Are my GPU memory thermal pads contacting the block?

Prozn

Prozn

Prozn

Associate
Joined
16 Apr 2015
Posts
274
While my first time loop was leak testing I decided to look down the side of my strix 2080 Ti / EK block to check the thermal pads. I don't know why I did, I checked the instructions about 300 times while installing the block and all was correct.

Anyway, It looks like there may be a gap between the pads and the block, which is obviously quite concerning, but I cannot tell for sure. Do these look normal? Am I stressing over nothing?

If I just go ahead with booting. What is the best way to check that my GPU memory isn't overheating/running hot? I wan't to avoid any unwelcome games of space invaders if at all possible :D

KkHfuYg.jpg

Ps. Sorry for bombarding the watercooling forum, but it felt like this warranted a new topic!
 
You can see the bottom pad looks like it's just drooping over the edge of the RAM IC, but seems flat going further into the PCB, so I'd say assuming your pads are larger than the IC's and just dropping over the edges, they look OK to me.

If I was worried, I'd overclock the GPU MEM, if I could get anywhere near a normal overclock I'd call that a success since if there was just an air gap it certainly wouldn't go as well as the stock cooler in terms of memory.
 
Demon said:
You can see the bottom pad looks like it's just drooping over the edge of the RAM IC, but seems flat going further into the PCB, so I'd say assuming your pads are larger than the IC's and just dropping over the edges, they look OK to me.

If I was worried, I'd overclock the GPU MEM, if I could get anywhere near a normal overclock I'd call that a success since if there was just an air gap it certainly wouldn't go as well as the stock cooler in terms of memory.
Click to expand...

Thanks. I managed a 15500 on the memory on the stock cooler, so will give that a go tonight! Starting at 15000....!
 
Prozn said:
I actually came across this completely by chance when searching for the installation guide on the EK website, so the 4 closest standoffs to the die have been removed from my block :cool:

Although I'm not particularly happy with it, you cant blame EK for ASUS making a change to their process I guess.
Click to expand...

asus does this quite a bit !
Gigabyte does this half way through but they are normally big changes - going from reference design to their own
 
orbitalwalsh said:
asus does this quite a bit !
Gigabyte does this half way through but they are normally big changes - going from reference design to their own
Click to expand...

Just another one of the perils of watercooling!

I had another problem with my block, the standoff on one corner of the block does not line up perfectly with the screw hole on the PCB. I'm not sure if it is a manufacturing fault or user error, but it didn't seem to affect the clamping force around the components that matter so I just went with the screw missing.
 
Prozn said:
Just another on of the perils of overclocking!

I had another problem with my block, the standoff on one corner of the block does not line up perfectly with the screw hole on the PCB. I'm not sure if it is a manufacturing fault or user error, but it didn't seem to affect the clamping force around the components that matter so I just went with the screw missing.
Click to expand...

worth dropping ekwb a message and snapshots to confirm etc .
could be just an off batch maybe
 
I think it looks pretty good, it’s tough to tell as the edges look like they hang loose but the pressure on the chips is fine.

If in doubt take it off and you will visually see the depression from the VRAM in the pads, reapply paste and reattach.
 
Smffy said:
I think it looks pretty good, it’s tough to tell as the edges look like they hang loose but the pressure on the chips is fine.

If in doubt take it off and you will visually see the depression from the VRAM in the pads, reapply paste and reattach.
Click to expand...

Thanks! I booted the system up last night and everything looks fine. GPU sits +1C over water temp at idle, and +10 to 15C at load, with core clock pegged at 2,100mhz. Cannot check memory temps, but I am running the memory at 15,400mhz (7,700/1,925 (why are there 3 ways of quoting memory speed?)) and all seems stable, so I guess that is a good sign the memory is being cooled.
 
Prozn said:
Thanks! I booted the system up last night and everything looks fine. GPU sits +1C over water temp at idle, and +10 to 15C at load, with core clock pegged at 2,100mhz. Cannot check memory temps, but I am running the memory at 15,400mhz (7,700/1,925 (why are there 3 ways of quoting memory speed?)) and all seems stable, so I guess that is a good sign the memory is being cooled.
Click to expand...

Its looking good then speed is because 15,400mhz is derived from 7,700 x 2 (Double Data Rate, aka GDDR6), divide it by 8 for conversion to bytes. Its the same reason your broadband is sold to you as 50mb but you 'only' get a download speed of 6mbps, bits vs bytes :)
 
Prozn said:
Thanks! I booted the system up last night and everything looks fine. GPU sits +1C over water temp at idle, and +10 to 15C at load, with core clock pegged at 2,100mhz. Cannot check memory temps, but I am running the memory at 15,400mhz (7,700/1,925 (why are there 3 ways of quoting memory speed?)) and all seems stable, so I guess that is a good sign the memory is being cooled.
Click to expand...

GDDR6 is QDR or quad data rate.

Most memory is DDR hense the 7700.

GDDR6 is 1925x4 (QDR) then x2 essentially due to how qdr operates. So in your case it will then have an effective speed of 15400mhz.

With regards to contact, it looks fine to me. Looks the same as every Ekwb block I’ve had when checking memory contact.
 
Necroing this thread for an update: turns out the pads were not fully contacting the memory/vrm! I was getting random black screens under high load situations. I took the block apart, lo and behold - no depressions in the pads!

Having reseated the block and being more careful with the installation everything is working perfectly.

I'm an idiot :rolleyes:
 
Prozn said:
Necroing this thread for an update: turns out the pads were not fully contacting the memory/vrm! I was getting random black screens under high load situations. I took the block apart, lo and behold - no depressions in the pads!

Having reseated the block and being more careful with the installation everything is working perfectly.

I'm an idiot :rolleyes:
Click to expand...

I took one of my old ones apart about 2 years on and noticed I hadn’t removed the plastic on one side of the memory pads :).

They worked perfectly regardless ha!
 
You must log in or register to reply here.
Back
Top Bottom