lebel

NucleusX
Can anyone else out there comment on whether to use long strips or singular squares of thermal pads on VRAM ?
The strip can cover multiple VRAM modules at the same time and will have more surface area to contact to, but
I'm not entirely sure of the downsides of this, if there is are any. Your input is appreciated !

The higher the thermal conductivity (W/m-k) the better and If possible use the lowest thickness whilst still making full contact i.e 0.5mm.
Another positive effect (see link) in lowering the temperature can be achieved through the use of thermal pads together with (non conductive) liquid thermal pastes, i.e; MX-4. This looks interesting...
 
Interesting link on the subject
https://hw-lab.com/therma...lity.html#introduction
 
 

Great read in that link, thanks for that. Although, i don't think he really said much about strips vs individual squares
on the VRAM specifically unless i missed it. Now i'm contemplating the use of compound with the pads, but tbh, i'd 
prefer to do this by over-compensating with high quality thermal pads alone. All that extra compound would be messy. 

NucleusX

lebel

NucleusX
Can anyone else out there comment on whether to use long strips or singular squares of thermal pads on VRAM ?
The strip can cover multiple VRAM modules at the same time and will have more surface area to contact to, but
I'm not entirely sure of the downsides of this, if there is are any. Your input is appreciated !

The higher the thermal conductivity (W/m-k) the better and If possible use the lowest thickness whilst still making full contact i.e 0.5mm.
Another positive effect (see link) in lowering the temperature can be achieved through the use of thermal pads together with (non conductive) liquid thermal pastes, i.e; MX-4. This looks interesting...
 
Interesting link on the subject
https://hw-lab.com/therma...lity.html#introduction
 
 

Great read in that link, thanks for that. Although, i don't think he really said much about strips vs individual squares
on the VRAM specifically unless i missed it. Now i'm contemplating the use of compound with the pads, but tbh, i'd 
prefer to do this by over-compensating with high quality thermal pads alone. All that extra compound would be messy. 

 Based on simple thermal dynamics, the heat transference would be greater from point of contact to the plate/heat sink than to excessive pad use. So I would think sufficient squares or strips would make no difference.
Adding the paste seems doable, probably just spread the thinnest layer possible to avoid spill out.

lebel

NucleusX

lebel

NucleusX
Can anyone else out there comment on whether to use long strips or singular squares of thermal pads on VRAM ?
The strip can cover multiple VRAM modules at the same time and will have more surface area to contact to, but
I'm not entirely sure of the downsides of this, if there is are any. Your input is appreciated !

The higher the thermal conductivity (W/m-k) the better and If possible use the lowest thickness whilst still making full contact i.e 0.5mm.
Another positive effect (see link) in lowering the temperature can be achieved through the use of thermal pads together with (non conductive) liquid thermal pastes, i.e; MX-4. This looks interesting...
Interesting link on the subject
https://hw-lab.com/therma...lity.html#introduction

 
Great read in that link, thanks for that. Although, i don't think he really said much about strips vs individual squares
on the VRAM specifically unless i missed it. Now i'm contemplating the use of compound with the pads, but tbh, i'd 
prefer to do this by over-compensating with high quality thermal pads alone. All that extra compound would be messy. 

Based on simple thermal dynamics, the heat transference would be greater from point of contact to the plate/heat sink than to excessive pad use. So I would think sufficient squares or strips would make no difference.
Adding the paste seems doable, probably just spread the thinnest layer possible to avoid spill out.

I'm aware that using the thinnest possible pad is preferable, but that seems to be a problem here. After deconstructing the
ACX 3.0, and although i didn't find much in the way of gaps, the original VRAM modules hadn't left a reasonable indentation
in the pad which caused me to choose a 1.5mm thickness to fill the slight void. Now the stock VRAM pads are 1mm, and i
don't see a way i could go below that seen as tho the GPU die itself won't allow me to go any lower. I know 1.5mm isn't
ideal, but I can't see another way around that ? I thought i could compensate for this with wide strips for extra surface area. 

NucleusX

lebel

NucleusX

lebel

NucleusX
Can anyone else out there comment on whether to use long strips or singular squares of thermal pads on VRAM ?
The strip can cover multiple VRAM modules at the same time and will have more surface area to contact to, but
I'm not entirely sure of the downsides of this, if there is are any. Your input is appreciated !

The higher the thermal conductivity (W/m-k) the better and If possible use the lowest thickness whilst still making full contact i.e 0.5mm.
Another positive effect (see link) in lowering the temperature can be achieved through the use of thermal pads together with (non conductive) liquid thermal pastes, i.e; MX-4. This looks interesting...
Interesting link on the subject
https://hw-lab.com/therma...lity.html#introduction

 
Great read in that link, thanks for that. Although, i don't think he really said much about strips vs individual squares
on the VRAM specifically unless i missed it. Now i'm contemplating the use of compound with the pads, but tbh, i'd 
prefer to do this by over-compensating with high quality thermal pads alone. All that extra compound would be messy. 

Based on simple thermal dynamics, the heat transference would be greater from point of contact to the plate/heat sink than to excessive pad use. So I would think sufficient squares or strips would make no difference.
Adding the paste seems doable, probably just spread the thinnest layer possible to avoid spill out.

I'm aware that using the thinnest possible pad is preferable, but that seems to be a problem here. After deconstructing the
ACX 3.0, and although i didn't find much in the way of gaps, the original VRAM modules hadn't left a reasonable indentation
in the pad which caused me to choose a 1.5mm thickness to fill the slight void. Now the stock VRAM pads are 1mm, and i
don't see a way i could go below that seen as tho the GPU die itself won't allow me to go any lower. I know 1.5mm isn't
ideal, but I can't see another way around that ? I thought i could compensate for this with wide strips for extra surface area. 

Wonder if using copper shims to take up some of the difference and then using the thinnest thermal pads?

Previously when I put pads on my 480 GTX cards when I attached water blocks and a water block on my old x58 EVGA mobo on the VRAM chips I cut the blocks. But when I reseated the VRAM & mobo block I used a strip cut long & wide enough to cover them my temps dropped a little. I made sure the strip was wide, long, & plenty thick to allow plenty of compression. I don't know if the strip that exists on the block between the VRAM chips allowed for that 1/2 degree extra heat dissipation but in my little brain it did! :-)

Their will be little to no difference when if you use small squares or a strip of the thermal pad.  This is from previous experience trying this on a old ati card with a water block back in the day.  I always use small squares if it is the size of a ram chip, and then a strip of thermal pad if it vrm or smaller.  Because when you put on the block or the cooler, if the square of thermal pad is really small, there is a chance it will move when you put that cooler on, and you dont want that.
 
If the thermal pad is good quality, which they all pretty much are anymore, the heat transfer will be good and no worries about having more that just covering the chip (it wont hold heat or build up heat).
 

Well my idea was to make them 1.5mm x 30mm wide strips covering multiple VRAM modules at the same time while having the
extra surface area contact. The Thermal Grizzly strips i was looking at are rated at 8.0 w/mk, and i was hoping combining that with
larger 30mm strips would yield enough of an improvement to counter the 1.5mm height that I'm pretty much forced to use.
If it wasn't for the absence of acceptable indentations in the original factory thermal pads, it wouldn't be such a problem.
 

One thing i have done in the past, that worked just fine, is if the gap between the chip and the cooler is a bit thicker than the thermal pad, is just to put a dab of thermal paste on top.  So it would be vram chip, thermal pad, thermal paste, then cooler.  Believe it or not, this does work pretty well and it is better than having a space gap with no contact.  The thermal paste does not care what it is placed on, it still has the same thermal characteristics.

dakon
One thing i have done in the past, that worked just fine, is if the gap between the chip and the cooler is a bit thicker than the thermal pad, is just to put a dab of thermal paste on top.  So it would be vram chip, thermal pad, thermal paste, then cooler.  Believe it or not, this does work pretty well and it is better than having a space gap with no contact.  The thermal paste does not care what it is placed on, it still has the same thermal characteristics.

Yeh ok, well, the addition of thermal compound to 1mm is starting to sound like a better idea to go with. I was hoping to avoid
it to be quite honest. If i had my way, i'd thermal compound the VRAMs and bring the base-plate down to flush with them, but
i don't think that's gonna happen. I was already planning to get me some Kyronaut TIM for the GPU die itself, so having that
on hand for the additional use on VRAMs should keep this thing well within tolerance. Hell, i was going to add even more strips
to other surrounding surfaces to sink as much heat as possible into their respective surfaces where air flow wouldn't be a concern.
You could say I'm on a mission from god to get this thing cooler than its ever been lol. I want it to live long and hard.

Have you seen the free thermal pads EVGA are sending out ? The VRM pads are HUGE. It had me thinking that
if EVGA deemed those sizes sufficient for the fix, then it might be wise to pay attention to the rest of the card.