Need guidance on optimizing Ryzen 5900X with PBO and managing thermal limits.
Need guidance on optimizing Ryzen 5900X with PBO and managing thermal limits.
Hi everyone! I have a few questions about my setup and recent changes. I’ll start by explaining what happened before I ask your points. My system includes: a Ryzen 5900X CPU, Aorus Pro B550 V2 chipset, HyperFuryX RAM at 3600MHz with XMP enabled, an NZXT X73 PSU, the 3090 Founders Edition GPU, and a 850W Corsair RMx 2021 cooler. After assembling everything, I noticed the CPU temperature would spike to around 90°C. I tried adjusting frequencies, voltages, and BIOS settings to see if it would help. Eventually, I updated the BIOS and followed a YouTube guide on using PBO. I remember setting some values like PPT EDC and another number, but I’m not sure of their exact meanings. After that, I enabled curve optimization and managed a stable 25-step temperature drop. Now I’m running Cinebench R23, Blender rendering, and multithreaded apps. All cores stay around 4.45–4.5 GHz with temperatures at about 75°C. I’m planning to use this build for Blender work, which requires lower temps and more stability.
Questions:
1. Is a temperature of 75°C acceptable for the Ryzen over long periods?
2. Is lowering thermal limits a good way to protect my PC? I’m unsure if it’s a solid long-term fix since I still don’t fully grasp what these settings do.
3. Based on what I know, if I switch to a less powerful cooler, will the temps stay at 75°C but frequencies drop? Does this mean performance depends more on temperature than frequency?
4. My current results show a max temp of 75°C and core temps between 4.45–4.5 GHz with about 1.1V. Are these satisfactory?
Thanks for all the help, especially since I’m still learning! (I’m currently at work, so I can share screenshots later if needed)
75C fits perfectly within the range. Values below 80-85°C are ideal and won’t affect longevity. Keep temperatures under 90°C for optimal performance. The Ryzen will adjust its speed based on demand and voltage limits. You don’t need to push it beyond its design; it handles itself. Avoid swapping out your cooler for a weaker model—it’s sufficient. A less powerful cooler will cause temps to rise into the 80-90°C range, which can slow the processor. You purchased this unit for a performance boost, so extreme cold isn’t an issue. Unless you’re dealing with NAND drives, they prefer warmth during use and coolness when idle for better retention. The controller needs cooler conditions, so read up on your drive specs if unsure. Let the machine run as is; don’t interfere unless you understand what each setting does. If it’s functioning smoothly, leave it alone.
Hey! Thanks for the input! Don't worry, I don't intend to change the cooler, the only thing I want to change is the fan/pump curve. I set them both to 100% for the moment, because I wanted the best cooling available while I figure out the what's and how's. I asked that because I intend to lower them to a more acceptable ~50-60%. The thing is, the cpu didn't reach 75C on its own (or to say otherwise: 75C is the temperature my cooler manages to reach for my cpu). No. I actually set in BIOS that the cpu can boost its frequency until it reaches 75C and after that no more boosting. This has resulted in 4.45-4.5 ghz all cores under load. And this is what I'm asking, if I was wrong setting it to never pass 75C
You’re dealing with a situation where you shouldn’t be exposing the stock to extreme heat, particularly with a 360mm rad AIO setup. The recent chip update probably only tweaked power limits, not actual performance. Ryzen processors rely on the motherboard’s data to gauge how much power they can safely use; if it underreports, the CPU might draw more than intended, leading to overheating despite the promise of better efficiency. Undervolting with a curve optimizer is beneficial for cooling, but for Ryzen—especially Zen 3—it’s about finding the right balance. Power and thermal limits are the main constraints; otherwise, performance will suffer. A per-core approach is recommended rather than a blanket reduction. Use OCCT to adjust one core at a time in BIOS to -30. Perform a test run with intensive AVX2 workloads on that core alone (advanced settings, skip all else). Observe for 15 minutes. If errors decrease by five levels and it runs again, proceed. If not, lower the setting further and retest. Continue until stable, then move to the next core. This process may take time but ensures reliability.
For PBO tuning, start with BIOS limits set to "motherboard" and run Cinebench several times. Employ a tool like HWINFO64 to track PPT, TDC, and EDC values (in amps). In BIOS, switch to manual mode and apply these numbers. Gradually cut EDC by 5-10A each time and rerun Cinebench multicore multiple times to find the optimal balance. Watch for a noticeable drop or spike in scores. The EDC range can vary widely—some users report values between 110 and 220 for peak performance. Explore different settings to discover what suits your setup best. After stabilizing EDC, bring PPT down to 100% (verify with HWINFO64) and ensure TDC also hits 100%. Once confirmed, repeat the process for all cores. This method guarantees thorough optimization without risking instability.
Wow, thanks a lot for the thorough explanation! I'm working with the Lian Li O11, having an AIO mounted on the side to prevent hot air from passing through it. There are nine fans in total—three top and three bottom. It’s still confusing why, with the old BIOS, I was getting around 90°C even at stock settings (I remember running at 4.5 GHz all cores). After updating the BIOS, under load the CPU hit about 81°C, which is quite high given the room temperature. When I adjusted the PBO settings as per the tutorial, it stabilized to around 81°C, though it’s still a bit hot. I’m planning to find the stable cores at 30 negative steps next, but I didn’t have time yesterday and realized it takes a while. Also, with PBO on and the curve optimizer active, I’m seeing 90°C, which seems normal? I’m not sure, but I set the thermal throttle limit to 75°C in BIOS, and that’s the current max temperature.
The post shares insights on optimizing performance with specific hardware configurations. It discusses running a new build using a 5950x processor and a Silent Loop 2 280mm cooling solution, noting peak temperatures around 88°C during testing. The author suggests that these chips can handle high temperatures, so they’re not overly concerned about overheating. They mention experimenting with radiator placement on the front panel to assist cooling. The focus shifts to maximizing core clock speeds for demanding tasks like Houdini simulations and Mantra rendering. The writer is currently working on optimizing PPT, TDC, and EDC settings. After a significant drop or sudden improvement in scores, they recommend returning to the previous high score as a stable baseline. They plan additional benchmarks at lower amperage settings and seek clarity on what qualifies as a "dramatic" Cinebench jump—expecting improvements in the hundreds or thousands if the trend continues.
If the drop is notable, revert to the last stable setting. If it increases, keep it unless you're still evaluating options. Record the strong results together with your scores and decide at the end which performed best. A jump usually means a change of a few hundred points, not thousands, though smaller shifts are possible too. The goal is identifying the value that delivers the highest performance. This might be a slight improvement over another setting, but it's still advantageous. Consider balancing factors like temperature as well—slightly higher EDC might mean a lower score but less strain on the CPU. It depends on personal preference.