25 stages curve optimizer featuring maximum power caps.
25 stages curve optimizer featuring maximum power caps.
I’m feeling really confused about this question. To be truthful, it’s bothering me a lot. I just bought and installed a Ryzen 9 5900X, and I wanted to experiment with curve optimizer and try out CB23. I’ve done this before on my older 5800X. I set the optimizer to "per core," changed all settings to negative -25, and set power limits to the motherboard—of course that pushed every limit to its maximum. Then I ran a multi-core benchmark on Cinebench 23 and went back into the optimizer in BIOS. I noticed one of my cores was actually running at positive 25. So I increased that core’s voltage by 25 steps while keeping all limits at maximum. Could this have harmed the chip or weakened that core? As someone who works with PCs, I’m a bit anxious because I feel a bit out of my depth here and made an oversight.
It won't harm the chip, especially if this was just a few runs—it mainly lowers your performance by adding +25 on one core. Most of the time, the curve optimizer maintains the same voltages while tweaking the frequency curve to use more voltage than necessary for each frequency (still within safe limits). Also, keep in mind that hitting the maximum PBO values isn't ideal. Generally, better results come from reducing them to something like PPT 300, TDC 177, or EDC 188 (these are the settings I use on my 5900X). I also have some unusual curve optimizer tweaks—like -30 across the core with +31mV DVID offset—so those details matter.
Damage is just a misleading word. Unless you physically dropped the component, it’s not really degrading. Running above 1.45v won’t hurt much, and you won’t need more than 1.4v to see any real impact. Under heavy load it might drop to around 1.2-1.3v, so you’re not causing damage. There’s also idle boosting that can handle up to about 1.5v on a few cores, but it only lasts a short time—usually when the system is just starting up or doing simple tasks. It doesn’t cause long-term wear, though it does affect performance when pushed hard.
Interesting observation. At 5900x the best cores, crashing when CO is below -20 definitely puzzles me. I’m not sure how CO interacts with other voltage adjustments like DVID offset. Can you explain more?
My setup remains consistent, the DVID adjustment removed the issue during stress tests at -30°C, even though it wasn’t the main focus of my work. PBO relies on VID and temperature data rather than actual chip voltages. Ryzen experiences clock stretching when PBO isn’t active, leading to reduced speeds despite stable OS readings. I tested a DVID offset because the chip’s cooling was sufficient, allowing me to use CO’s tighter frequency range. This improved my score by 500 points in R23 across all cores, offering better stability under light loads. I also adjusted PBO limits at the same time, finding that raising one affected performance significantly. At around 188EDC (close to optimal), I settled on 11A increments and a +31mV setting, which helped level performance. I haven’t encountered others discussing this approach with CO, but it seems counterintuitive—overvolting while undervolting. With good cooling, it might help some systems, though I’m unsure if it’s universally useful. My current rig includes a custom loop, a 5900X (or newer models), X570 Aorus Master Rev. 1.2, and 2x16GB Ripjaws V at 3800MHz CL14-13-13-21.