M15x & 6970M / 6990M - What you need to know - upgrading, problems, solutions

What does Memio monitor anyway? It seem like many people have problem with it. Mine is always like 6 degree lower than other.

 

hello,

I've flashed the vbios with the dell one (first page), and installed catalyst 11.7. Here's what it looks like.
http://gpuz.techpowerup.com/11/09/17/2fv.png
Then i've run furmak: shutdown after 25 sec when MemIO reached 86,5. I've run another furmark (stopped before shutdown)
http://forum.notebookreview.com/attachment.php?attachmentid=70614&stc=1&d=1316267618
HWinfo32 tells the 4Gb RAM's speed is only 667Mhz(pc 10600)

 

Looks good, though your min CPU temps seem to be on the high side, what are your idling temps for both CPU and GPU?

The RAM speed gets reported correctly, it's DDR RAM - double data rate, so 2*667MHz equals 1333MHz.

 

i reboot the computer just after the shutdown, that's why idling temps are high on HW32 screenshot. Usually, idling temps are : CPU 50, GPU 46

 

Ah, ok, then everything seems to be fine.

 

So what should i do now? I can't play crysis more than 1 minute before computer shuts down. Is this card defective?

 

Flo, just to be sure, do I understand correctly that the computer shut itself down after 25 seconds with the Mem temp at 86.5C? You did not intervene?

 

Yes that's it. I did not intervene

 

Now that's weird... did you already try to repaste? It's possible that some areas of the die aren't properly covered with paste.
Do you happen to have a picture of the die?

 

I honestly think its the thermal pads. Can you take a picture of how you have placed the thermal pads on the card? The issue definately seems like a heat problem, but you said you repasted already so im guessing you havent correctly placed the thermal pads

 

That is bizarre. I cannot figure out what could trigger the shutdown.My guess, mostly because no alternatives occur to me, is that either some registry setting is the cause (although I do not recall any program that sets a temp-based GPU shutdown point) or that the 6970M is drawing more power at the shutdown point than the PSU can handle and the GPU temps are not related at all. I would suggest a clean reinstall of the Cat 11.7 drivers after removal through Control Panel, a safe mode Driver Sweeper session and a CCleaner registry cleanup. If you have any questions about the process, let us know, though you sound knowledgeable about these matters. If you have another compatible PSU available, you might try that as well. It's possible that the extra power demands from the 6970M are exposing a pre-existing defect in the PSU. I'll try to come up with some better suggestions, but for the moment that's all I have. Sorry for your troubles.

 

I think its your copper heatsink. I had this same problem with my old Sony laptop. In those copper tubing is some fluid and thats burned out due to high temperatures over a long period of time. Try changing the copper heatsink.

 

I can't take a picture. I've put thermal pads on every chipsets (with old 260m's thermal pads and new ones). The thermal paste covers the gpu.Can someone post a picture of what it should look like?
@Revelator: I tried catalyst 11.5a , dell drivers, 11.7 with a clean reinstall, shutdowns occur evrytime. I haven't another PSU

 

Flo, I ran a more or less duplicate effort with Furmark for 30 seconds to get a sense of normal temps at that point. At the 30 second mark, my 6970M temps were all at ~70C, and we both started from approximately the same base temps (low 50C's). Clearly, the GPU cooling needs attention, but that doesn't explain the shutdown. But I also note that at the 30 second mark, my CPU temps were still in the mid 50's (using the hotter running 920xm), while yours were already in the low 70's (1 core at 79C) on the short run. CPU temps can escalate very rapidly, much more rapidly than GPU temps, and trigger a shutdown at lower temps (100C generally, not certain about the 720qm). Since your fans are at max levels, I suspect CPU temp may be the culprit. However, I cannot imagine why it would reveal itself only after the new card was installed. Every theory so far has a hole.

Plus -- when you have a forced CPU shutdown, it is usually accompanied by a series of beeps, which you do not mention. Further, Dells usually throttle at about 85C to prevent temps from reaching the critical point.

I give up for the moment.

 

Ok Rev', i'll try repasting the CPU. Can someone show me what the GPU's die looks like when you remove the heatsink? Thank you

 

Dell 6970M:

 

Hey flo i have had the same problem as you since i got my 6970m.

My computer can randomly just shutdown on me when playing crysis/entropia universe. And i have tried everything, repasting 20 times. I tried many variations of RAM memory which just delayed my shutdowns.

And then i started getting the "A clock interrupt was not received on a secondary processor" BSOD.

So far i havent had any shutdowns at all when i just use "System Auto" fan controlling for CPU Fan and GPU2 Fan in HWiNFO fan control. I just use Custom Auto on GPU1 Fan ONLY. So it seems its HWiNFO that is messing up my system.

And now on to another problem, when my CPU fan runs at say 4500rpm at the same time as my GPU fan. I can feel with my hands and notice that the GPU fan side clearly blows out much much more air than the CPU fan side. So it might be my CPU fan that is defect and can on some rare occasions now give me random shutdowns.

Still my cpu temps only reaches 70-80c on very rare occasions (read very very very rare) which shouldnt be bad for my cpu, Usually its at 60-70c (all temps on load). I have a theory though that since HWiNFO reports cpu fan running at say 4500rpm but it feels like its running at 2000rpm if i feel it with my hands. Hence the motherboards own fan monitor senses there is something wrong with the fan and shuts down system to prevent damage. Is this a stupid theory?

 

@rev': I meant a picture with padded card .
@MrValle: I'll try the HWinfo auto fan, thank you!

 

@flo6813: Doesn't your card have automatic fan control? It's a Dell card after all.

Check out this post here. There's a pic which explains which areas are most important to be covered, make sure to read the description as well, you don't need pads on the heatsink and on the card

 

Thank you svl7, that's the way i put the pads. My card has automatic fan control,(so it's a dell one) but strangely when i flashed the vbios (to the dell one in first page) an error occured (command atiflash -p 0 Dell.bin) and i forced the flash (atiflash -p -f 0 Dell.bin).What does that mean?

 

That indicates that a subsystem ID or hardware ID in the r3 VBIOS is different from the stock VBIOS of your card, don't worry, doesn't really matter in this case.

 

I have repasted gpu, put new thermal pads on the vram, repasted cpu.
Here are the temps after 30 minutes (normal computer use, no games)

After 1 hour

 

Those look fine, particularly GPU temps. Have you tried Crysis or Furmark again?

 

Some comparisons between the Overclocked AMD 5850m@ 800/1000 (Stock Clocks are 625/1000) and my AMD 6990m Stock also some other random scores with the 6990m.


Wanted to give people a REAL idea of what to expect from the upgrade and not just Vantage or 3dmark 11 synthetic stuff, though they are included .


I was going to do a whole new thread but people seem to enjoy this thread alot so its probably best here.



The M15x is 100% stable thus far with the 6990m and my lowly 720qm, also i undervolted the card to .975 to bring down the temps and power consumption.

Spoiler

5850


6990


5850


6990


5850


6990


6990 but no pic for the 5850 as i forgot to run the bench prior to yanking it out


5850 DX11 everything on high no AA


6990 DX11 everything on high no AA


5850


6990


6990


6990

 

Great work! Added a link to the first post. What were the 5850m clocks?

 

5850 was 800/1000 as i was finding for gaming the memory was just bringing more heat and no performance in games.

I will add the OC number to my post so people know .

 

Nice boost my friend specially since the 5850 was already overclocked higher than the 5870 clocks that really shows the power of the 6990

 

Hey guys im now getting the same issue as flo is. Running furmark, it lasts around 3:50 with the temp at 88-90 and just shuts down and restarts. I checked event viewer and the error is Kernel-Power event id 41 task category 63. Problem is now is trying to figure out what is causing it, any ideas? My card is also the exact same one flo has.

 

Just had a brain fart, i was gaming for a couple of hours on friday night and had no shut downs, then last night i flashed the vbios and today im getting shut downs gaming and benching

 

Just ran furmark again, but this time i got it to run for just over 5 minutes, but i have noticed that each time it hits 90 degrees is when it shuts down. I thought 90 degrees is a perfectly acceptable temp for a video card, and have no idea what could trigger the shut down at a certain temp. Im stumped

 

Exactly the same problem! When temp hits 90 degrees, computer shuts down!

 

Just done a repaste and ran furmark for 12 minutes and no shut downs due to the temps staying below 90. I still need to find how to solve the issue, but for now hopefully i can game in peace.

 

I cannot game in peace! MemIO is the highest temp, shuts down when the temp hits 86,5. Should i try to undervolt the card? How do i do that? Has someone a dell vbios (like in the first page) undervolted? Could post it?

 

Apparently there is a fail-safe in bios that when the gpu temp sensor reads a certain amount it cuts power. Im using A08 still and wondering if i flash to A09 will it fix the issue. Might give it a try when i get home tomorrow

 

I'm using A09, same problem

 

I've checked windows event viewer: error is Kernel-Power event id 41 task category 63 for me too. The software "tune up utilities" tells peripheric NVR0FLASHDev (from ntune which is not anymore installed) doesn't work well.
What does that mean?

 

That's a nvidia system tools file. Make sure everything nvidia is uninstalled from your system, then run driver sweeper in safe mode to get rid of the rest of the NV files.

 

I had already done that, so that isnt what is causing the problem. I cant for the life of me figure this out, nothing happens in the event log prior to the kernel-power critical error. Ran furmark again today and the temp got to 90-91 for a few seconds and just shuts off.

 

I've gone 93c on the MemIO and 89 on the core without shutdowns or BSOD. When you do resolve this bug do update. I'm currently looking into a copper heatsink mod to bring down the temps. Will update when I get the stuff to do it successfully.

 

Im not really sure if this is relevant, but i just took a screenshot of trixx and open hardware monitor, trixx is showing voltage as 1100, which i assume is 1.1V, but open hardware monitor is showing 0.900V. Would this have any effect on the card if in fact it is only running at 0.900V? Just thought since it is a kernel power issue, this is the only thing i can find that is related to power issues

 

After you game, the monitor should say 1.1. It is only saying max 0.9 because that is the max it has read since it loaded. Open up furmark or something and you'll see it go up.

 

To those of you who may have heat issues, I had this too and I realised that my heatsink wasn't making proper contact with the die. This was caused by one of the voltage regulators pressing against the copper heatpipe. I did a small mod:

Which gave room for the regulator and allowed my heatsink to sit properly and help my temps by about 10C.
I recommend buying another heatsink so you have one spare.

 

Im done with it, i've tried absolutely everything tonight, i even managed to keep the temps down, but now it just cuts out at about 83 degrees where it seemed to be going to hold at. Tried uninstalling things, checked all drivers, nothing has worked. Im just about to chuck my 460m back in. Im pretty sure the card is just faulty, big waste of $350!

 

Have you tried my suggestion I posted just before your post? May work

 

hi DAv0415,

Where do your card come from? Mine's from waresphere. People tell me in a french forum that the psu's the culprit. Could this one work?
Alienware Area-51 M9750- Power Adapter- 0415B19180- Adapters- Alienware Laptop- Notebook Parts- Splus Direct

 

Hey bud, yeh i checked to make sure the heat sink was sitting flush. I got the temps down by using thinner thermal padding, but then it was just cutting out at 83 rather than 91.

Yes mine is the waresphere one, i've used both the 150W adapter and the 240W adapter, made no difference at all

 

So you flashed another BIOS already on the card ?.

Reseated the card card in the slot ?.

Did you pad the back of the card with some type of thermal pad ? the flip side of the card .

Do both people having problems come from a 460m ?. Maybe the retention bracket is causing problems somehow.

 

The card comes with the proper 6970mbracket. Waresphere told the card was "brand new" but there was thermal paste on the die, so it's a second hand. I do believe the card is faulty.
@dav:Yours from waresphere? Have you written to them?

 

You may have uninstalled it, but if you're getting that error, there are still files left over. The problem could be the old nvidia files are interfering with the thermal controls.

I would uninstall your ati drivers as well, update sweeper and clean everything out in safe mode again. After that, use ccleaner to look for and remove reg keys. Or better yet, if you have an old hdd laying around, swap it in and just run a base windows install. That would remove the possibility of software issues.

 

When i go into safe mode i cannot do anything. Cannot run CCleaner either.