MSI GT683DX(R)/MS-16F2 AMD RADEON HD 6990M Succeed!

So yeah I bought the card from Quadzilla in NBR marketplace and it works nicely.

Booted into windows, installed the catalyst 11.10 preview 2 BC3 - RAGE beta.

Below is a few pics of the 2 cards, heatsink and a screenshot after I installed the driver and reboot.

The stock heatsink in its default place, notice the warranty sticker is kinda of no consequence, you can just put it out and stick to anywhere you want and then put back in later, and it seems that the original sticker was already tampered with.


GTX 570M just pulled out the heatsink, in its original slot:


The heatsink pulled out completely:


The backside, notice the 2 RAM chip space not covered with thermal pad and the themal pad right above it is slightly not in the correct position for the 6990M (that small thingy I dont know how to call it )


Backside fullview:


Behold the GTX 570M! (is there any screenshot of it on the internet yet?)


Backside:


AMD Radeon HD 6990M!


Backside:


GTX 570M vs HD 6990M side by side:


Backside:


I have to pry out the back faceplate thingy off the 6990M with a knife while being very gentle with the card ( though not with myself as I cut my thumb finger ) for it to fit in the slot.
I was going to leave it as it and brute force it into place but the cost of it reminded me to be careful and leave no mistake. Thank god I didnt do that.

About the missing thermal pad on 2 RAM chips, for now I temporarily put thermal paste in them, not sure if thats a good idea. Thought?
And the slightlt out of place grey/brown thingy. Should I do anything about it? I just put the thermal pad on it then the heatsink so it would half contact with the heatsink (i think).

The system boot up nicely, windows show up standard VGA as expected and I installed the catalyst driver right away.
First reboot it show 48*C idle at first then continue to stay at 52-53*C at 90V. Thats not very good so what should I do? GPU observer no longer have the memory load option and funny that the fanspeed now always show at 30%.
Just run Crysis 2 for 10-15’ and it stay at 85-86*C and Im hesistant at playing any longer for fear the card may get damaged.

The first triumphant pic I posted in other topic:



Another pic of Hwinfo and 3dmark vantage result. Hwinfo run while I was typing this and after benchmark. As you can see the temparature is a bit too high, and the score is not that great, what can i do? (just uninstalled the rest of nvidia drivers but havent reboot yet)



AMD Radeon HD 6970M video card benchmark result - Intel Core i7-2860QM Processor,Micro-Star International Co., Ltd. MS-16F2 score: P11580 3DMarks

So yeah my concern is about that miscontacted grey/brown transistor or something.

The missing thermal pad on the 2 RAM chips.

Overall temps.

Any optimization tips? since this is literally my first time using an AMD/ATI card.

Most effective pasting method? Pea, tube, etc,...

Last but not least the way I repaste CPU/GPU is with the material remover – surface purifier then apply the paste, is that correct?

And currently I have the Artic Silver 5 and GELID GC-Extreme (which im using)

Newegg.com - GELID Solutions GC-Extreme Thermal Compound

is it any good? Should I keep it or switch back to AS5 or another brand?

Can you recommend any specific thermal pad for me to purchase?

Thanks for reading guys and if possible a little rep for the effort wouldnt hurt

 

okay rebooted and run 3dmark vantage again and its still at 118xx. lurking around and most are around 14xxx. What should I do :-s

 

On the M15x and M17xR2s those guys never have any thermal pads except for the RAM and when i had the M15x with a 6990 i tried to pad the card as close to my R3 as possible but the heatsink was not designed with that in mind so really im not sure it helps or hurts.

Its been an on going debate between me and a friend about what those backend pads are actually doing for the temps/card etc .

Are they keeping the heat away from heatsink since they are so thick ?.

Are they transferring heat to the heatsink and ultimately getting cooled be it very little since the pads are so thick ?.

Since there are quite a few people running 6990s out there with 0 thermal pads except on the RAM i cannot truly tell you whether long term of having no pads or having them is going to help/hurt the card

For the moment i would not worry to much about that stuff and mainly just focus on getting your temps down on the things you can monitor via Hwinfo.

As for paste on the memory that is fine and i have used paste plenty of times on VRAM but i always use a silicon based paste like Arctic Alumina not Arctic silver. MX-4 or OCZ freeze would be fine on the RAM but i am not sure id want to put say IC diamond on the Vram cause that is stuff is messy as heck..


Most people on the Alienware forums are using these thermal Pads fujipoly 0.5 mm . Some people buy the 1mm ones but most have found the best temps from the 0.5s

http://www.frozencpu.com/products/1...ermal_Conductivity_60_WmK.html?tl=g8c487s1288

Right now i am using the stock ones myself and my temps are great.

 

Did you completely wipe the Nvidia drivers before installing the AMD card ?.

My stock score on the 6990m with the RAGE driver is around 13,400 on the GPU portion.

You may want to run Driver sweeper and make sure all remnants of the Nvidia drivers have been wiped clean and go from there.

Also what does Does GPU Z say ? is the card running in 16x PCIexpress 2.0 . Highlight that area with your mouse and see what the window says that pops up with the cursor hovering over it .

 

I think im gonna pull it out and repaste with AS5 and put back some thermal pad that still got stuck on the GTX 570M I just noticed.

umm not really in a rush I just installed the catalyst driver right away and choose custom to skip some components, can that be problem? Only after the system rebooted that I uninstall nvidia related stuff. Then After benchmarking that I rebooted again.

Will try driver sweeper but theres no link on guru3d :-ss. Should I uninstall them all nvidia + amd, sweeper clean than proceed with catalyst again?

I hovered over that area and it said: "It is currently running at PCI-Express x16 v2.0."

 

Skipping stuff like Hydravison and a couple of those other things wont have any effect so dont worry about that.

Yeah you need to do some cleaning though of the old drivers and worst case scenario if you are still not happy with the performance back up your current OS with something like Acronis and start a fresh OS install and go from there.

This download should work for Driver Sweeper.

Phyxion.net - Driver Sweeper

When installing it uncheck that crap 3rd party thing it asks you to install its some search bar or something so pay attention when installing it

A fresh install of the OS though will be the easiest way to trouble shoot vs spending hours trying other things but if your PC is loaded down with all kinds of apps that you will have to reinstall and download
then you may not be in the mood to do all that . Thats also why i suggest backing up the current OS so if you do a fresh install and find you have the same exact performance then you can just roll right back to your old OS.

 

hmm I repasted both the CPU and GPU with the AS5 and the CPU temp seems to decrease a bit while the GPU is now 60-61*C. I did line method on the CPU and pea on GPU.

Very tempted to do a fresh windows install since driver sweeper doesnt help and I cannot load catalyst control center up. But have to deal with the temp first. Time to pull it out again.

 

You dont need much paste at all on the GPU and you may want to try the other paste you have as well.

The last time you pasted if it made good contact and spread all the way across the core use less this time. It usually takes me a few times minimum before i am happy so you may have to as well.

As for Catalyst Control Center not opening hmm yeah try the fresh install for sure because i dont know what all you have on your machine.

 

I'm not sure what the temps are typical for that, but aren't the 6990m and 580m supposed to be hotter than the 570m in general? Hmm, not sure why the stock scores are lower, but I'll do some looking around. I think this is the first picture of the 570m. It looks like it has the same setup as the 580m (with a wider surface area on the core), and less 2 VRAM chips. I'm starting to think that the 580m WOULD work as a plug n play! Shouldn't even have to switch drivers too. Great job on the upgrade process so far dragon, and thanks for the pictures!

 

Yep me too but still too expensive

 

They make ES versions of the 6990m? Just wondering because there's one on ebay for $450!

 

every CPU/GPU has an ES version as they send out to OEM/resellers and press for testing purpose and preview. But most of the time they were recalled before announcement or official release but some got out

So I just read around a bit along with Quadzilla's advice as well turn out the idle temp for this card is not the same as nvidia card, and 50-60*C idle is usual.

http://forum.notebookreview.com/ali...-you-getting-6990m-when-idle-stock-paste.html

its like moving to a whole new different country but im getting used to the catalyst control center.
How can I disable or change the AI settings that everyone keep talking about?

Gonna do a fresh windows install soon as im still moving some stuff around back and forth. Will take this chance to see about the SATA transfer rate on my RAID too (did a fresh install few days ago actually).

But seems like AS5 is not good for this as the temp now stay idle at 60-65 and if i try to load even just a youtube video it will full load to 740-900 and temp reached up to 105 and it throttle down, dont even dare try to launch game yet. Gonna put GC Extreme back in and see.

 

This may come in late, but congrats bro!

 

@Dragon, remember that thermal paste needs time to cure before it's really effective. Were you able to get the vantage score increased or are you still working on that? =)

 

I would steer clear of ES Nvidia cards. ATI ones seem to be pretty good still. There are some gtx 460m however that are stuck at 500/1250 and no better than a gtx 280m (cant change voltage or clocks).

dragon, I think what you should do basically is compare the heatsink to the back of the case. It should be perfectly parallel if the heatsink is making good contact.
The simplest way might be to use a copper shim, if you have it.

There was one thing you failed to take a picture of. That is the empty space where the GPU lies in the laptop.
The 570m has 12 vRAM modules total, and the 6990m only 8, correct? I think you have that extra piece for 2 vram modules somewhere.

The grey brown thing is called a VRM, its a voltage regulator. Yes, I think the general consensus is, its impossible for the actual VRM to be damaged from overheating. However these components can get the board VERY hot.

Heres a thought: Nvidia cards don't have thermal sensors on the vRAM. AMD cards do. The vRAM would be the first thing to take damage from the heat. So essentially you could monitor the temperatures pretty easily and see if cooling them was helping you overall or not.

But I guess that is not your problem. The place where the 1 VRM is on the 6990m and the 3rd on the 570m, is blocking contact with the core of the 6990m? Theres only 1 way to really fix that (although a copper shim could work perfectly) ...grab a nail file and start filing down that piece of the heatsink.

 

The 6990 actually idled lower in my M15x then it does in my R3 simply because the M15x idle fan speed is constant but the R3 basically shuts off the fan till it hits a certain temp then it spins back up and drops it down to around 50-53. The R3 will let it hit 60-62 before that fan kicks in.

So every notebook will act abit different but yeah 51-55 average idle temp for this card is very normal.

As for your card spiking up to 105 instantly thats just bad contact and nothing more because all thermal pastes generally perform very much the same +/- a degree or 2.

MX-4 is my current favorite because it spreads so easily and its easy to clean up as well so for someone who likes to repaste alot looking to get the absolute best temps i think its a great choice .

Hopefully you get the performance straightened out soon though so you can start enjoying it ..

 

thanks man!

nah I think I pasted it wrong, time for a repaste! nope, i couldnt even watch a youtube video without fear of overheating please wait some more.

can you explain it a little bit further? Im clueless about that and the copper shim thing though I have read about it quite a few times . Should I just get a roll of copper on ebay then cut it out put on the heatsink? Pardon me if that sounds terribly stupid cause I know nothing (.
If you want I can take pic of the space behind the card but its just empty and facing the mainboard. You’ve seen pretty much all there is of the heatsink, thermal pad, etc,…The GTX 570M doesn’t have anything cover it back, it was just like that in the picture.
Yeah I also see one or two regulators on the mainboard as well just out in the open with nothing cover it up or reduce heat. Pic coming up soon.
Are those sensors GPU temp #1 2 3 as can be seen in the sensor tab in GPU-Z? If so the first one is identical to GPU temperature. 2nd one cooler by 2-3*C. The last one is 1*C less.
Nah at first I feared that 1 VRM may cause problem but now im not sure any more. It doesn’t block any chipset. Its just that the elevated area of the heatsink to face it only fit half of the VRM as the position is not aligned. If that’s the case, do I still need to use the file nail to trim it or anything?

You guys also have optimus to begin with. Can you show me your preferred thermal pasting method? I tried line, cross and dot and maybe its too much or too little that it wasn’t effective (maybe the first time I put in but not really sure). If there’s a video or pic demonstrating that would be a big help too.

I found out this chart of top thermal paste, do you think its legit? According to it the AS5 is still fairly good 
Thermal Compound Roundup - October 2011 | Hardware Secrets

And yeah I just did a fresh install of windows and the GPU seems to run at its correct clock 715. But I couldn’t benchmark yet due to the obvious reason.


----------
Bonus this pic to meaker:



I changed place of the pair to the other's slot and bam! SATA III 6Gbps now

 

Lol those SSD speeds are crazy. I need an SSD badly again. Once i can free up some cash that will be my first purchase but nothing like what you have .

As for the paste method its really nothing more then size of a grain of rice in the middle of the GPU .

Set the heatsink ontop and if possible push down and twist a little to the left and right then follow the screw pattern 1,2,3,4 and tighten each one down fully, not a little bit each till they are all tight.

Its really hard to explain or even watch a video and get the same results as everyone else. Its all trial an error and even though i have done 6,000 paste jobs i still dont get it perfect on the first try
every time.

If you want to hop on a plane and come to Michigan ill be happy to do it for you .

 

I'd recommend you to pick a Crucial M4 any capacity, you wont regret it.

Last update for the night:

So I pulled the laptop out one last time and repaste, this time I used Gelid GC Extreme, put paste down the size of a grain of size then use the spoon/applicator to spread it around evenly on the surface. Then i put the heatsink back in and put as much force as possible into it while securing the 4 screws and well this time it seems to work.

But another problem arise!

This was while I'm watching youtube, before last repaste, as you can see the gpu run at 300/900 with crazy temp. Notice temp 1 and 2 are 8-10 degree higher than temp 1. ----> after the repaste the system idle at 54*C. Raise to 64-67*C while watching youtube so Im supposed thats normal.


Now after the repaste, all seems to be well and good and then I tried to launch the witcher 2.


As you can see the GPU temp #1 is way lower than the other 2 but just get dragged along and cimb up as high as in pic. It stayed there but I just quit after 3-5" in fear of further damage to the card. With this I assume the thermal pad on the VRAM is not effective, but not sure where exactly is the 2 thermal sensors listed there. I have ordered the thermal pad quadzilla recommended, lets hope putting them in will remedy the situation.

--------------
This is the space behind the card, as you can see, theres nothing there, should I put anything in there?


And then about those out in the open VRM on the mainboard.


I may attempt to launch 3dmark vantage, hopefully it wont blow up.

 

Behind our cards on the R3 there is a HUGE thick green thermal pad its probably 2mm to 3mm thick and this thing is like a small blanket .. If you have any kind of pads laying around it cant hurt to put something behind the card .. The reason i believe it is so thick on the R3 is to keep heat away from the motherboard.

Also when doing the paste method all you should have done is drop that grain size in the middle and the heatsink will do the rest. You dont need to spread it with a spoon or anything else for that matter.

I wish i could see exactly what is going on but sadly i cant so your just going to have to keep trying. Just remember dont spread the paste and let the heatsink spread it . Also this will give you a good idea how well the heatsink is actually laying down the core of the GPU by how well the paste spreads just by putting that drop in the middle ..

I have been looking at the Crucial M4 and eventually i hope to get one . First time in a long line of laptops i do not have an SSD but priorities come first- bills .

 

now the GPU is idle at 54-54*C with all 3 sensors stay the same at 53-52.5-54. But as soon as I load up a game the same situation with witcher 2 happen again.

I think I need to replace the thermal pad on the 2 columns of VRAM and maybe put a 2mm thermal pad underneath the card similar to M17x R3 (watched the youtube clip and boy is it thick and large).

Also I think this time the paste is applied correctly, havent bothered to do again but also found no reason to. Its just that the GPU sensor temp #2 and 3 raised horribly high and dragged the whole card along.

I ordered the Arctic Silver Ceramique as well to see if a different type of compound would help at all.

 

I think im quite satisfied with the idle temp now, it is currently 52-50.5-52.5 +/- 0.5*C. If only those thermal pads are on sale locally.

Also do you think I should undervolt the card to 1.0V to make it run cooler?

 

Undervolting will help in the long run. I ran one of my cards at 0.975 stable at stock clocks(temps lowered around 7 degrees IRC) but for now that is not going to help with the crazy temps you have . So yeah eventually you can undervolt but for the moment dont waste the time on that.

The way your temps are spiking it sounds and probably is the core not making proper contact at all. Something on that heatsink i guess is lifting the card up and stopping it from sitting flush.


Was it possible to use the original backplate i sent you or are the screw sizes different ?.


Looks like the GX660R with the 6970m had to do some grinding to get the heatsink to lay flush but his heatsink also looks a bit different. Those 4 blocks along the back end of the card though might be propping up the the card so it cannot lay flat. Pads and paste are only going to get you so far and they usually will get the cards temps to drop a few degrees but what your experiencing is totally different from what i am seeing and your card is not even getting close to the heatsink from the temps in Witcher2.

http://forum.notebookreview.com/msi/568347-ati-6970m-gx660r.html

 

wow i was going to post the same thing and then see you edited your post .
Im gonna pull up the card and heatsink completely and compare it out.

What can I use to trim it away if needed?

 

That guy used a Dremel (google it if you dont know what that is and you can buy them locally) .

But yeah it seems like you might have to do some grinding in the end to get it working correctly .

Also see if you can secure a spare heatsink in case you end up having to go back to the 570 as well . RKtech deals with alot of MSI parts so they might be able to get you one and having a spare cant hurt in case things go south with the modding.

Maybe message Meaker because that guy knows his stuff on the MSI and has tried so many mobile GPUs he might have a better understanding of what needs to be done to get the heatsink to lay flush.

I dont know what your dealing with without being there in person so i am running out of ideas/suggestions.

 

Oh wow so i pulled the thing out and yeah theres some trimming need to be done, pics coming soon.

 

A note about the raid.

All I did was set the bios to raid and create my array. It should just be default SATA III, strange one of your drives is not (in raid0 both are limited effectively by the lowest).

Both of mine read III.

 

You can get thermal pads at radio shack....but I don't know how good they are compared to Quad's suggestion. Probably not as good, but they can be found locally.

 

AMD Radeon HD 6970M video card benchmark result - Intel(R) Core(TM) i7-2860QM CPU @ 2.50GHz,Micro-Star International Co., Ltd. MS-16F2 score: P14804 3DMarks

 

wowww you must be happy eventually ! How are you getting on with the temps ?

Note: the Vantage CPU score is a bit low; isn't it ?

 

I guess you got those temps under control .

Congrats on the score .

 

Another one:
AMD Radeon HD 6970M video card benchmark result - Intel(R) Core(TM) i7-2860QM CPU @ 2.50GHz,Micro-Star International Co., Ltd. MS-16F2 score: P14835 3DMarks

yeah I got those under control..sort of. Please wait while I post a full detail report.

 

What are your CPU temps while your at it ?.

Your CPU score is a couple thousand off so i wonder if its throttling.

 

okay so follow Quadzilla's advice I pulled the system apart again, this time to align the GPU with its heat sink as you see below:

The right part faced the heatsink well while left seems to be blocked by something.


The other side:


Turn out this part (highlighted in pink) of the heatsink blocked a black transistor (or something). I need to trim it away somehow. The yellow highlight dont really block anything, its just half-touch with the regulator.


So I went to Target and bought a rotary tool.
Walmart.com: Black & Decker 3 Speed RTX Rotary Tool (Model# RTX-B): Tools

Using its sanding point I just whet away the highlighted part (carefully of course). After that I just reuse the stock thermal pads though they are already kind of tattered but its not like I have anything else at the moment. I kept the paste as it still looks good then boot the system up. The temp at idle is still the same. I set to high performance mode, turbo fan for it to run at full speed while benchmarking.

This is after I run 3DMark Vantage with all the temps and WEI score.


So yeah when the new thermal pads arrived maybe I will try to trim it down a bit further, replace the new pad, put the 2mm one underneath and see if it improve.

The CPU seems to be overheat or something that I'm still finding solution, consulted with Xonar a bit:

------------------


Just better images of the space behind the card and out in the open regulators on the mainboard, wonder what does that white port do.


@Meaker: I deleted the old array and create a new one, install windows and Intel RST say that both disks are at SATA III 6Gbps now.

 

That 96 degree temp in there that shows up in 2 spots looks like your chipset getting overheated.

I am glad you made some progress though.

Also on your CPU if it is making good contact with the heatsink try to use as little paste as possible on it.

On my R3 i only use such a tiny amount and in this instance i spread it across the core with a tiny piece of old credit card.

Regularly i dont ever do the spread method but on my CPU i have found i got the best results with just a tiny amount of paste and spreading.

MX-4 spreads like butter so its great to work with but AS-5 also spreads very easy and you can use just enough to cover the core.

Keep trying man your getting there .

 

yeah I figured since im still recycling the old stock thermal pads on those and the surface may not touched fully yet. But until my new thermal pads came in, theres not much I can do can I?

 

You can try to figure out where the CPU chipset is located on that machine and see if there is something you can do to bring the temps down.

Meaker might know where it is. On the Sandy bridge laptops i have notice the Chipsets get passively cooled mostly so they dont have a fan on them generally.

In case you dont know what i am talking about the Chipset is not the CPU but a separate small chip somewhere on the motherboard and it probably has a small aluminum heatsink over it .

I think that one temp in your reading that says Package should be the chipset the one that is 96 degrees. I assume that is where the throttling is coming from but i am not sure what your
motherboards CPU temps are for throttling but either you gotta get those CPU temps down .

 

You mean the HM67 chipset right? I will try look around.

 

Yep exactly that.

 

I do not know what the Package temp represents, but mine is only 2-3*C higher from my core temps under load. At idle its ~10*C higher. I figure it's just the whole CPU'd (including the iGPU) temp, but our's is disabled so I don't know if it really is what I think it is. I have yet to get my throttling under control (I am able to keep it at x29.75 max out of x31 which is better than the original x28 I was getting).

Because THRM and Package appear so similar to each other, I think they might be a predicted/estimated value than an actual temperature reading from a sensor. My 7811FX was like that and it would trigger throttling because the calculation was off by a good 15*C and there wasn't a simple way to fix it.

As for the multiplier being stuck, I've never seen such a thing happen unless EIST is turned off, which it isn't.

 

Yeah mine is generally 2-3 degrees higher as well but his is way off the charts and like you i dont know exactly what it represents either but it should be similar to the CPU temp.

I rememeber the Gateway FX issues all to well.

 

maybe those chips/blocks right next to the CPU.

There's something weird I just noticed. Dont know what triggered it but now when I load up the bios, in the front page it said in the marketing name line: instead of GT683DXR like usual it is now GT680R???

 

It's probably an editable string in the BIOS and it's the same notebook in the end.

Those chips (gray cubes) are the VRMs. I'm pretty sure it's impossible for them to overheat, but they do get extremely hot. There should be thermal pads attached to the heatsink where it presses down on them.

 

yeah maybe they got defective from quite a few times I repaste the CPU, again not much I can do until my new thermal pads arrive (

 

Multiplier is stuck at x21 on an i7 920 - Other-Components - CPU-Components

i7-950 multiplier stuck at 24x - Overclock.net - Overclocking.net

they seems to be similar to my problem...

 

What does core GT mean? - AIDA64 Discussion Forum

more info about the CPU package thing.

This evening I did one last repaste for the GPU and this time I take out all the thermal pads and put in paste instead and it works wonder. Tried Witcher 2 and temp never reached over 80*c, all 3 sensors are similar +/-1*C. The reason I tried paste in place of thermal pads is because of the review of IC diamond 24 on tech inferno. They have no thermal pads on their 5870M, instead only paste and it seems to work well for them.

But the CPU package temp is still crazily high and I have no idea where the chip could possibly be.

 

The CPU package is usually considered the little PCB the chip is attached to.

As a QS maybe it's not quite calibrated right?

 

If you do this (thanks MXM-upgrade):





Then the 5870 heatsink covers the vrms as well.

 

about the CPU, if thats the case is there any solution I can do by myself? I have contacted the seller and if its indeed defective, I will get an exchange. I will try put in another CPU tomorrow

About the GPU, you mean the top left chip/VRM with the plus sign?

 

the Package thing doesnt seems to be caused by CPU, or else all 3 of my CPU are defective.