addition to being undervolted, with the CPU running at a slower speed as well. I can play some games (like Skyrim) with Crossfire enabled with no/infrequent shutoffs, but that's only the case for games that can run at/near 60 FPS on one card to begin with.

This is with a 330W adapter modded to use the ID chip from a 240W, which doesn't seem to make much of a difference at all.

 

Wow... thats an incredible amount of info for R2 users. Thanks for the detailed response, its much appreciated. Although you just burst my upgrade bubble I was also planning to undervolt and underclock with the 330W. Guess the only way to run two cards then is the dual PSU mod which isn't very practical.

Are your cards Clevo or Dell?

 

An extra 90W would cover it, it's likely more of a motherboard limitation.

 

Yeah I agree another 90W should suffice. Chances of getting a higher wattage adapter at this stage from Dell is diminishing though considering the new fabrication processes and lower power requirements. Hopefully AMD's newer range of cards draw 75W like the 5870M's.

I don't think the motherboard is a limiting factor since StamatisX measured a total of 16A draw from his dual 240W mod with 7970M's in xfire with no shutdowns. That equates to around 320W if I remember correctly. Thats 320W directly from the PSU not taking into consideration the efficiency of the PSU. So the 330W by default isn't going to cut it unless you do some sort of power reduction mods which by the looks of .Camerons post isn't helping

edit: For more detail read from StamatisX's post here and the response by iamsolidstate who made the ID chip transplant mod.

 

I think they're Clevos, but I'm not really sure. I bought them from MXM-Upgrade and they came pre-flashed with the Clevo VBIOS.

If anyone has an idea for something to try, let me know. I'm in Europe now, but when I'm home again I think I might put my original 720QM back in to see what happens with that configuration.

Sorry to disappoint, but in the end I think you'd rather I save you the anticipation and expectation of it working after going through everything! If you want full technical details on my painful experience trying to get it to work, check out the thread here.

 

An easy way to tell would be if the fans are on auto or do you need to use HWinfo to control the fans? Clevo cards come with auto fan control, Dell's don't.

I think the 720QM idea is great cos I've heard these XM chips are power hogs. Theoretically you could mimic the 720QM with Throttlestop by dropping the TDP and multipliers but if that doesn't work the 720QM would be ideal to test.

No worries I really appreciate the info you've brought to the table and I'm sure anybody who was contemplating the exact same mod would find it very helpful. Thanks for the link, great read.

 

Ah, yes, it probably is a Clevo then. I use HWiNFO for fan control, but it seems to be adjusting the speed even when it's not running.

I think the lowest I've run the processor at while testing was 16x (versus about 27x on Turbo), but I feel like putting the 720QM back in would put a much stricter ceiling on its consumption. Knowing my luck, I can say with a lot of certainty that it won't shutdown with Crossfire and the lesser processor, since it'll probably bottleneck the two 7970Ms and make it a pointless solution.

 

Maybe try dropping the TDP as well? As that determines how much power the CPU can draw before dropping its clocks to stay within the power envelope. I agree with it being pointless though, two 7970M's is overkill for a lower end CPU. That said, the 840QM is very simular to the 940XM at stock except for the TDP of 45w vs the 55w of the XM.

Try your current CPU at default multipliers and turbo boost but drop the TDP using Throttlestop to 45W if you haven't tried already.

 

Depends on the resolutions and details to be honest.