Yes, it is possible to adjust the voltage for E cores independently from P cores.
Yes, it is possible to adjust the voltage for E cores independently from P cores.
Hello, your goal is achievable. You can boost only the E cores while keeping stability at 4.9Ghz. Adding extra voltage to the whole CPU can increase heat, so focus on adjusting the voltage for those specific cores. This approach helps maintain performance without overheating the system.
It seems you're exploring ways to improve stability by adjusting voltage levels. Consider testing an offset to see if it resolves issues caused by voltage rollover. If not, it might be due to other factors. Be aware that keeping the system cool is crucial—lower temperatures reduce power consumption and allow higher clock speeds, though achieving subzero temps requires a powerful phase change cooler. The relationship between temperature and performance is inverse: higher temps mean lower power use for the same voltage, and faster clocks if managed properly.
This would still be manageable as long as the system stays insulated, because any heat would just stay frozen. Making your own or tweaking an existing cryocooler would be difficult since normal AC units don’t reach the needed extreme cold. It seems the temperature should be near or below -100°C to make it worthwhile, otherwise you’d face serious challenges in maintaining performance. Once you have some extra money, experimenting with subzero cooling could be possible, but it would likely cost a lot—hundreds of dollars—to build an effective phase change cooler that works down to -100°C or colder.
Have you explored positive offset on the ecores yet? It’s a bit of a stretch looking at the Reddit discussion, but who knows if it’ll work. If it doesn’t, and there are no other options, then your only choice seems to be upgrading to a better cooler or searching for a higher-end 13900ks or 14900ks sample if you’re really committed to a 4.9GHz setup. The problem here is that the device just can’t dissipate heat quickly enough, so cooling becomes your main concern. This means you’d either have to handle condensation risks and maintain it below zero constantly, or you’d need someone to build and install it for you. It’s like a hard tube custom loop but much worse in terms of performance and upkeep. And if you’ve already ruled out the 13900ks, finding a better sample might be even harder—maybe something with 5GHz ecores and higher PCORs, or possibly a 6GHz option, though that would likely require sacrificing RAM speeds. You’d still face the challenge of finding chips that work at those frequencies without overheating. It’s tough, especially when you consider the long list of components you’d need to source and assemble. As someone who really loves overclocking, I’d probably go for a phase cooler instead, aiming for something capable of -100°C or lower. I’ve seen phase coolers using the Joule-Thomson effect reach -170°C, which would be great for liquid nitrogen applications. Hopefully, someone can afford the time and money to replicate that setup soon—extreme subzero overclocking sounds pretty cool!
These new chips seem problematic. Either they’re too rigid for your setup, forcing you to overclock heavily or resort to extreme cooling methods like subzero temperatures. Or you’d have to go all-in on phase cooling and spend a lot of money building a top-tier cooler to maintain stable temps. The only practical way forward is to keep things cool consistently, which demands significant effort. Unlike older systems that handled heat better or dissipated it more efficiently, these newer designs generate excessive heat and require costly solutions. Commercial phase coolers like the Prometheia M.II GT are already available, but you’d need an even more powerful unit to actually make the CPU coolable. This would involve a lot of work, higher power consumption, and likely result in a costly project. Also, pushing temperatures below 0°C risks thermal failure or crashes. A cascade cooling system could help, but it would need substantial capacity and special refrigerants, adding complexity and expense. If you’re only building a CPU block for testing, connecting it to a standard AC might be the cheapest option, though it won’t handle extreme cooling needs.