Notebook goes black screen when gaming

I recently re-pasted my notebook with AS5 just to practice, while I'm looking for Gelid GC Extreme to buy as the paste I had in place was a mess, but I've got a BIG issue. I used 99.8% isopropyl alcohol to clean everything and then went with the apply and spread with plastic card method using a thin paste layer.

After everything was done I booted the notebook and tried to run FurMark, 3D Mark and then a consuming 3D game like GTA V, all of them go black screen within 2-3 secs of actual 3d action on the screen and then the computer enter a restart cycle and I have to turn it off forcefully. I then tried to run what I believe to be a 2D game called Inside and it worked 60fps like usual, I can browse the Internet, use Office, etc but I can't do anything 3D related.

Tried to reset the driver already. Used DDU during Windows safe-mode boot and cleaned everything then installed the Nvidia driver I had before which was 100% working. I'm kinda desperate, did I break something?

GPU temps on idle (Chrome + some background softwares) are AVG 38c/42c. Photos of the current paste job attached!

Edit 1: Maybe I used too little paste? It was indeed a small quantity and I spread it with the card to cover all the die. Maybe the distance between the die and the heatsink is bigger than a desktop so I need to use more paste?

Edit 2: I'll put some more paste above the one I recently put, I really feel like it's a paste issue, the notebook feels too damn cold on the higher area, just below the screen and this area generally feels a little hotter even on idle, maybe this is indeed cold because the heatsink isn't in contact with the CPU/GPU die, hence it is not transfering any heat, so the notebook is turning it off to avoid issues when I start anything demanding.

 

Looks fine, but if this it the first time you've done a repaste; after you've applied the paste and screwed down the heatsink make sure to remove it once. You should see all little 'peaks' of paste all over the die surface, indicating the paste made full contact with the heatsink. If one corner doesn't show these peaks then the gap was indeed too wide for the amount of paste.

Spoiler: the left heatsink is what you want to see

Also make sure that the vrms and memory chips are properly covered by the pads and push in the mxm card with a bit of force into the connector. Some cards are rather thin and don't make a nice 'squeezy' contact on all pins. When the card is activated/stressed it'll require more power from the board and could result in this sudden shutdown if the mb can't deliver that due to a semi-complete connection.

 

Ok, but if I remove the heatsink to check this I'd have to clean everything and re-paste again or could I just put some more above the one I already have? The paste job in question was done yesterday, so it's pretty recent.

 

No need to clean; just remove the bit that spilled over the edge (that was too much) and apply a little extra where there wasn't enough.

 

Can you please clarify? Which image are you talking about, perhaps img_0622.JPG? I saw that little bit out of place when I posted the image here, rsrs.

 

No, that was a general comment. Only applies to after screwing down and removing the heatsink. Your images show only the pre-heatsink situation, right? At least, they look like a clean, credit-card applied paste job.

 

Oh ok. Yes, I took all the three before putting the heatsink in place. When I get home I'll remove them, take some photos and apply a bit more of paste if I see that the heatsink is too "clean", in other words, didn't really touch the die...

Could this be related to the isopropyl alcohol 99.8%? I mean, maybe some of the liquid went to the wrong place? I'm using the same power outlet I've been using for yers and the PSU is the original one, this couldn't be a power issue, right?

 

Update: It seems I did a terrible job again, this time with too little paste (my first attempt I put too much). Two heatsinks (Master GPU + CPU) were below average, but the Slave GPU was really in a bad situation, maybe this was the cause of the black screen + restart?

OBS: I forgot to take a photo of the CPU die before the new application, so that's why you're seeing paste in there. This paste is from the new paste job. This only applies to the CPU, the GPUs were taken before this new application.

Photos: http://imgur.com/a/QePOu

IT'S WORKING. I ran FurMark v1.17.0.0 and it worked.



I believe the temps are a bit high but people say AS5 takes some days to really work, but the good thing is that it's working. Applying paste is a really difficult job for me, I hope I do it right the first time with Gelid GC because this thing isn't cheap where I live :|

 

Just an update on the temps:




I played GTA V on high/very high for some time and got what I think to be good temps. This same game with the same settings was putting CPU + GPU1/2 on 90-94c all the time and I was getting GPU limit 1 for both within minutes of gameplay, now it's all gone.

The SAGER-NP9370 have 1 heatsink for GPU MASTER and other that's shared for CPU + GPU SLAVE, is that right? That's why my GPU2 always have higher temps than GPU1, right? I just checked Nvidia CP and saw that Physx settings is set to "Auto-select" and it chose GPU2, should I enforce it to GPU1 instead?

Right now I'm using AS5, but I'm looking forward to buy the Gelid GC Extreme, which temps should I expect when using this one?