If you have ever taken a graphics card apart, you will have met two very different thermal materials. There is the paste on the main die, and there are the soft, rubbery thermal pads sitting on the memory chips and power components around it. Repasting is the job everyone talks about, but repadding is the quieter, more advanced sibling — and on a hard-used card it can matter just as much.
This walkthrough assumes you are already comfortable opening a GPU. If you are not, start with our honest look at the risks and reward of GPU repasting in Nigeria first, because repadding always happens alongside a repaste. It also helps to understand how much VRAM you actually need, since the cards that benefit most are the high-memory workhorses.
Paste and pads do different jobs
A GPU cooler has to draw heat away from more than just the core. The main die gets thermal paste — a thin layer that fills the microscopic gaps between the chip and the cold plate. But the surrounding components sit at slightly different heights and cannot be reached by that same cold plate directly.
That is where pads come in. The VRAM memory chips and the VRM power-delivery stage each need their own bridge to the cooler or backplate, and a soft, compressible pad of the right thickness fills that gap. Over years of service these pads dry out, harden, or pump out from under pressure. Many cards also shipped with thin or mediocre pads from the factory in the first place.
In Nigeria's climate the timeline is shorter. Sustained ambient heat accelerates the drying, especially in the hotter builds of the north and Sahel, where a card never really gets a cool night to recover.
Why it matters: memory junction temperature
Modern GPUs report a separate memory junction temperature — the hotspot reading from the VRAM itself, distinct from the core temperature most people watch. On a healthy card this tracks reasonably close to the core. On a card with tired pads it can run alarmingly hot and, in the worst cases, force the card to throttle to protect the memory.
The cards most likely to suffer are exactly the ones worked hard: those used for crypto mining, sustained gaming, or AI and ML training workloads where the VRAM is hammered for hours at a time. Fresh pads of the correct thickness can drop memory junction temperatures significantly. That measurable drop is the entire reason to do this job.
The one variable that decides everything: thickness
This is the part that makes repadding genuinely advanced. Thermal pads are sold in specific thicknesses, measured in millimetres, and matching them is make-or-break:
- Too thin and the pad never touches the cooler — no contact, no cooling, and you have made the card worse than before you started.
- Too thick and the cooler cannot seat properly. You get uneven pressure across the board, the core paste contact suffers, and you can physically stress the chips and solder joints.
- Common values run roughly 0.5mm, 1.0mm, 1.5mm and 2.0mm — and a single card often needs different thicknesses at different spots. The VRAM might want one value while the VRM bridges a larger gap to the backplate.
So you cannot buy one sheet and pad the whole card. You either carefully measure the old pads before removing them, or you research the exact thicknesses your specific card model requires at each location. Getting thickness wrong is the single most common way this job goes badly.
Alongside thickness, choose a quality pad with a good thermal conductivity rating, quoted in W/mK. A higher figure moves more heat — but it is worthless if the thickness is wrong, so settle thickness first.
The process, step by step
Because the cooler has to come fully off, repadding overlaps almost entirely with a repaste. Do both in the same sitting:
- Fully remove the cooler and backplate, working slowly and keeping every screw organised by length.
- Before touching anything, note which pad sits where and its thickness — photograph each one in place if you can.
- Peel the old pads away gently so you do not lift any tiny surface components with them.
- Clean off the residue from both the chips and the cooler with isopropyl alcohol.
- Cut the new pads to the size of each chip — neat rectangles, no overhang bridging two components.
- Place the correct thickness on each VRAM chip and each VRM block.
- While the die is exposed, repaste the core with fresh paste — there is no reason to do this twice.
- Reassemble carefully so nothing shifts the pads out of position as the cooler comes down.
- Tighten the cooler screws gradually in a cross pattern to keep pressure even across the whole board.
After it boots, watch the memory junction temperature under load. A clear drop tells you the pads are seated and doing their job.
The risks and the warranty
The honest framing is identical to repasting, only more so. Repadding voids your warranty on most cards. The teardown is fiddly, the components are delicate, and it is genuinely easy to fit the wrong thickness and walk away with a hotter card than you began with.
This is firmly an advanced job. If you are not confident measuring pads and reassembling a cooler with even pressure, the cost of a mistake — a stressed VRAM chip or a cracked solder joint — far outweighs the few thousand naira a professional charges. There is no shame in handing it over.
When it is actually worth it
Repadding is not routine maintenance. Reach for it only when the symptoms and circumstances line up:
- An older card showing high memory junction temperatures or throttling tied to the memory rather than the core.
- A card that is out of warranty, so there is nothing left to void.
- A hard-used card — ex-mining, heavy gaming, or sustained AI and ML work — where the pads have genuinely had a beating.
- A card you intend to keep for several more years, making the effort pay back over time.
If your card runs cool and behaves, leave the pads alone. And think of repadding as a partner to the die repaste rather than a separate ritual — the same cadence thinking in our repaste schedule by cooler type applies here. When you are already inside the card for paste, that is the moment to assess the pads too.
Frequently Asked Questions
How do I know what pad thickness my card needs? Measure the existing pads before you remove them, or look up your exact card model — community teardown guides usually document the thickness at each location. Never guess; a single wrong value undoes the whole job.
Can I repad without repasting the core? You could, but it makes little sense. Removing the cooler to reach the pads already exposes the die, so repaste at the same time rather than disturbing everything twice.
Will new pads always lower my temperatures? Only if the thickness is correct and the old pads were genuinely degraded. Fresh pads of the right size on a hard-used card can drop memory junction temperatures noticeably, but a card with healthy pads has little to gain.
The One Thing to Remember
Thickness is everything. A premium pad in the wrong thickness will cool worse than a cheap one in the right thickness, because cooling depends entirely on solid, even contact between the chip, the pad and the cooler. Measure first, buy second, fit third — and never reverse that order.
Not sure your card is worth the teardown, or want it done safely by someone who has repadded dozens of them? Spec a cooler-friendly build with our configurator, or contact us to talk through your card's temperatures before you pick up a screwdriver.