Timing is everything.
Timing is everything.
Could you use a gold leaf sheet or three pieces instead of standard thermal paste? You're looking to test it safely without damaging components.
It would be fine, though they could be too delicate for this. Gold leaf is incredibly fragile, so thin it won’t add much impact compared to the dry surface.
Unless it's liquid, then not. The purpose of using a paste is that it can be crushed and reach every tiny groove and hole on the cooler’s surface and the IHS of the CPU, ensuring a strong bond between them.
Carbon offers greater thermal conductivity compared to gold, making it a preferable choice in certain applications. Gold leaf is extremely thin, requiring large amounts to achieve noticeable thickness. To make it practical, the gold would need to be embedded within the microstructures of the plates, which could compromise their integrity and performance. This approach might not provide advantages over carbon cloths, which are already considered adequate.
You move smoothly over the CPU HSI plate and cooler while using gold leaf. Make sure there are no wrinkles, folds, dents, or imperfections. The heat advantage isn't strong because copper and silver actually transfer heat more effectively. It's better to opt for Silver Leaflets. I'm sorry, Bombastinator, carbon performs much worse in thermal conductivity compared to gold.
You're referring to graphite. Correct, it ranks below gold in value. Not every form of carbon is graphite. Hex link carbon exhibits much higher thermal conductivity—ranging from 1000 to 2000 compared to gold's 300.
I’m familiar with MX-4 being carbon-based and it performs well as a paste. I used a silver coin on both sides of a Morgan silver dollar during testing. People often ask about thermal pastes and metal choices, but the difference between Wm.k values is significant. For instance, MX-4 sits at 8.5 Wm.k, whereas our copper plates and blocks offer 385 Wm.k thermal pads—slightly higher at around 6 Wm.k, maybe closer to 6.5. The real challenge now is cooling efficiency. Transistor heat spikes are much more noticeable than other factors. When running Cinebench or even light loads, temperature rises sharply. There’s no way to completely avoid these spikes. Based on general testing, we can narrow the spike range from 40°C to 20°C using certain methods. I’ve tried a few approaches, but nothing works reliably at 7nm yet. Some discussions on TPU suggest maintaining a better temperature gradient with cooling solutions. This could help lower spike temperatures, though the gap between idle and load remains large—especially since modern chips can run under a single watt while still drawing less than a watt when idle or in standby.