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

No shut downs while playing Swtor. I use to get shut downs every 5 minutes in Swtor, 1 hour later still no shut down.

 

Good job apoc2k8, on a successful heatpipe mod.

Hey johnny did u get proper screws with that backplate or are u using the original screws

 

Original screws, but it's a dell plate... they should be the same, no?

 

Thats what im running, 6990m backplate and original screws this alone will not provide proper contact, you will need eclips or cut the stands down or switch backplates.

 

Thank you.

 

My search function seems broken so I hope this hasn't been addressed. How thick do the thermal pads need to be for the RAM? TIA.

 

Ya, I plan on filing down the posts. I'm going test contact first and then go from there; I also have copper shims.

edit: Pulled some screws from my dead w870cu, heatsink fits really tight now. Just need to figure out where to place pads and such, then put it back together.

 

Widezus heatpipe mod definitely helped in my case.
I had reboot/shutdown very often until I did the mod, and after it almost never and it has become entirely game/situation dependant and corresponds to a game related bug now(map/save skyrim shutdown bug).

Now I can play Skyrim for hours without shutdown, and the metro 2033 test proved to me that the hardware is stable. I can also run benchmarks and things like this without shutdown.
Temps do get very high on occasion, highest I've seen sofar was 94C playing metro 2033 on Dx11.

 

Holy cow, now I noticed something really bizarre in GPU-Z...
it says it's running at PCI-E x1 , what does it say for you guys?

 

hmm saw a screenie from someone with a M18x with gpu-z saying 8x...
HWinfo32 says "PCIe v2.0 x16 (5.0 Gb/s) @ x1 (2.5 Gb/s)"

 

How are you guys running metro on your cards dx11... my card runs that at 10-20fps? How do you do it?

 

here is my screen shot and some info too

 

I'll upload my screenshot once I'm back home. By the way Mexi00ls one is 6990m, so can't compare yet. Mine is 6970, so if it shows the same as yours, then i guess it's normal (i hope).

My heatsink mod sure needs to be reviewed again, Should i hammered all the way down flat ? because i just only made a little bump on that area.

 

My copper part is flush with the heatsink, I think as long as there is no contact with the card in that particular area it should be fine.

My GPU-Z reads PCI-E 1.1x16

 

Ah, now this gets interesting, my bus interface shows PCI-E 2.0x16 @16 2.0

mine is 6970m bought from Ebay - Eurocom with no soldered post. no backplate given

oh well, this means that bus interface has no effects on solving shutdown.

 

Specs as of writing this: 940XM | 6970m | 16GB Corsair | 500GB XT

I spent the last two days in my office working on this thing. I've got it setup now where there are no shutdowns or throttling of any kind. Everything is 100% stable. For me, stable means it passes OCCT's GPU test in DX10 mode with shader complexity set to "8" and error checking enabled for 1 hour. Clock monitoring was done with the newest GPU-Z and voltage monitoring was done with the newest Hardware Monitor. Once I was finished with everything, I loaded up Skyrim. I went to Dragonsreach and looked out over Whiterun, left it running from that position overnight, and woke up to the game still going.

When installing the card I did widezu's heatsink mod and I used 1mm Fujipoly pads on the ram and most chips on the top of the card(see pic). I didn't put anything on the bottom of the card and I also used a 1mm copper shim on the GPU.



The next thing I noticed was a large gap at the bottom of where the GPU fan meets the body. There's enough space between the fan housing and the heatsink fins to allow that to hurt airflow through the fins. I took a two pieces of foam padding from my W870CU, cut them, and glued them to the body in the center of the gap.



I also used electrical tape across the top to seal the gap there; this made a big difference in airflow and noise.

Next: Undervolting. I used Ati Tray Tools 1.7.9.1573 and Catalyst 12.1.

This was trickier than I thought it was going to be and took up the most of my time. First off, the voltages that are allowed and work are steps of 0.5v between 0.8v and 1.1v. Voltages between these settings round up; I tested this extensively. A value of 0.91, 0.93 or 0.95 all spit out errors at the exact same clock speed and time frame. The voltages of 0.8 and 0.85 work, but not 0.75 or lower.

The max stable core clock for each voltage on my card:
0.8v = 440
0.85v = 520
0.9v = 580
0.95v = 650
1.0v = 715
1.05v = 770

Testing all of that took about 10 hours, or 14 episodes of Murder She Wrote on my netbook. The next step was to reflash my card using RBE. Below you can see how I changed the settings.



After I flashed and rebooted, I was surprised to find two things: First, the 100/150 settings was running at 0.9v, not 0.8v. Second, there's a phantom 250/900 0.95v setting that's not listed in the vBIOS that runs in dual display mode. The first is an issue with the vBIOS. The setting of 0.8v is valid, but in the vBIOS any setting below 0.9v is changed to that. I solved both of these problems by going back to ATT. I made four profiles: 100/150, 250/105, 400/400 and 710/900. Now I'm setting my clocks with ATT, but flashing the card with the other clocks wasn't in vain. I noticed at times that card would throttle briefly when the system was under heavy GPU and CPU load. When set to 710 @ 1v it would switch back to 680 @ 1.1v, which actually uses more power. It also did this at times with the settings, and just briefly for the most part. However now, it uses the clocks in the vBIOS, or rather: it doesn't throttle.

I think the most important thing is the shim, which allows the heatsink to sit a touch higher and miss contact with some of the other components. I had issues before I used the shim, even with the heatsink mod. For temps, even in OCCT I don't break 90C and using 100/150 @ 0.8v the fan turns off. Hopefully this can help some of the people having shutdowns.

 

^^^ great stuff Johnny, go to see that you have everything double checked, very interesting modded bvios, good find on the fan too.

 

Awesome stuff JohnnyFlash, I will try some of that soon...

Ony one question: which backplate are you using? gtx-260 backplate?

@Spidert622 I set it to Dx11, Very High, then turned AA&AF down a bit if I remember correctly, and it ran mostly around 30FPS, it did get a bit choppy a few times but some tweaking between driver and game settings should resolve it imo. I think the lowest I saw was 15-16FPS when the sparkly ball thing was flying around killing 15nosalises or so but I was standing still anyway so it didn't matter.

One thing tho, check through HWinfo32 or some other tool that you ARE getting 680mhz core clock... Stealth or battery mode will bring it down to 400 which would probably get 10-20FPS max.
(I guess most here HATE stealthmode as much as I do... Who the #$%$ came up with that?)
Mine pops into stealth mode every so often seemingly at random, and I see this as FPS dropping to barely playabe...

 

Yeah, that's a lot of info there, thanks Johnny! Maybe I'll try this over the weekend.
MW3 free weekend on steam as well haha, anyways I ran that for a bit and so far no shutdowns with that game.

 

I set to dx11 checked my clocks everything and I am getting 20-30fps on very high with aa but sometimes it drops to 5-10fps when looking into light.

 

Damn, I see there's still no fix for this issue. I haven't bothered with it because i've been playing SWTOR with no issues for hours on end however today i fired up Alan Wake and after 5-10 minutes it shuts down ever time. I tried the heat sink mod hammer mod but that didn't fix anything. I'm surprised nobody has figured out what the hell is wrong with this card yet.

 

I am pretty sure shutdowns are heat related as while I was on battery life I set my M15x on the bed and it shutdown after about 10 minutes. Probably wasn't getting enough air flow anyways I didn't have any load on the PC so it's not load related.

Hope this helps

 

@Johhnyflash i nearly stop reading after the first line when you say 920, because that card is in the clear, luckily i read on.

Thanks for your time testing out the possibilty and pin point the issues that might help those (included me) who had shutdown issues. I might take a look at the foam part and the copper shim mod.

 

Have you guys tried undervolting the GPU?

 

I have the Alienware backplate from the M17x R3 with different screws that came from the DVD tray in my old Clevo.

 

Okay guys here's something that might help. I'm actually pretty excited. I've tested it a few times now with the same results every time. Every time i've managed to predict when my computer will shut off and also avoid it.

It has to do with the external monitor i'm plugged into. I'm testing this with Alan Wake since it seems to shut off within 5 minutes of playing it at almost the same exact part in the game while i'm plugged into my external monitor running a resolution of 1920 x 1080. When i unplug it and just run it off of my laptop monitor at 1600 x 900 I had no shut downs at all.

I ran this 3 different times all with the same results. Thinking back the ONLY time i've had shutdowns is when i'm here at work plugged into the external monitor. When i'm at home just running off of the laptop i haven't had a single shut down.

Not sure if it's just because it's an external monitor or because it's the resolution, i'll try running it off my external monitor at 1600 x 900 and see if it shuts down, that will confirm which it is.

EDIT: Okay, it's not the resolution. I dropped my computer down to 1600 x 900 as well as made sure the game resolution matched and it still shut down at the same part. So now it looks like it's definitly something to do with the external monitor or the connection between the two. I'm using a display port to HDMI adapter in case that might be the issue.

EDIT 2: Just tried it again with the same results. No shutdown when disconnected however it shut down 5 minutes into the game while connected.

 

Can you pass OCCT for 10 minutes on the internal screen?

 

i don't use external monitor...

 

Well, from my personal experience an external monitor has no effect on me. I get shutdowns either way.

 

Not sure if i ran it correctly, I selected GPU: OCCT on the left then changed the resolution to 1600 x 900 and checked full screen then hit OK. The test ran for 4 minutes then shut my computer down.

So if i ran it correctly then No, I was not able to run it for 10 minutes. The only thing i'm sure of is that Alan Wake shuts down at almost the exact same part of the game EVERY TIME when i have an external monitor plugged in. It hasn't shut down on me yet when it's unplugged. I've tried both ways at least 8 times each with the same results every single time. I wish there were another game i could try this with that the shut downs were so predictable.

Hel20eS - You don't use an external monitor but your laptop monitor is 1920 x 1080, mine is only 1600 x 900, maybe that's the difference (it's grasping at this point)

EDIT: I just ran the test again this time keeping the resolution as default and not checking full screen. Same results, ran for about 3.5 minutes then shut down. I was monitoring my temps though and holy ! My MemIO got up to 101.5 then the system shut down.

 

Ya. I mean no disrespect by this, but this is why people to need to correctly test their systems. The human eye can't tell if the machine is spitting out errors in a game, unless they're very pronounced. The mentality of "It ran game X for two hours and didn't crash, so it's stable" drives me crazy.

I really do think that the heatsink mod + shim is the answer. I've had 0 problems and my temps are nowhere near 100, even after 10 minutes. It looks like you're just not getting good enough contact.

 

looks like i'm going to try the shim and see if that does anything. Till then i'm going to try running skyrim both on a monitor and on my laptop and see if I have the same results.

After running OCCT it seems like my system is operating like it should be. I mean with my temps getting that high it should shut off right? So as far as that program shutting me down, that's what should happen and it's a matter of trying to keep the system cool. Now if OCCT shut it down at 80c or something like that then i know there's a problem with my machine.

For now, until my machine reacts in a different way, i'm going to go with my theory on this having something to do with the external monitor

 

Are you using ATT to control your clocks, or are you just using the stock vBIOS? The main difference between external and internal is the driver clock settings. It could be that it's downclocking when you run on the internal screen, but it doesn't on the external. If you lock the clock speed manually, you can remove those variables.

 

I haven't touched the clocks with this card. What is ATT? I'll try it if it will rule anything out

 

ati tray tool, works like CCC (catalyst control center), just with extra features like overclocking and such.

 

Hey guys I'm not sure this is abnormal but HWinfo says that my max cpu clock speed is 6775mhz. I find this really weird as my clock speed is 1.73ghz (i7 740qm) Anyways I am just curious because that is is highest clock speed I have ever seen as my max speed.

 

Does anyone have extra copper shims they may be willing to sell or donate? On eBay, they don't sell individual ones, and I don't want to pay the $15 package since I already have MX-4 and would prefer not to buy another bottle of thermal compound

 

Okay so i have this ATT Program installed and i'm looking at the overclocking section. It shows my values as the following

Bios GPU: 680
Bios MEM: 900
Current GPU: 250
Current MEM: 900

Right now i'm running on the laptop monitor. Any suggestions on what to change the speeds to? I'm not familiar with OC'ing and i don't want to F anything up so if you could help me out that'd be great. Thanks

 

Alright, so I tried the foam mod and played BF3 for a couple minutes (no crash) but according to GPU-Z my GPU Temp#2, I don't know which that is, got up to 98C while the others work low 70s - 80s. I think this is what JohnnyFlash told me about but I didn't see it before adding in the foam. How worried should I be about this?

I didn't want to keep running BF3 in case it is bad for my card >.<

I should add that I just used regular foam, does that make a difference?

 

Okay so turns out i am wrong. However, this has forced me to try something else. I tried Alan Wake while at home on my laptop monitor and sure enough, BANG! it shut down as the same part as before. That made me ask myself the question of what i did differently yesterday when i disconnected from the monitor to make it so it wouldn't shut down.

The answer is air flow/heat. Yesterday when i would disconnect it i always moved the laptop to a different location which had better air flow. I just took my laptop and placed a couple of coasters underneath it to make it so the back was about an inch off of the desk and sure enough no shutdown. Without elevating the back, i get shutdowns.

So i guess the shim would keep the temp down some so that might be the answer. Any other suggestions?

 

Just contact him and ask for just one of the thick ones and nothing else. I got extra stuff for no cost, I'm sure he'd work with you.

The easiest thing to start with is set the clocks to the default 680/900, but set the voltage to 1.00v; that works for majority of people. Test it with 2 min of OCCT with error checking enabled.

I still think your contact is bad though, if you're getting temps above 100. The undervolt should help a little with that, but it's not the best fix. Use GPUZ to monitor your clocks, to make sure they're actually staying.

 

Alright Johnny, i changed the setting exactly to the ones you have in the picture and ran the test. In fact i ran a 10 minute test and it passed without any errors. The computer didn't shut down or anything. In fact, the temp only got to 87.5c. Yesterday after 3 minutes i was above 100c.

Other than that the only differences i've done since my last shutdown were that I added about a 3/4 of an inch of space to the back of my laptop for better air flow.

I just tried Alan Wake and it ran perfectly. Got past the shutdown point without a problem.

Should I keep my clock speed like this? It has the option to run it like this at startup, should i check that?

 

Yes, if you keep it set to 1.00v that will keep both power consumption and heat down. Make different profiles and save them, then you can switch between very easily.

Here are the setting I use:



There's more info about it in my big post a couple pages back. Just be sure to test each setting you use before you save it; every card is different.

 

Ordered my copper shim, will post results once I get it installed ::crosses fingers::

 

I've found a new bug: Crysis 2 seems to reset the clocks and voltage to 680/900 1.1v after a few minutes. I don't get any shutdowns or issues from it, but it's weird because I don't have that setting available in the card's vBIOS. I'll look into it more tomorrow.

 

Good work JohnnyFlash! Just one little remark... there's really no need to pad the capacitors (I'm referring to this)

 

Ya... I just had a lot of extra pads.

 

Watching youtube vidoes or flash based contents and my card's temp is constantly around 60-62.

Is this safe?

 

Sure it is. And if you lower the memory speed it is even cooler, resulting in less noise. No idea why AMD decided to set the lower 2d clocks profile to 900MHz memory freq...

 

just an update. I kept the clocks as suggested and tried Alan Wake again with success. It doesn't shutdown on me anymore. Hopefully this issue is resolved but there's no saying for sure yet.