The curve optimizer is making educated guesses, but you haven’t provided enough evidence to confirm it.
The curve optimizer is making educated guesses, but you haven’t provided enough evidence to confirm it.
Hey, I run a Ryzen 7 5800X asus TUG X570 gaming plus. I’ve been trying to achieve a stable CPU overclock, starting with +200 and +150 settings while using the per-core optimizer. Every time I restart, it gives wildly different results. At +125 it showed a huge drop of -30 across all cores except the fastest one, which was better at 8-28. I also ran Cinebench and 3D, getting impressive scores. Sometimes I get random blue screens in regular games, which suggests instability for other cores. I wonder why this isn’t a bigger problem—why doesn’t the optimizer run for longer sessions? It seems like Ryzen Master isn’t very intuitive, and I’m not sure why AMD hasn’t fixed this more. I’m trying to figure it out myself, but the auto-curve optimizer is really frustrating.
It seems you're suggesting the curve optimizer is making random guesses, but I can't confirm that. Clock stretching might be involved, and cinebench isn't a useful performance metric. The process is straightforward: adjust values like -30, -25 until stability is achieved. If issues persist, the CPU may struggle, so consider removing the device and resetting the CMOS.
Thanks for your feedback. I noticed the numbers sometimes displayed lower values, but overall they were consistent. The core 4 and 8 each have their own distinct performance characteristics. My late-night thoughts focused on extending the 100% core runtime to ensure stability, as it seems the system mainly prioritizes the fastest and preferred cores. People often run extended stability tests to verify overall reliability. Regarding your cooler question, the Master MA620 CPU cooler should be sufficient for your setup unless you have a high-performance build.
I took a look at your profile, you have a 5800x rating. When you pushed the overclock, what results did you achieve? Your Asus TUF board uses a 7x7 VRM configuration, which means 14 phases—quite powerful. Which stress test do you prefer for CPU stability? I’ve been using this system for some time, but adjusting these newer chips feels very different from older models like the R3000 series, and it requires a lot of fine-tuning.
I don't think it's reliable in every game, though it works in most (except a few). I've tried different temperatures and it hasn't improved much. For stress testing, Prime95 or ROTTR are my go-to, but they seem very sensitive to instability. 3DMark FIRESTRIKE is my main benchmark and stress test because it's also prone to issues. My 5800x3d rarely exceeds 86°C (100% core usage), which suggests the problem might not be overclocking but rather cooling. Better cooling generally leads to better performance, so lower temperatures tend to help more than pushing the CPU higher.
PBO on zen3 Ryzen offers some intriguing advantages during optimization. You can push the clock speed to its maximum before instability across all cores, then fine-tune power settings to preserve stability—though this may cause overheating, so robust cooling is essential to keep temperatures below 85°C for optimal performance. For extra gains, you can slightly undervolt the CPU to manage heat, which might improve core benchmarks by a small margin since Ryzen 5000 models perform better at lower temperatures. Another useful approach is adjusting power limits to enhance efficiency, allowing the CPU to run cooler and potentially boosting scores in certain benchmarks. Be cautious of reviews that compare stock and PBO results without detailed tuning; actual gains could be around 15% depending on the workload. Avoid relying on Ryzen Master or automatic settings—manual BIOS adjustments provide the best control.