Upgrade a G73-jh to 6970?

Did you try to reseat the 6970M? Maybe it just got loose and the laptop couldn't recognize it...
I remember when I had my old Clevo m570TU it was Christmas night, I was waiting for the next batch of games on Steam game sale when BAMM, everything went black and laptop didnt boot again...the VGA got loose (I didn't move the laptop) and I had to reseat the GTX 280M.

 

When i had a first go at swapping to the 5870 into the D900F i also had some trouble. It booted to windows, but shortly after the screen went black (on 2nd try white before i shut it down) as it overheated - even though i had a missleading good paste print on the heatsink the core actually wasnt properly cooled (which i noticed when i took a closer look trying to fit the heatsink on the gpu outside the laptop), so i needed to cut up the heatsink in order to get it working properly. The difference though was that i could boot again with the same thing happening unlike in your case where it wont even boot again. Its strange, overheat protection just shuts off the gpu, it shouldnt prevent booting.
Id still recheck the heatsink to see if you have to mod it, i find it hard to believe that it would just fit as some of the coils are usually differently positioned and different heights.
The second cause i could think of and would explain it not booting anymore would be that some metal filing from your drilling shorted something on the gpu...

 

Was just moding my heatsink and was mesurig the GPU and Vram hight. Look at the 6970 and use a ruler of some sort and balance it on the side rams and GPU if its flat (ram with GPU) then there is your problem. With the 5870 the GPU is higher about 0,5-0,7mm. + found a nother flaw with the 5870 heatsink. The part that is near that HDD that cover somesort of a voltage regulator and some solderd parts, it sticks out more then it should.

I removed all the thermal pads and just put it on the 5870 and the side of the heatsink sick out so much that form the side you can see that the gap is at an angle.


If the 6970 is flat then with the 5870 heatsink the GPU should have a gap of about 0.5-0.7mm and if it is at and angle then no wonder it overheated.

 

I remember one user had some problems with his keyboard after repasting and had similar shutdowns becouse keyboard underside shorted with something underneath. You posted that your keyboard didnt worked so maybe it is something you should investigate. Tried to find thread here but its like looking for needle in haystack

And rep for you man you deared to do something I just have dreamed for last couple months...

 

At the end of the day we got what we wanted it booted without any need to flash or update so it shows the chipset is compatible.

It is all trial and error from here now to correct any issues that may occur just like all other modded cards.

A worthy +1 from me.

@Tudordeworf: Do post an entire thread of your results though and as many photo's as you can to help others who try or if you should attempt again.

 

A thermal shutdown doesn't happen that fast as long as the heatsink sits properly. And the system should boot after such an emergency shutdown, more or less normally (it will give you the "Windows did not shutdown properly" error message, but that's it).

If it really was because of heat, then probably because there was a gap between the die and the heatsink.
I really can't believe that the heatsink fitted well enough without modding. A 1-2mm gap between die and heatsink is much too big.
The card is hardware-wise identical to mine and I used a JH heatsink...

But like I said, thermal shutdown doesn't explain that the system doesn't boot anymore, a card won't die because of a thermal shutdown, there must be something else wrong.

 

out of power cos g73jh take a lot of watt compare to jw/sw according to notebookcheck

 

Did he even try to take out and put the 6970M back in?

 

150w is more then enough for the JH and 6970. With a 920xm....now thats a bit to much to handle.

 

I know this may sound weird, but check that the thermal pads in for the GPU VRAM isn't too thick. I tried stacking thermal pads on the memory banks on my 6970M and the cards went crazy; it would shoot up to 70 degrees C in idle and would skyrocket to 100 degrees C in like 2 seconds and then cause an abrupt system shut down. I think stacking thermal pads (or making them too thick) messes with the GPU sensors or something. So make sure that the thermal pads are properly leveled with the die contact site on the heatsink.

 

Yeah, having to thick pads will result in a bigger gap between the die and the heatsink, the card will run very hot. Thermal isn't supposed to fill a gap more than about 0.5mm, and even this is actually too much.

Take a closer look at the heatsink next time tudordewolf, I really think you have to mod it.
Still, this doesn't really explain why it didn't boot anymore afterwards.

 

After reading the thread responses to so far,
I'm inclined to believe its the heatsink. What kind of mods would I have to do? There are no chips pushing it to an angle; to me it appears the thermal pads are securely pressed against the VRAM; perhaps they are too think.
I'll try using thinner VRAM pads as well as helping the chips on the back of the card. As far as I was able to tell, the only parts of the heatsink making contact were over the die and VRAM.

I thoroughly dusted, checked my card after the dremel op, I doubt any flakes or shavings made it.

 

Wow this thread is moving on incredibly rapidly.

From what I have read I also think your heatsink just isn't sitting flush with the die of your card. The hard shutdowns are typical side-effects! If it crashes once it won't let you start for a good while as the card has to cool down first.

Other than a mod to make sure the heatsink sits absolutely flush on the die there are other things that you can do to help it boot successfully.

1.) Try a larger PSU such as the 180W targus although unlikely you will need the extra power (I use one but have a power hungry desktop quad)

2.) Lock your CPU to the lowest possible speed and voltage possible. Preferably in your bios if you can if not then with throttlestop.

3.) Completely and utterly remove your GPU drivers off your machine. After unistalling them re-boot and use a free driver wiping program to remove any remaining traces then run CCleaner to clean up your registry. Finally re-boot again with your 5870 after all this and then switch off and change the card.

4.) Run without battery on first boot until new drivers are installed.

5.) Keep laptop casing off and run a desk fan to cool the card actively from underneath until you have the drivers installed so can monitor temps (this might not be necessary but if the heatsink is badly seated this could prevent a hard shutdown allowing you to install drivers and then at least find out your temps. If very high then you know the heatsink isn't fitting rather than being kept in the dark)

6.) Another simple thing to do is to download the latest drivers before you initiate the swap so that immediately after boot up with the new card you can start the driver install.

Then cross your fingers and hope it boots/works! Hope this helps

PS; Dallers I would love to take your Porshe out for a spin!

 

I have great hope for this, it definitely sounds like a overheating issue due to the heatsink. I don't know enough about this kind of stuff to be any help though.

I'm very excited that it booted up properly with it in there, that means it is compatible, we just have to work out the kinks!!!

 

Here's a pic where you can see what I had to do to make the heatsink fit.



On the left you see the original JH heatsink, I marked all the spots where I had to do some cutting/grinding to make it work. All of these small modifications were necessary to make proper contact with the die. I really took a close look at it.

Compare it to the picture of the modded heatsink to see all the small differences.

Here's another pic which shows the installed 6970m, you can see the grey coil (R56) next to the heatsink, I had to do some grinding there. Easy to overlook.



Uninstalling the drivers in advance is a good idea, but not crucial. If you still have the old driver installed, Windows will simply give you an error and just run without loading the driver.

 

Those pics should be really useful to the OP svl7 great work!

I know windows will give an error but it is always best to give both hardware and software a kind of clean slate when doing something new to give the best chance of success.

Every little helps!

 

Looking at those, I can see how I missed it. The only cut I don't understand is the bottom-right - there's nothing there on the card, and nothing in the case that gets in its way. The top-left ones, however, I could see getting in the way just enough to give me trouble. The top-right coil, interestingly, fit, it was close, but it wasn't a problem.

So here's my next plan of assault: Get the heatpads (ordered), grind / check / recheck the heatsink on the 6970, and boot (without a password in place) with the 6970 drivers sitting on my desktop.

 

tudordewolf when can we expect attempt number two? I guess its safe to assume when the thermal pads arrive but when are you expecting them?

 

according to this clevo reseller the 6970 is being phased out as told to him by clevo, apparently some supply issues..

http://forum.notebookreview.com/sag...d-radeon-6970m-goodness-d-11.html#post7610331
read the rest of that thread for more info..
If that's true i hope we this works so we can order em before they're gone.. (again assuming his info is accurate..)

 

D=

Its a move to stop us from upgrading our JH's!

Second attempt: The shipping estimates range from July 26-August 8th. That looks like a painfully long time, maybe we can hope for sooner, OR I can take a big leap of faith, put AS5 on my VRAM (since it has to be so thin anyways) and use my RAM therm pads to cool the trouble spots on the card =/

 

For delivery of thermal pads? Where are you buying them from the Antarctic?

Just pick some next day delivery ones from Ebay for a fiver. If there is a gap then thermal compound will not do the job because it needs to be compressed as mentioned earlier a 0.5mm gap is large enough for thermal pads let alone thermal compound.

 

I just picked the cheapest XD (Taiwan)

I mean, yeah, but, the GPU is our main concern - and the heatsink springs / backplate do apply a lot of compressive force - using Arctic silv. on the VRAM may be the best way to ensure that as much compression as possible is going onto the GPU. - no backpressure from any therm. pads.

What does NBR think of this:
Pull the VRAM pads from the heatsink, slap 1 on the back accross the chips MXM-upgrade mentions, put the other across the top DC-DC converters just in case, and use arctic silv. on the VRAM - the heatsink sits so close already, and since the 6970 GPU is slightly lower than the 5870 GPU, we might get better cooling on the VRAM than the pads would have done anyways...
I'm out for a bit tonight (saturday ) starting 4PM, might make an attempt when i get back, or tmo.

EDIT: OR OR OR I could use a copper shim 1mm thick on top of the GPU, to bring it closer to the HSF assembly.

 

Im not gona lie, im all about taking precautions, but my heart just ached a bit when you said up to august 8th.

It would be great if you can pull off a second attempt within the next few days as opposed to (potentially) over a month from now...who knows second time around could be a success......

 

Surely paying an extra $1 for it to get to you a month+ sooner is worth it??

 

None of the ones on ebay I see even estimate shipping in June - I tried US only, but that cut out all of the .5mm matches. I think I'll try my VRAM-paste / repurposed pad idea.

 

At this point, any reasonable attempt is a great one.

 

try frozencpu for the thermal pads they have wide selection and will ship quickly..
link

 

Oh yeah, the metal thing I cut off on the right side was because the heatsink didn't fit otherwise if I remember correctly. I tried this with a JW, probably it's not necessary to cut it for the JH mobo.

Don't you have thermal pads on your heatsink? They'll do their job just fine as long as the heatsink fits.
I'm still using the stock pads in my system as well.

Otherwise here's a reseller for thermal pads, but like I said, the stock one should work fine.


Edit: Lol, kurtcocaine just mentioned the same reseller.

 

You can use thick thermal paste on vram without any problem. I have used it on my 5870 and it did bring my vrams temps one degre down. Only problem with thermal paste is that you cant have it for long time becouse of high temperatures and bigger gap mading it to dry quickly and lose thermal conductivity.

 

@tudordewolf
Were the fans running fullspeed? I don't saw tha answer.

@jmhdj
The VRAM has no temp sensor. All are inside the die!
MemIO is the memory controller.

 

@tudordewolf
Maybe you already answered this but did you use AS5? Looks like way too much thermal paste to me and maybe shorted the board. You might try cleaning up the GPU and retry. Good Luck I hope this works.

postimage.org/image/tbkrmrt0/

ASUS G73JH
i7 720QM @1.9GHz (3.2 GHz Turbo Boost) 5870M 825/1100 8GB DDR3 1333 @ 1600MHz Western Digital Black 500GB Seagate 500GB

 

Back from a rather extensive retool. I didn't use AS5, I used ICD
slv7: you were right about the heatsink. I modded it, got a MUCH better fit, was able to locally acquire some thermal pads, and used the original ones on the VRAM. Seated the card, put it all back together. Few seconds of blue light, then nothing. I feel like at this point it is likely something with the card =/

If only i'd first modified the heatsink, I may have had full success from square 1.


I am willing to give it one more shot: clean off the card, update to 213 BIOS, (i don't know, just maybe it'll help) reseat it all, and see if it boots.

 

Does laptop switch off on its own after this few seconds are past or it just stays like this?

 

Lets hope not because if you have drilled it you have no chance of returning it becoming an expensive test, unless you are rich then its ok.

 

It switches off on its own, before anything else happens. No kb lights, screen, fans, just the case lights for a few seconds.

upgradeyourlaptop advertises a VERY clear no-returns policy on their ebay page.
I'm not rich, but I'm also not one to get TOO attached to physical things. I will consider the money on it lost, but not wasted. My G73 still works, and the 5870 is still a kickass card.

I got to see the 6970m boot once, which makes this at least a proof-of-concept capable experiment, though we now know that was with a badly seated heatsink (which may have destroyed it) (which means if someone is willing to try AGAIN, with a proper heat sink this time, they might succeed)

If anyone has any ideas for resetting a card, somehow, something I could do, that would be awesome.

 

Are you going to try installing it again with a proper heatsink? I hope so

 

Sounds like a dead card, very sorry to hear this! If you have the chance to put this card in an Alienware or Clevo system you could see whether it's working or not.

He just did.

 

It looks like your gpu card got damaged somehow, sorry. I recomend you to find some other laptop that has support for 6970m and check it that way first.

Looks like we had same thoughts here slv7 at same time

 

If you can somehow confirm a dead card, i already have a spare JH heatsink and 1mm thermal pads from frozencpu waiting for a 6970m 2nd attempt .

 

For now, that's a plan-on-hold. I'll take the card with me, maybe I'll find a kindred soul to let me see if it deserved a proper burial or another trial by fire XD

 

I think there's one last chance. 1341 mentioned here that when the w860cu has more than one RAM module than notebook does not start with 100w GPU. I know it's longshot but maybe it can help. If it was me I would have tried it again with only one RAM without optical drive, second (or both) HDD(s) and battery.

 

Aww this is a really sad outcome! Really hope the card isn't dead!

 

If this was the issue, it shouldn't have started the first time. Something went wrong after that.

 

No dice on the follow-up re-try.

Oh well. Guess my next GPU is going to be an ATI 7k or 8k series in a few years XD

I just feel that if someone else has as much spare hope lying around, modify heatsink, and try ANOTHER 6970, we'll have our successful completion.

 

What about your card? Are you selling it? I could easily test it, but I don't think the shipping costs are worth it.

 

Its sad to hear the card sounds like its dead. I'd offer to test since im minutes from the DC area but a g73 is a uncertainy for the card.

Your first boot having lower resolution is indeed very promising. I'd go for an attempt if I had the money to use but I don't. And I just bought a 920xm that will be here soon

So if you want a 2nd g73 tester not far from you, I'm game. Although best to confirm in a machine that we all know the card works in 100%

 

Here is an idea. Remove the card and try to boot it. If it does the same thing as with the card then you my friend have got a dead card unfortunatly


Lucky me i have not orderd my 920xm from the uk. Now have my eye on the 6970 from Laprop monkey for about 378£ and free shipping. But not so keen on the fact that the G73 can kill it....if it was not heat related fault that killed tudordewolf´s card( if its dead).



Can not get the clevo 6970 upgrade out of my mind as they added thermal pads to the bottom of the card. The problem might be the lack of cooling on the bottom side of the card. Can not find a good picture AW and Clevo to see what is under the card.

We can not do this on out G73 as it is on the motherboard. :/

 

It's very unlikely that the card is dead because of heat... I already had thermal shutdowns with my card and it's still alive and performing greatly.


Anyway, I have nothing at the back of the card, there's the mobo. The M17x 6970m comes with a big thermal pad on the back... no idea why, maybe it really helps. However, my card works fine without.

 

Well the clevo and MSI-s 660 seems to have thermal pads on the back + the clevo upgrade needed the pads for it to run.

 

I have one question here guys. How would windows know that his card overheat if right video drivers were not present/installed and then temperatures couldnt be read.? And if I remember fans are controled by EC on our g73`s.
I am bit confused here and if someone with more insight on this can clarify it would help to future upgraders .
Thanks