[Guide] Installing AMD Radeon Mobility 6970 / 6990 in your M17x R2 (Single GPU and CrossfireX)

The noise added with the second PSU, is due to my second PSU being slightly faulty - nothing due to the fact that I'm using two PSU's. If I only plug in the second PSU the noise is still there. It's an issue with the second PSU, not the set up.

To answer your other question, a PSU being rated at 240 watts doesn't mean it can't output more than 240 watts. The most people have seen on a single 240 watt PSU is around 270 watts I believe. This is because a tolerance of about 20% is commonly used in the industry when determining ratings for electronics. If you buy another PSU, and it turns out that your first PSU was faulty and was causing these issues then it should solve your problem. If the problem is due to something else completely, then buying a new PSU probably won't fix anything. Ideally, I would get ahold of another PSU just to try it out without purchasing it. Maybe someone you know has a M17x PSU?

You mentioned your gpu's were at 1.02 volts. I can't remember, did you ever try setting the voltage back to stock?

Try disabling CFX and running 3D mark. If it is indeed a bad card, that might help narrow it down.

The error with 3D mark you got seems to be network related to me. Reinstalling 3D mark might fix that. First, make sure the "Futuremark System Info" service is running though.

 

Ok, get back to us when you've reformatted and you're using the new PSU.
Did you try running the laptop without the battery like I asked? Simple Yes or No answer

 

Yea, I did. Also, I can't imagine the cards are faulty when all 3dmark tests minus the last one run perfectly. I don't know anyone who has an m17x to use the PSU. I'm still under warranty. You don't think Alienware will just send me one for free under the condition (they hold the funds) that I'll return the current one? I really hope that between formatting and a new psu, something will work.

EDIT: Should I just reformat anyway? Or should I get the PSU first to rule out that issue? Also, yes - I did try putting the 6990's back to stock voltage.

EDIT 2: For reformatting does this look like a good list? http://forum.notebookreview.com/ali...er-install-order-guide-alienware-systems.html for some reason I thought I was supposed to install the video drivers last. Also there's a bunch of drivers on the dell support website that I have no clue what to do with. Not sure which removable storage drivers to get, not sure which network drivers I need ( I just use the default windows wireless application) and which harddrive is mine for the specific harddrive driver. Hopefully a new PsU will fix the problems but with my luck I'll get a new psu and reformat and I'll still have issues. Thanks

 

Hey guys!

Finally got to installing Greywolf's 6970m in my R2!

I did the following choices, tell me what you think about them:
- use Arctic Silver Ceramic paste on GPU die
- use the allready existent thermal pads from the 5870m heatsinks
- left 1mm of available screw space on all 4 screws

I tried furmark in 1920 benchmark settings and got à shutdown at 50% when temps reached 100°C

Wondering If I should have tightened the screws all the way to the bottom?

 

Hey Tom,

I would highly recommend buying new 0.5mm thick thermal pads that you cut to the size of the dies you want to cover.
Artic silver is great and that's what I used on the dies as well.
My temps never cross 80 without any extra fans; the laptop is flat on the desk.
Try the free vantage or 3D mark 11 benchmarks to see how well they perform in normal conditions.
The temps seem high to me.

 

@tomX,

If i recall well i've used the AS Ceramic to stick some passive heatsink over some mosfet some years ago, on an Abit MoBo (overvolting an XP 1700+ >2.6 Ghz, watercooled ).

If it is the same paste that i recall, then avoid it, it will bound the heatsink with the gpu die and it will be a major PITA in the future to be removed.

look for Artic Silver 5, Artic Cooling MX-4 or other paste

 

I definitely would have tightened the screws all the way. Just tightening the screws an extra half a turn on my slave 6970m saw temps fall by about 7C.

 

Is their no harm in bending the 4 metal pins that hold the heatsink, down all the way (tightening the screws all the way) ?

What's the catch on using the original thermal pads (they seem ok) VS new pads?

 

Hi Tom,

My experience as well that aarpcard mentioned. You'll definitely want to tighten then screws down more. On mine the 4 metal pins are bent to make good contact.

Be careful not to tighten it 'too' much. But you definitely want to tighten it down further to make better contact with the GPU

 

Wondering what the "too" much is..

is it safe to screw them all the way down (with the set of screws you sent me)?

I reckon the only thing that can happen is a stronger contact between GPU die and heatsink right?

 

http://www.amazon.com/Original-Dell-Alienware-Notebook-Computers/dp/B0048HOIIG Does that look acceptable?

 

Hi Tom,

Yeah you should tighten them down so that the x-bracket bends slightly on each corner. If temps are still high, try tightening them down a bit more.

Hi Intense,

Looks like the right one to me

 

I don't think it's possible with the type of bracket and screws we are using to tighten them "too" much.

 

I dont know if this is applicable to you with your eurocom card\screws etc, but...

If you are still using the retention mod then be careful when tightening the screws...

I did not undo my retention mod from the 5870's so I dont tighten them all the way, but i make sure it is flexing the brackets with sufficient pressure all the way around. This means that some screws I actually tighten a little more than others..

 

No retention mod needed with the Clevo cards and screws

 

Cool, just making sure

 

Thanks for double checking though

 

With the screws Eurocom sent me, my screws are all the way in on the two centremost brackets and less tight on the outer one to make sure I have full contact with the thermal pads and that the card isn't bent...

 

Tightened all 4 screws... still 99-100 °C on furmark at 35% of 1920*1080 benchmark run...

haven't changed pads yet...

 

I use 1mm pads and tighten the screws all the way down. First I put them all in just a little to make sure the heatsink doesn't move while tightening and then I tighten down one by one. I don't over tighten, I just go down as far as it'll go. My temps are around 45C and 40C Master / Slave respectively and they never get higher than 78-80C in benchmarking or gaming like skyrim with everything turned up. Those were my numbers a week ago when everything worked. I'm getting a new PSU in the mail tomorrow. Hopefully that fixes everything.

 

If you haven't changed the thermal pads from the 5870M setup, then that's probably the problem right there since the stock pads are a bit too thick, which compromises contact between the die and the heatsink. I would recommeding purchasing .5mm thick thermal pads instead of the full 1mm (too thick).

Oh, and Happy New years everyone! I've been MIA for a bit...

 

Yeah Intense, you have to run .5mm pads on the RAM chips, or you'll be sacrificing contact with the GPU die. I've tried a lot of different combinations. Premium .5mm pads with good TIM, and you shouldn't be overheating. Remember what FurMark's job is though...

 

I've personally run both 1mm and .5mm with mixed results. I will say however that I'm currently running .5mm pads and my temps never go much above 82 degrees.

I also don't run furmark, so I've never tried to stress my cards to the point to see how much higher they can go.

I agree with the other folks here though and would recommend that putting the .5mm pads on the memory could help significantly.

Also what method did you use to paste the card? Personally, I had to apply the paste a tiny bit thicker on the GPU, then I did when I replaced the CPU
to make good contact.

On a side note

Hi Ichime...welcome back!

 

Welcome back ichime!

 

I had the .5mm pads on my cards when I ran out of my 1mm sheet because of reseeding too much to figure out problems. I get 5C better temps on both ends with the 1mm. As long as the temps are good I should be fine right?

 

Since you're out and about, thank you ichime

 

Farmer you don't have a second psu do you? Do you have everything working fine? What's your undervolt? I had nearly 2 weeks of perfectness with 1.02 until ( assuming hopefully so that tomorrows package will amend it) my psu went bad.

How high does your wattage get on your psu? Now that I realize that even without overclocking the wattage gets to around 260 or so I'm contemplating getting a second psu just to make sure I always have enough power.

 

Aarpcard I am curious as to how you dont get throttling with your GPU's clocked up like that as well as running a 920?

At my slight GPU overclock @ 1.03v I get throttling in BF3 unless I drop my CPU TDP\TDC to 50 watts with stock multipliers forcing turbo. I also cannot have my monitor brightness on full, which isn't totally a bad thing as prolonged gaming on full brightness gives me eye strain.

You mentioned that you could handle around 280 watts before throttling? Maybe you R2 has a little more give here as mine throttles when it goes a slightly over 270.

Where's that BIOS update!?!?

 

I thought two psus is supposed to prevent throttling?? Also if you left everything at stock would you get any throttling with 1.03? I was fine a week ago with everything at stock + 1.02 so hopefully the new psu will fix it again. I'm just curious about the double power supply because if it eliminates that potential for throttling id do it just so I can rest easy knowing it won't happen again. From your post, douse, it seems like that second psu is worthless if you're still getting throttling.

 

Having dual PSU's will not remove throttling as this is done by the laptop itself in an attempt to maintain power consumption at 240 watts. The reason for this is that the laptop still believes that there is a 240 watt PSU attached due to the ID of the power supply. The BIOS will not recognise anything higher.

You can see this yourself if you look in your BIOS and it will say what type of adapter is connected.

In real life though, the laptop throttles at higher value than 240 watts. So it actually still stresses a single PSU and this is when the PSU can trip to try to prevent any damage to itself. Worst case scenario it may even become damaged depending on its health. Say for example the PSU had a pre-existing condition that was not a problem until it started to operate at its full capacity and higher. It could theoretically start to malfunction.

However, most of the time it will trip and you will know when this happens by the laptop suddenly turning off AND the blue light at the end of the PSU will also be off.

So in effect having two PSU's eliminates the stress on the actual PSU’s when exceeding 240 watts as the power draw is roughly shared between the two.

However, until a BIOS update is released to address the issue (By The Wiz and if it is even possible, we don’t know) the laptop will still throttle when your power consumption exceeds approximately 260/270/280 watts depending on your laptop and its limits.

 

Thanks. I think my PSU got damaged because everything was working great a week ago and now it's not. Hell my 5870s made the computer shut down while benching. Not sure why my 6990s aren't.

 

Also I think the Delta branded power supplies are meant to to be better then the Flextronics ones

 

from my experience i have a flextronics power supply and a dell one. The flextronics one is making a soft sound when in power. The Dell one doesnt. But it could be just it was not glued accurately. But it still works perfectly.

 

Yeah it is one of those things I guess. All up to luck and quality of the components used.

 

I am currently undervolted to 1.05V on both cards and my 920xm is at stock. Single PSU.
Powerdraw may go as high as 250W in some instances but I haven't had a system crash in about 2 weeks now which is a good sign that the laptop is enjoying its upgrade as much as I am!

Edit: Attached screenshot will show you computer temps after running Metro 2033 for 30 minutes at 25 degrees celcius ambient temp with the laptop sitting flat on the desk. Game settings are DX11 High with 4 x AA and 16 X MLSAA. Average fps = 70

 

Imho you are running to hot.

CPU, not even overclocked it is not supposed at all to run like that in a game, even metro 2033. Also your cards should run way cooler than that.

If you don't wish to repaste at least rise the back of the laptop with some caps, book etc..

 

Raising the laptop will make a huge difference.

With my system as is, I wouldnt dare run it without a cooler or at least being raised

 

I'm using an 820qm though - not a 920xm. The 2.12ghz on the cpu is the result of an obscene fsb OC. For normal gaming, I run the 820qm at 2.03ghz

Here's the news on my power draw. Just tested it.

With the clocks in my sig on the gpu's, furmark will pull 288 watts. I tested it for 10 minutes. No throttling. None. It hovered between 285-288watts and stayed there. No throttling what so ever. I've also observed the power draw in games like Metro to be a stable 280+watts without throttling.

Next I ran 7 threads of Prime 95 and furmark at the same time - again with the clocks in my sig. The power draw peaked at 340 Watts! It stayed there for a few seconds, then throttled down to 240 watts. In a few seconds it vamped back up - but only increased to 318 watts and stayed there without throttling until I was satisfied it was stable.

So it seems to me that my system definitely won't throttle at 288-290 watts with the upper limit being perhaps 318 watts.

That is a crapload of current.

 

Wow, maybe I will pursue my 330W adaptor idea. Then I can keep the CPU OCed with Throttlestop and the GPUs OCed +1 for the testing.

 

WOW!!!

Congrats on those results!!!

I definitely do not have that sort of power at my disposal. I can hover at max around 270-275 watts in game with the occasional throttle (once an hour etc)

Man, if I were you, I would be stoked!!!

LOL, I am beyond jealous

 

Nice findings and thanks for your time performing the test, +1

I asked Greywolf to perform some test too, since i don't own a watt-o-meter and i'm unable to purchase one where i am at the moment, but he has my same config. He has been so kind to do as i've asked him. Well the figures he has got are that with CPU @133x20 (2660Mhz across the 4 cores), 68W TDP, both cards clocked @715/1000 0.96V it stay withing the 240W range.

Here are the figures he has got, cudos and rep to Greywolf for the findings

While running 3DMark11 :

AVG : 208WATTS
MAX : 258WATTS (a few peaks wasn't sustained)
LOW: 112WATTS

While running Vantage

AVG: 130WATTS
MAX: 231WATTS
LOW: 114WATTS

 

I got a new powersupply and reformatted and both of those didn't fix the tearing. Switching the xfire cable, however did (as per greywolf's advice!) Hopefully the PSU will allow me to game without throttling, but I haven't put any games on the newly formatted drive yet. I'm working on that now. I at least got through a 3dmark 11 test with no tearing.

Douse, didn't you have a serious issue with overdrive? Now, for whatever reason, Overdrive is in my CCC under performance and when I click on it, I have to accept the terms of service I guess to activate it. I assume it wont do anything as long as I don't click on it, right? I don't want any sort of OCing.

Also, this driver is on the dell site for my m17x r2 and it won't allow me to install it. It says my computer doesn't meat the pre requisites or something like that. Do I need it for Turbo Boost? Drivers and Downloads | Dell [United States]

In http://forum.notebookreview.com/ali...er-install-order-guide-alienware-systems.html it says nothing about installing that driver. So I'm really not sure. Thanks.

 

Hmmm about the turbo boost driver...I'm not sure what that is (or if you need it). Do you have anything in device manager still showing up as unknown? If not then I'd leave it up to the folks here as to what they think.

I'll look at mine and see what that is, isn't ringing a bell off the top of my head.

 

So I have two unknown devices. When I click properties on them under general and location one of them says on Intel(R) PM55 Express Chipset LPC Interface Controller - 3B03

The other says on PCI Bus

 

Dude, I told you to check the CrossFire cable ages ago, HERE.

 

It'd be cool to see that come to fruition. I half-heartedly tried taking apart the casing of one of my PSU's. I gave up after a bit because I didn't have the right tools handy and I wasn't confident I could do it without damaging the case.

But if it were possible to replace the ID chips (which i think it is - probably rather easily too) then that would be the ideal power solution!

Thank man

I think that throttling at a certain power draw is reliant in some way on your video card. When I had my Dual 6970m set up with the Dell primary, it would reach 270 + watts of continuous power without throttling. If it went above 282 watts, then it'd throttle guaranteed. Now I'm using the 6990+6970 configuration and as far as I can tell (done some more tests) it's fine up to ~318 watts. Very interesting indeed.

 

Guess what? After a new PSU , a new xfire cable, and a reformat I HAVE MY PERFORMANCE BACK! SC2 is the only game I have installed so far, but even with 4 maxed armies colliding on ultra / ultra I don't drop below 55+ fps, 55 being the lowest I go (I stay @ 60 tops because of v-sync).

Hopefully I can get skyrim and portal 2 installed tonight to check them too, but I'm sure it'll be fine. Thanks for the help guys but I am back in business!

 

Nice! Glad everything is working well now!

 

Awesome!!! Congrats!!!

 

For Tasselation and Cat AI, what settings do you all use? Also, what about tripple buffering? Whats the deal with that? Ever worth using?