What is the most effective way to overclock an AMD system?
What is the most effective way to overclock an AMD system?
I have configured my 3900x in BIOS with the following settings:
D.O.C.P. 3200mhz FCLK 1600
PBO Manual
PPT/TDC: 240A
EDC: 10
Scalar: Auto
C-states: Enabled
Power Supply Idle Control: Low Current Idle
CPPC Preferred Cores: Enabled
CPPC: Enabled
Power Plan Balanced (set with processor 85% min 100% max)
Running cpuinfo shows performance jumps to around 4.350Ghz – 4.375Ghz depending on the tool used. These results appear in tools like cpuinfo, Aida64, or games such as Minecraft. I’m unsure what’s causing this variation or how to optimize it for stability and longevity.
My main activities are streaming/recording and gaming, with occasional video rendering and photo editing. I rarely change settings but might consider animations in the future.
What’s the optimal approach—static, PBO, or something else—to achieve top performance while protecting the system?
Already have some 'how to' ideas, but what really works for you is still unclear. For me, the most effective approach will be PBO. This is because it seems to be the safest option, as it keeps the boost algorithm in charge, automatically reducing power and voltage when needed to safeguard the CPU.
But also: trying to overclock a second or third generation CPU often doesn't lead to genuine overall improvements.
Pass the specifications to your build in the body of the thread. If you think others should check your sig space, just let me know. The reason is that these specs can change over time, so the thread and suggestions might not be useful anymore for someone currently in the same situation. You can also check the video by Jayz2Cents if you'd like more details.
That's correct, I'll make the necessary adjustments with the system specifications. Thousands of different configurations are possible, and I'll omit the specification details. Sorry I haven't been active here for quite some time. The video clearly outlines the voltage and other settings to follow. I'll also review that Reddit post, as it might help clarify things. I didn't find a similar one there about a year ago.
There are already several suggestions provided, but what truly matters is what works best for your situation. For me, the most effective approach will be PBO. This is because it appears to be the safest option, allowing the boost algorithm to manage and reduce voltage when needed to protect the CPU.
Additionally, static overclocking on a second or third generation CPU seldom leads to genuine performance gains. It tends to benefit only more demanding multithreaded tasks; otherwise, lighter threaded games may actually suffer. This happens because the stable clock speed achievable with standard cooling is usually lower than the peak boost speeds the CPU can reach when optimized with PBO on lighter workloads (such as gaming). Most people I know who have statically overclocked to maximum frequency either used excessively high voltages (which shorten CPU life) or gave up on stability for heavy threads, or both.
It’s crucial not to overlook cooling when choosing any overclocking method: use the best possible CPU cooler you can fit and ensure proper case ventilation. Although you might not notice the difference at lower temperatures, better performance with PBO becomes apparent because the boost algorithm actively seeks temperature, increasing output until it reaches the workload's threshold.
I’m also aware of the "EDC = 10" trick, which works for my 3700X as well. You could experiment with adjusting it—some users report optimal results around EDC = 15 for a 12-core CPU, but results can vary depending on your motherboard and BIOS version.
Keep in mind that clock speeds fluctuate too much to rely solely on them when testing PBO effectiveness. The most reliable method is to use benchmark scores like CB20. It’s important to check both MT and ST settings, as static overclocking can cause one to spike while the other drops significantly. Also, test various EDC values to see what gives you the best balance.
This is very helpful thank you for all this information. I will try with EDC I usually just go in increments of 5. Scalar wasn't doing anything for me so got rid of it. I'll probably go with PBO in the end. I remember seeing 4.3 boosts occasionally in CPU intense games. Recently been feeling slower than it should be. This is a fresh windows and motherboard just got RMA return. Maybe I'm missing something.
At EDC level 10A you generally need to adjust in small steps. When working with the 3700X I noticed that lowering it from 6 to 5 brought noticeable improvement. At around EDC 12 it wasn't performing well, and by 15 it was barely functional. I also discovered that turning off Advanced C-States hurt single-thread performance significantly. I’m unsure why it still worked for MT but not ST tanks, but I remember this EDC issue as unusual behavior in the boost algorithm, so it shouldn’t be expected to work consistently. Just try different settings and see what gives the best results.
Regarding voltage, I haven’t found much guidance on a safe setting for 2nd and 3rd generation Ryzen CPUs with fixed overclocking. Besides the warning to stick to AUTO mode, AMD hasn’t provided much advice either. We do know it can handle higher voltages up to 1.5V, but only when core temperature and current are low—such as during single-core boost on light threads. People used to believe a max of 1.25V was safe under heavy loads (with high current), as long as temperatures stayed down. But determining what’s acceptable is tough; cooling a 7nm chip is extremely challenging due to limited surface area for heat dissipation.
The boost algorithm is the only one that can quickly reduce voltage and clocks when temperatures rise too much on light threads. It relies on FIT data to maintain safe limits on voltage, frequency, temperature, and current. That’s why I prefer PBO.
I have one technique I’ve learned that helps me configure my overclock: I’ll also share the article if the link becomes unavailable. This approach works for both the 3000-series and the 5000-series. "For regular use, aim to match the voltage used by the CPU when it’s under load and during overclocking. Enable PBO and let everything else auto. Run Prime95 with a 128k FFT, leave it in place without ticking, and monitor the v-core (SVI2 TFN). This will be your chip’s fitness voltage. Variations exist due to cooling, which significantly affects the boost process." I’m familiar with the Buildzoid method. I watched his video on it. Choose the lowest voltage your CPU reaches—it’s your fitness voltage. My 5600X was around 1.285v during initial FIT tests, so that’s what I targeted under heavy loads. I adjusted the LoadLine Calibration (LLC) to reach that level. Some chips stay below 1.3v, others exceed it. Each unit behaves differently because of cooling differences, which plays a big role in the boost algorithm." I know this as the Buildzoid technique. I tried his video and followed the steps. The static voltage is set at 1.3v, but with my LLC settings, it drops to 1.285v during heavy loads. The CPU runs smoothly at 1.3v under normal conditions. Stock PBO usually keeps the CPU at 1.4–1.45v during light use. I adjusted the temperature a lot to achieve this. At idle in Windows, with PBO, it’s around 50°C; with static OC, it’s about 35°C. I also perform a static overclock for stability reasons. The fans would fluctuate when using PBO. A frequency of 4.45GHz is what works best for me. Your clock speeds will change, and so will your fitness voltage. Every silicon chip is unique, regardless of model. Even if you skip the static overclock, the fitness voltage limits still apply. Prime95 and SmallFFTs are useful tools. Cinebench isn’t a heavy load, so don’t rely on it for overclocking. If you push beyond Cinebench, you might face instability or crashes. To put this in perspective, my CPU reached 70°C during Cinebench runs, while Prime95 hit over 90°C. That’s a big difference. There’s a Reddit community dedicated to overclocking: https://www.reddit.com/r/overclocking/ You can search there for your CPU model and see what others recommend. You might also post your own questions or threads—some experienced users often contribute helpful insights.
I had EDC on 10-12 for a while and now have PBO as well. My cooler performs well, maintaining idle temperatures around 36-38°C. I’m struggling to reach over 4050 MHz in prime 95, regardless of what I do with PBO and settings. I’m not getting any meaningful results. It might be time to switch to a static OC. I used to get around 4.3ghz, but I don’t know why that changed. That observation came up while gaming, though. It was basic PBO enabled; only D.O.C.P. and a high FCLK made a difference. Of course, clocks on prime95 were stable at around 4.1ghz, with voltage at 1.280 in the "current" section after 20 minutes of testing. I’m trying again now. C-states are disabled. Offset might help as before, but I’m not sure.
Report: I’m still hoping it might be stuck at around 4.075ghz. The highest stable speed I achieved was about an hour on prime95 with Vcore at 1.280. I’m uncertain about the LLC setting, since it seems unused on my Intel chip. Probably just expecting too much—my Intel only managed 5.0ghz with 1.33V, which was a lucky result given the low voltage. I definitely won’t reach 5, but I wanted at least 4.4.
It might be about the most promising outcomes from a 3900x in P95 with PBO, though I'm not entirely confident. Clearly, much also relies on the FFT size being examined. Core currents and powers change significantly depending on the FFT used, as they appear to interact differently with the CPU's internal components.
When performing tests on P95 using a fixed OC, it's essential to turn off AVX. It seems almost impossible to find such tightly packed and efficient AVX instructions in code, and doing so could risk damaging CPUs that can't adjust their clock and voltage to safeguard themselves.
Another point: P95 serves as a stress test. It helps verify stability under extreme conditions but mainly challenges the cooling systems of PCs with limited thermal capacity—most modern high-performance desktops are likely affected. This is particularly relevant when running PBO on Ryzen, as lower clock speeds might indicate the boost algorithm is conserving power by keeping the CPU within its ideal operating range.
Personally, I don't see much value in P95 beyond stress testing. For real-world performance evaluation, I focus on typical workloads that match your needs. A benchmark video rendered in Handbrake works well for this purpose.