Are you checking if PBO exceeds the specified maximum clock speed?
Are you checking if PBO exceeds the specified maximum clock speed?
Need to handle idle temperatures carefully, especially with Ryzens.
With Intel, during idle the CPU reduces speeds and voltages, yet all cores remain active. Background tasks, services, and processes are distributed across every core, causing only a slight increase each time it starts. Typically, idle temps around 10°C above room temperature are observed, since each core is considered the hottest one at that moment. Usually, one core runs at about 31°C while another reaches 33°C, then both return to 31°C as another process begins and a different core rises to 33°C. You'd notice just the frequent switching of cores as the hottest.
For Ryzen, only one core is running; the rest are idle. The full workload is placed on a single core, making it the most likely to reach the highest temperature. Every time a new process starts, it adds to the load, and others release, resulting in small spikes in temperature depending on the demand. This can lead to idle temperatures ranging from around 20°C to over 40°C above ambient. It's not unusual for a spike to jump from 40°C to about 60°C, and certain software might display abnormal readings due to how it interprets temperature data. Ryzen Master averages the last three seconds of activity, so a sequence like 40, 40, 70 would register an idle temperature of roughly 50°C. Some applications only check once per second; if they detect a spike at 70°C, they might report a temperature of 70°C for idle, which can cause confusion among users. If it reads 40°C, it might suggest unusually good cooling performance, but in reality, most cores are closer to 30°C and only one reaches above 40°C.
What’s particularly surprising is that when you trigger a small task—like moving the mouse or loading a webpage—the remaining cores quickly activate, causing temperatures to drop from the higher idle level as the load shifts.
Temperatures should be interpreted with caution; compare them to your specific cooling setup, CPU characteristics, and airflow conditions. Only worry significantly if they change dramatically.
Hey everyone, I wanted to check if higher idle temperatures were harming the chip. But all of you and finally AMD confirmed it's just normal Ryzen behavior. The real issue is under load temperature, and with PBO I'm seeing stable performance at 4.5 across all cores even at 72% max load.
I could try pushing it further, but it doesn't seem necessary.
Idle temperatures are also lower than expected.
I just hope AMD improves the chipset installer—updating drivers is really frustrating.
I observed something interesting when PBO was enabled at Advanced level. With all cores at 15, using negative curve optimiser, the temperature reached 85 during thermal tests with Prime95 small FFTs. That’s close to the TJmax I’m comfortable with. Lowering it to 8 brought it down to 82, which is still decent but not ideal. However, running Prime95 small FFTS with PBO set to auto gave temperatures around 60, and no thermal problems appeared—idle temps were 58 under full load without increasing. Is this the expected behavior?
Various tests in prime 95 generate distinct loads and temperatures. If you aim to assess the maximum thermal load, perform the stress test. It will execute different procedures. The initial few tests are lighter, but the heavier ones activate afterward. Keep monitoring the temperatures; if your cooling isn't sufficient, it may overheat and reduce performance.
I believed the test using small FFTs was the optimal choice for thermal testing and stability?
I perform it with standard 5900X configurations and the results are satisfactory. However, when I apply a minor -5 offset with PBO, the thermals increase by nearly 20 degrees.
It's puzzling since running it at stock settings doesn't seem to impact the thermals much, contrary to what I expected.
Small FFTs put the most strain on the processor section of the CPU.
However, this applies only to an AMD CPU.
The remaining component, the I/O die, contains the memory controller, PCI-E, SATA, USB, etc.
This combined stress tests all parts of a CPU and memory system.
A more comprehensive evaluation for overall system stability.
It's quite unexpected since typically only a few cores need higher voltage during boosts, especially those with better performance. These are often the top performers. Some users have even noted that applying a positive curve offset to these cores can enhance their boost results for CPUs.
Testing for stability usually doesn't lead to crashes under heavy threading or full-core stress because the cores aren't pushing for maximum clock speeds. This issue tends to appear in lighter, more intermittent workloads. My machine may crash during a Defender quick scan. A useful tool is Core Cycler, which runs a single-threaded Prime95 FFT test per core, clearly identifying failures and allowing precise offset adjustments.