Thermal Paste vs. Thermal Pads
These are not two options for the same job. They solve different problems. Choosing the wrong one causes overheating regardless of how carefully you apply it.
The key difference: paste fills microscopic surface imperfections on flat surfaces where the gap after mounting is near-zero. Pads bridge a real, measurable gap between a chip and a heatsink plate. These are fundamentally different tasks.
For flat die-to-heatsink contact
Thermal Paste
- +Maximizes heat transfer on flat surfaces
- +Works on CPU die, GPU die, laptop die, console APU
- +Lower thermal resistance than pads on flat-to-flat contact
- −Does not fill gaps; semi-liquid, will not stay in place
- −Wrong choice for VRAM, VRM chips, or M.2 SSDs
For gap-filling between chips and plates
Thermal Pads
- +Fills gaps of 0.5–2mm+ reliably
- +Works on VRAM chips, VRM chips, M.2 SSDs
- +Pre-cut to size; no spreading required
- −Higher thermal resistance than paste on flat surfaces
- −Wrong choice for the main CPU or GPU die
Why most GPU repastes need both
A GPU has multiple components that need thermal management, not just the main die. The die gets paste because the heatsink sits flat against it with near-zero clearance after mounting. But VRAM chips and VRM components sit at different heights and have a real gap between them and the heatsink plate, typically 0.5–1.5mm depending on the GPU model.
Applying paste where a pad should go means the paste cannot bridge that gap. The VRAM chips run hot, throttle, and may fail over time. Applying a pad on the main die instead of paste adds unnecessary thermal resistance at the most critical contact point.
Standard GPU repaste procedure:
- Thermal paste on the main GPU die (center dot method)
- Fresh thermal pads on all VRAM chips (measure original pad thickness before removing)
- Fresh thermal pads on VRM components if the originals are dried out or damaged
Two mistakes that cause overheating
Mistake 1: Paste where a pad should go
Applying paste to VRAM or VRM chips results in near-zero thermal contact. Paste is not designed to fill a gap; it will simply sit at the surface without bridging to the heatsink plate. The chips will overheat. This is a common cause of GPU VRAM failure after a repaste.
Mistake 2: A pad between CPU and cooler
Using a thermal pad on a flat CPU die instead of paste adds significant thermal resistance. Pads have lower thermal conductivity than high-quality paste and a higher contact resistance on flat surfaces. CPU temps will be noticeably higher than with paste.
Quick reference
| Component | Paste | Pad |
|---|---|---|
| CPU die (desktop or laptop) | Yes | No |
| GPU die (main chip) | Yes | No |
| GPU VRAM chips | No | Yes |
| GPU VRM components | No | Yes |
| Console APU (PS5 / Xbox) | Yes | Sometimes |
| M.2 SSD | No | Yes |