High temperatures occur during single-core operations on the AMD 5900X processor
High temperatures occur during single-core operations on the AMD 5900X processor
Hey everyone! I just wrapped up my first AMD CPU build with these details: processor – AMD Ryzen R9 5900X MB, motherboard – ASUS CROSSHAIR VIII DARK HERO X570 (BIOS 4402), cooling – LIAN LI Galahad 360 (first time using AIO), RAM – G.Skill Trident Z Neo DDR4 32GB 3200MHz, graphics – MSI RTX4080 Gaming Trio X SSD M.2 Samsung 970 EVO Plus 250GB, power supply – be Quiet! Dark Power 12 850W case, case fans – be Quiet!, PSU – pure WinFS 2 with 3 intakes and 1 outlet OS, Windows 10 x64 22H2. The CPU isn’t overclocked yet; the only change in BIOS is D.O.C.P enabled. AIO pump spins at 90% (3100 RPM) up to 70°C, then hits 100% after that. Case fans run around 850 RPM for 60-65°C, and AIO radiator fans stay at 1000 RPM in the same range. Room temp sits between 26-27°C. Core temps hover 35-40°C – looks normal. Multi-core Cinebench R23 shows cores up to 60-70°C and CPU at 70°C. Single-core tests push individual cores and CPU to 65-70°C, though it’s puzzling why they don’t average. In Cyberpunk2077 with OBS recording in the background using NVENC, one core hit 82°C on the package while CCD1 stayed below that. Similar patterns show up in other games – for instance, during Cyberpunk encoding, CPU usage dropped to ~10% but temps spiked to 90°C. Oddly, the AIO radiator didn’t get hot during those peaks, unlike in multi-core runs. I moved the water pump and checked cables, re-installed OS, even swapped power cables – nothing changed. Pins are intact, no bending, OS fresh. It seems okay for the 5900X, but I’m curious. Are there any red flags? If you notice anything unusual, let me know what else to check!
You might be using an outdated cooling solution on your AIO. If you haven’t already, consider reapplying thermal paste and reinstalling it—this time applying a thin layer across the IHS for better contact. The 5900X tends to reach higher temperatures when running with just one or two cores, which is normal. This happens because lower-core configurations allow higher clock speeds, increasing power consumption and generating more localized heat. Running all cores distributes the workload, allowing heat to dissipate more evenly through the IHS, resulting in a lower reported temperature. Tdie/Tctl reflects the hottest area on the CPU.
Thanks for the feedback! I attempted that again to confirm, but the first time there was a poor connection between AIO and IHS in one corner. The next time the thermal paste formed a solid layer on the AIO, it seemed the issue wasn’t there. I don’t need to worry unless I experience consistently very high temperatures without multi-core load or CPU throttling. @YoungBlade, I’d like to ask more about your point. You said Tdie/Tctl shows the hottest area on the CPU—what about CCD temperatures? Why can the CCD1 temperature be higher than any core inside it, even though it’s based on more than just core data? Spoiler: it’s something else.
Looks fine overall. Just a note—don’t like water cooling much, so I’d suggest opting for a good air cooler. You might get lower temps, which helps with performance. My setup has a small 'ish Noctua, hotter CPU, and max around 70°C, so air cooling still wins. Plus, fans are barely audible unless you push hard.
The temperature pattern you describe is typical, but thermal throttling could be the concern. The Tdie reading should represent the peak temperature, yet with HWInfo64, a single core or CCD might briefly exceed it momentarily. It's worth noting that the average Tdie value tends to be higher than others, ensuring most reports capture the hottest point. I'm puzzled why another sensor might show a higher figure; it likely relates to how HWInfo64 collects data. Tdie is meant to reflect the CPU's hottest location, while HWInfo64's maximum reflects the highest ever recorded, so CCD1 might have briefly reached 82.3°C before dropping, whereas Tdie never reached that level.
I don't notice 'high temps' in this area, sorry. Your readings look typical for their specs. For instance, my 7600—just a modest 65W chip—reaches 78°C during Cinebench R23. That's thanks to the overkill 200 TDP cooler and great airflow case.
I was used to air cooling before, but previously I had less power-hungry CPUs, so I assumed the 5900X required something more efficient. I decided to try AIO Money and will keep using it until it stops working... Thanks for helping ease my concerns! And @YoungBlade for the clear explanations!
Sure, I was just offering an idea from my perspective and experience. I still don’t fully grasp why you’d want to keep it if the performance isn’t up to par. Big IF might be normal, but as mentioned, my 5800x3D runs much cooler on a small and affordable noctua. Maybe lowering the voltage could help if the problem continues.
You're asking about the temperature readings and performance differences between your CPUs. It sounds like you're trying to clarify whether higher temperatures are typical under single-core workloads or if they indicate something unusual, such as a peak in any situation. You also mention comparisons between models like the 5800X3D and 5900X, noting differences in boost clock speeds during single-threaded loads. Your intuition about the 70°C being a high peak is understandable, and you're wondering if that's what contributes to the observed performance gap.