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

Well I am back to CF 6990s again. Didn't feel like selling the second card after getting the replacement so I threw it in. Gonna run it until it dies. Which is inevitable considering my luck with the second card in this laptop. Temps are good so far though maxing at 82 after 1.5 hours of BF3. Used 1mm pads this time and noticed better contact on memory. May undervolt when I have time. But BF3 ran smooth last night with only 2 noticeable small drops in FPS.

 

Definately undervolt. It will treat you well.

 

Do you think it will fry a card if I don't undervolt? I am curious to know what has caused so many of my second cards to fry.

 

It shouldn't, at least not that I'm away of. The main purpose if it is just to reduce the throttling as undervolted cards will draw less watts, but it couldn't hurt.

 

Is the recommended undervolt still .96? Haven't been following this thread religiously so not sure if this number has changed.

 

.96 works for some and not for others. I had it at .96 working well, but was getting some drops in BF3 that were driver related. Before I found out they were driver related I went to 1.0v and it's been there ever since because I never remember to drop it back to .96v. I also run Folding@home 24/7 so my cards and CPU are always at 100% load when I'm not using the system. 1.0v should be good.

 

no it shouldn't fry at stock voltage, if it does it's due to heat or a defective card.
i've had a card fry (also second/slave card) and it was definately the temps.

one of the main issues with 6970/6990 crossfire is heat, its considerably more than the 58xx series was and we're using the same heatsinks. undervolting helps a lot with that, we only need enough voltage to ensure the chip is stable, at stock many have found 0.96v to be sufficient, some require a little more at 1v. i also never play without raising the backside of the laptop (9c battery and notebook cooler to boot). ensuring that the room isn't so hot helps a lot as well.

best way to avoid frying the card, monitor your temps like a madman. i have hwinfo + msi onscreen display server showing all my card temps all the time. i know them inside out.

a quick and dirty way to test is to touch the top of the laptop just above the touch bar on the left and right sides. if the plastic is hot (not just warm) check the temps.

 

so guys i've installed a single 6990m into my m17x r2 unit now. Question now is what do i do with my 5870's? do i just take the second 5870 out or leave it in there?

 

Youre only going to use 1x 6990 ? Theres no need you will keep the slave card still in there... unless you wanna 6990 crossfire to 5870' lol i dont even know thats possible ^,^ !!!!

 

I dont see why why you couldn't have a 6990 and a 5870 they are only one generation apart, and it will improve performance some what.

 

^ Agreed. If 6970/6990 crossfire works, it's worth a shot with 6990/5870.

 

OK. So I am going to undervolt my cards. I have opened my vbios in RBE. It looks like there are 8 clock voltages that I can change. And they are all at different numbers. See attachment below.

Should I change all of these 8 values to 1.0? This will increase voltages on some of the values. Others have said .96 but DR says 1.0 is good so I would like to try this first. If anyone can help, let me know.

Also, is there anything else I need to changes or just these 8 clock voltage values? Never used RBE before but following DR's vbios flash guide. Just want to do it right the first time.
Thanks

 

For me I just adjust the clock info for the ones that are over 1.0v. I bring them down to 1.0v. Since the 715/900 is stable at 1.0v 500/900 will be as well. Defiantely keep the screenshot so you know what stock voltages are if you ever want to go back when you sell the card. You other option is to only adjust the first voltage for 715/900 (clock info 00) and that will do just fine. easiest is to just adjust 1.0v, so really, thats what I'd recommend. Your temps go up most at 3D clocks, and thats whats used in gaming, so I would just stick with that.

 

OK. So I save the new RBE vbios and just use that same vbios for both cards when I use ATI Flash?

 

Hi Ghost,

That is correct. Make sure you have a backup of your stock vbios just in case, but that should be all you need to do.

 

Thanks Wolf. +1. Operation completed successfully. DR I'd rep ya but gotta spread the love first. Thanks!

 

Awesome Ghost!

Hope you have better luck with these cards!

 

No prob, now get your game on! and enjoy!

 

Yeah thanks. Played some BF3 last night. I'm being really cautious due to my history with failing second cards. Trying to keep the temps under 85. Seems the second card hits 83 after an hour or so of playing. Been logging HWInfo each time just to be sure. Hopefully this one holds up.

 

Have a problem I didn't have before undervolting. Seems I keep crashing during BF3. Temps are good according to HWinfo log right up to crash so it's not overheating. Is it the 1.0 voltage change? DR you had said there was a driver issue you had after undervolting, is this similar? Do I just need to try another voltage and see or should I go back to stock? Or update the driver? Current driver version CCC 11.8, Cap 4.

 

Hi Ghost,

I'd try adjusting the voltage first. Some cards are more stable then others at lower voltages. You might try upping the voltage back to the default and see if the crashing continues. If not, then it's most likely a voltage problem.

Then you can try adjusting it in small increments until you find the best voltage that is stable for your cards.

Hope that helps.

 

I had problems with BSOD while undervolting @0.96, changed to 0.98 and all is well.

 

Thought I would put a post up regarding my recent shenanigans with my heat sink modifications.

Am a little pissy and it's late so probably the wrong time write this post

Anyway, was sick of filing so I bought a dremel equivalent and went to town on my heatsink. Focused mainly on the main component sitting at the back of the 6990 between were the two rows of ram meet as this is the main contact issue.

This task was not easy by any means....metal everywhere and I dont have a outside workbench....miss these luxuries from Aus

I dont know if was different for others but I had to file this down a butt load, almost through the heatsink to get it low enough for the heat sink to make better, most consistent contact with the RAM chips.

Took my about 4 hours of intensiveness to get it right, and even at the end it could have gotten lower but I decided it was filed enough and it was starting to come through.

So I put it all together using paste on the RAM chips to give a closer better contact.

I have noticed a massive improvement. At 780@950 I was getting thermal shutdowns in BF3 unless I had the window open. Now my primary GPU sits nicely at around 80 with the same clocks and stock voltage. My second one sit a little higher at around 85 with the same settings. These are in BF3. Furmark still gets too hot for my liking but it takes a while to get there. Acceptable imo.

A 10+ degrees drop is a nice improvement imo, but this mod is definitely not for the faint-hearted.

Hope this makes sense, I will prob wake up tomorrow and face palm

 

Think I'll try .98 tonight and see if that works. Can't go any higher than 1.0 without taking it back to stock.

 

I admire all the work you put into this. Sounds like a 10+ degree drop is WELL worth the effort though. I'm thankful I haven't had the same issues with the heat syncs that some of you all have been having.

Nice work Douse, glad to hear that after all the time and effort it's made a noticeable improvement!

 

Changed the 1.0 voltages to .98. BF3 locked up on me same as before after about 30 mins of playing. I was running HWinfo log. Some things from the log that look interesting.
In the last line of the log, my Core 0 clock dropped to 0MHz and my 2nd GPU went down to 100/150MHz. Temps on GPU2 were 70 max. Going to try changing all voltages to .96 and see if that works.

 

Thanks Greywolf

Always room for improvement! I should have take pics, but its doesn't give it justice and I also couldn't be bothered.

Just had gaming session with BF3 for many hours with windows closed and primary GPU hit 84.5 and seconardy hit 88.5. So happy with results, time for more over clocks

 

.96 failed too. Keep getting lockups during BF3. Flashed back to stock. Hopefully I won't have any troubles with stock now.

Stock voltage is good. Guess I just can't underclock. Which doesn't seem to be a huge problem as I am not getting any major performance drops.

 

you could always try running at 1.05v? every bit helps

 

Yeah, up that voltage a bit.

I couldn't get that low without losing stability. I found 1.02 was my stable under volt.

EDIT:: Latest temps, playing BF3 for about 6 hours, hot room as well. Loving the heat spread on my CPU. Barely any variation between cores

 

I did the same thing with a hand file. You don't want to know how long that took me. I just bought a cheapo $10 dremel and I'm gonna have another go at it. I'll probably destroy the dremel in the process but if it gets the job done then yay.

 

I didn't think anything over 1 would be worth it. But I guess I could give it a shot.

 

Douse, sorry I've been behind on following this thread over the last few days but what was the mod for? Did you purchase a heatsink that's bigger than stock or mod the stock? Also, is it necessary to kill explorer.exe when gaming with cf 6990's (with or without undervolt)? Sorry if I didn't go back and read... Early morning of pt, reading and posting from phone.

 

You will be surprised. I saw a fairly decent temperature difference running at 1.02 volts.

Yeah, it was very painful. Don't be shocked when the dremel doesn't actually make it much easier....

You will find it is OK on one heatsink as you can access the area with the dremel. However on the other it is very difficult and I had to use a combination of the dremel and file.

Nah, same heatsink as everyone else. Some of us have noticed that our cards are running hotter then they should and we noticed that the heatsink actually rests on a component just behind the ram chips. This results in improper contact on the RAM itself, and also can reduce contact with the Die. Turbod was the first to file down this area of the heatsink and he had pretty good results. Both myself and Aarpcard have seen some improvements as well.

I will try to find the post by Turbod regarding it.

Regarding explorer.exe, you shouldn't have to touch it at all! If you are having problems with it crashing due to an undervolt, I would say it has been undervolted a little too much

EDIT:: Found Turbod's post regarding the components hitting the heatsink: http://forum.notebookreview.com/ali...0-your-m17x-r2-single-gpu-crossfirex-263.html

The component designated with a 2 is the culprit.

 

Thanks! Can you please post pics of where to trim down and how much to trim down? I'm hoping I won't need to do this as I'd like to not mess it up and potentially destroy the heatsink, lol. If I can see the pics and figure it out, I may end up buying a secondary heatsink just for the mod (if its needed)

 

Hi man,
So you have to trim it in this place. (red square)
Put the heatsink on the card and mark the place where it touches the voltage regulator, then trim it, say 1-2 mm is enough.

 

Thats correct flingin.

As for the amount to file down, it seems that everyone's heatsinks are a little different. So this will depend on your set up.

The way that I did it, was to install the heat sink dry (with no paste) and get a torch and look at the contact points. This will give you an idea of how well it is making contact and roughly how much you may need to file.

Make sure that you screw it as tight as you would if you were installing it. As the card and heatsink do flex and make better contact.

 

yes, for example i didn't file it at all - i even putted 0.5mm thermal pad on it !
there is some pressure on this element, but in my case it just worked.

 

Yeah, it is strange. I tried putting pads on them and it made things much worse as it increased the gap.

 

Great thread surprised it's not a sticky.

 

Hello guys:

been monitoring the thread for a while. maybe some might rememebr me (i tried to upgrade my 5870's with 6990's from Eurocom, and coulnd not get the 2 cards to work). Anyways, after my failed attempts in december, I jsut decided to ship the laptop to Eurocom for them to do the isntallation.
anyways, tomorrow it should arrive at my house. So, i was wondering if anyone could help me out bu telling me what additional software i should get to make sure my computer runs safe and sound. Like i have read I need to monitor temps... si i am wondering what program to use, and what are the temps i should be aiming for (whast alright, and whats too hot?).
Well, again, thanks for the assistance last month, and look forward to joining the gamer ranks after a month of downtime. (ill post a thread of my experience and a review later in the week)

 

Hi Ricard,

GPUz and HWinfo64 (or 32) are typically what folks use to monitor the temps (among other things). As far as the common temps for the 6990's I think if maxed they don't run higher then 90-93C then you should be ok (for the folks here with 6990's, whats a good temp point to strive for)

Hope it all works out great for you.

 

My 6990s run well under 90C. I usually find that after a few hours of BF3, my second card gets up to 87C on the shader. That is the highest I have seen it. As long as it stays that way, I am happy. I'm also using the 1mm thick Fugipoly pads on the memory and so far I think it has been the reason I haven't killed this second card yet. It's either that or the fact that I think both of these cards are also Eurocom cards even though I originally ordered Clevo. Not sure how you tell the difference. But they have been stable so far.

 

Stock clocks and undervolted you should see 70-80ish.

Overclocked and at stock voltages anything up to 90 and even sometimes slightly beyond.

Temps seem to vary quite a lot from person to person, so hopefully your's run nice and cool

Good Luck!

Oh another thing, Furmark really hurts my temps. As in I get thermal shutdowns if left for too long. However the temps are more then fine during extended gaming in Skyrim\BF3 so I don't really mind that much. I personally think Furmark is a little intense for laptops anyway, but that is just my opinion

So if you use Furmark for testing, keep a really close eye on those temps!!

 

awesome! 90 degrees is top temp. From what I understand, eurocom puts pads on all the memories and components... so that should be good.

One more question, with the mod/upgrade... is the m17x still able to switch between the integrated and the discrete cards... i ask cause i travel sometimes, and I use the battery from time to time. And/or are there any changes i need to notice after the upgrade from my original configuration?

 

LOL, i hear ya on the futuremark. When i tried to install the cards myself, the first time i ran futuremark (with the 3d submarines), the computer crashed on me, and then basically the cards gave up on me... hence why I sent to Eurocom for them to do the upgrade.
I feel that this time around, I will stay away from runing benchmark tests and jsut do plain vanilla gaming. I do not oveclock.... and to help with power consumption, i bought a powered USB hub, so that the laptops psu can breath a little

 

Unfortunately there is no integrated video card with the R2's.

I also bought a powered USB hub to pull most of my USB devices off the integrated ports. Not sure how much of an improvement it makes, but figure something is better then nothing

 

ohh silly e, i must be thinking of the r1 i had before. But I could swear that every time i power up my r2, i get this message stating to disable cards to save battery,or something of that sort.... i do;p not remember the message exactly.
must be the rack (just kidding)

so goin back to my question, if i go through ati control utility and disable xfire (guess that means only use one card) how much battery life do i get? 30 minutes?

 

gaming about 5 minutes
i've heard that even onboard airplanes they don't have enough wattage in the sockets to sustain proper charging (>200Watt PSU) ....

 

I keep my cards @ 1.02. I had some odd freezes happen when I ran @ .96V in the past. I'm having some other odd issues ATM but I think they're hard drive related. When my fiance and I get back from our Costa Rica trip (leaving Saturday) I'll reformat with the new hard drives. For whatever reason I feel like my performance just isn't what it should be. My vantage and 3dmark 11 scores are what they're supposed to be, but in skyrim and sc2 I'm definitely noticing that my performance should be better. Hopefully the hard drives are at fault.