What method is most effective for verifying 3900x stability?
What method is most effective for verifying 3900x stability?
Hello everyone, I've been testing in Ryzen Master and achieved a 4.15 across all cores at 1.2V. I ran Cinebench R15 & R20, Timespy, and Handbrake encode for around five minutes and everything has stayed stable so far. What should I use for longer tests and how long should I run them? Please keep in mind my cooler isn't top-of-the-line—it's the CM Hyper 212 LED Turbo—and I don’t want to go beyond that voltage because I was seeing issues in Cinebench R20. My motherboard is an ASRock x570 Creator.
Also, CPU-Z shows a steady voltage of 1.216V even when the CPU is completely idle. Shouldn’t voltage drop when the CPU is off? I’m concerned that keeping it at higher voltage longer might affect its lifespan, especially since I often leave my PC running in idle mode for extended times.
EDIT: I currently have a stable overclock and good temperatures. My main worry now is controlling the CPU voltage and clock during idle periods. If you don’t have any suggestions in that area, feel free to skip your reply.
Auto sucks, I have seen voltage going into 1.4x stable and at load its always around 1.36x which easily push temps into high 80s and/or 90s and that's all at 4.05 GHz all core. Doing 4150 all core at 1.20 i have beaten multi-core score in cinebench R20 and almost reached single core (about 3% slower) while reducing temps by as much as 10-15C which is very significant. Now under stress i am not hitting above 75 C.
Also using auto i never saw single core going above 4.3 GHz. This is also my initial overclock. I have not pushed 4.175 or 4.2 GHz on 1.2V yet. I WILL if i am stable at 4.15.
Sorry but i am simply not going back to crazy 1.36x V. Just sitting at desktop i can see temps touching higher 50s or even 60s doing nothing.
Auto functions properly and stays within AMD's limits. The lower boost performance comes from high temperatures – the cooler isn’t doing enough. Its boost response closely matches Nvidia’s GPU Boost, which also depends on heat. The Ryzen 3000 tends to be quite unstable, keeping idle temps up while handling less load than rivals. Even when idle, the system remains active with background tasks. Your setup is typical for your budget cooler, and using it like an Intel CPU isn’t optimal either.
I don't follow you. What you're suggesting is essentially leaving it on auto at 1.368V @ 4.05 GHz, which is better than 1.2V @ 4.15 GHz.
It seems a large radiator in a custom loop would be needed to achieve less than 80s under stress when the auto voltage reaches almost 1.4V while not providing much speed.
I'm not comfortable with that voltage and am returning to my original question.
That was the best utility for long stress tests.
The performance increase is limited to 4.05ghz due to your cooler. The 3950X doesn’t generate much extra power, so a radiator becomes the only choice. The 9900K still consumes more power and typically requires a radiator of at least 280mm. The NH-D15S, Dark Rock Pro 4, and Freezer 34 esports Duo are solid air cooling options for your CPU; a Hyper 212 won’t suffice. You’ve already completed one of the top CPU stress tests, but it wasn’t long enough. A comprehensive system test is necessary eventually. After all, some GPU heat also flows into the CPU heatsink. If you observed the CPU reaching 90°C during short sessions with R20, Asus Realbench, AIDA 64, and Folding@Home, the heat will be significantly higher. Simply upgrade to a better cooler—large radiators aren’t essential for this CPU.
The optimal method for boosting the 3900x without losing control is using a chiller. This allows you to achieve high frequencies while managing heat effectively. Under load, water temperature would rise to around +3, and the CPU would spike to tens or even low twenties. A video showing a 4.75GHz overclock at 1.4 volts is typical. Maintaining low temperatures is essential for optimal performance but can be costly. Even with water cooling, reaching very low temps remains challenging. With my 3800x and a 360 rad 6 fan setup, an all-core 4.4GHz would push me over 80°C in Prime95 8FFT. More cores lead to progressively worse results as temperatures climb. As temperatures increase, clock speeds drop.
For the 3900x, strong cooling is crucial. I’d opt for the best AIO you can afford or even better. The largest radiator available, like the Watercool MO-RA3 360 PRO Black (25020), won’t help if it can’t maintain temperatures below ambient air. Money would be better spent on a chiller.
Maintaining low temperatures is key to achieving higher clock speeds. A stable low temperature environment yields the best results. Closer to zero is ideal for maximum performance. Within the 0 to 25°C range, you’ll see the most significant boosts from the CPU. At these cooler levels, expect around 4.26GHz all-core overclock with no additional effort.
I believe I didn’t clearly state my goals when starting this discussion. My aim isn’t to achieve the highest possible overclock for my chip.
My goal is to obtain the optimal multi-thread clock speeds from all cores while running at low voltages. I’m not interested in pushing my processor into the 1.35x or even 1.4x ranges.
I’m a user who works with multiple threads, especially during video encoding which requires full utilization of threads. If I can secure an additional 100 MHz at lower voltages, it would be acceptable. Performance improvements from single-core speed aren’t worth it for me.
<Mod Edit>
To me, that’s the only reason I’d consider a 9900K. In games, I’m often limited to 4K on a 1080 Ti (excluding the 2000 series), so I’m always constrained by GPU performance. When using auto single-core boost, I reached around 4300 MHz. If my single-core boost drops by another 150 MHz in games, it won’t significantly impact my experience since I’m already gaming at 4K.
Also, many people assume everyone worldwide is in the US or EU and can easily access top-tier components at lower prices. I never asked for recommendations on coolers, but I’m very aware of the best options available today. Since most discussions drift away from my specific needs, let’s check how much a Noctua NH-D15 would cost if I bought it on Amazon.
I’m confident in what works best for my workload and budget compared to others here. I won’t purchase the NH-D15 because it isn’t available locally for $370 USD.
No one has addressed my actual question beyond sharing their own opinions about coolers.
One of my concerns was whether changing voltage and core frequency in Ryzen Master locks performance even when idle. Is there a way to reduce frequency and voltage when I’m not doing much, similar to auto mode?
Finally, I was exploring silicon options and it seems I have a chip that can reliably run at 4.15 GHz all cores at 1.2V. It would be pointless to rely on auto settings that raise the voltage to 1.4V just to gain a few MHz on a single core.
In line with your goal of achieving optimal multi-core speeds at low voltages, I fully concur with others who recommend setting it to AUTO. You can find a useful guide in this video where Buildzoid demonstrates how to adjust PBO settings for peak performance across various conditions while maintaining safe operation levels. This approach ensures stability during demanding tasks such as heavy workloads or sudden bursts, and even under extreme stress tests. All devices respond well under these settings.
It's also crucial to recognize that voltage alone doesn't cause CPU issues under load; it's the accompanying temperature that matters. When AUTO is enabled, the CPU maintains its protective measures by reducing frequency and voltage as temperatures rise. Many users who stick with manual overclocks at voltages they deemed safe are discovering unexpected degradation. This might be due to the 'safe' voltage being higher than what the boost algorithm permits, preventing the CPU from safeguarding itself.
Ryzen's thermal behavior is quite restrictive, meaning better cooling directly improves sustained boost performance. However, operating below ambient temperature—such as with a chiller—has only been practical for short periods, especially when aiming for high benchmark scores in competitions. This is largely due to condensation on components that leads to early failure, regardless of the effort applied to mitigate it. For serious attempts, using liquid nitrogen is the most effective solution.
To address your query, you may try Prime95 v.26.6 or the latest version without AVX enabled. Let it operate for several hours to check for crashes or halted processes. When you increase the clock speed and apply a fixed voltage, it will maintain that setting continuously. You can explore P-state overclocking to determine if it suits your needs.