Temperatures are excessively high in PBO-ECO mode and it's restricting the power usage rating!
Temperatures are excessively high in PBO-ECO mode and it's restricting the power usage rating!
You notice temperature spikes when running tasks on one core. Even if you cap the CPU at 65W, the short bursts of demand still cause noticeable heat because a single core only needs that power briefly. The overall system doesn’t reach the maximum reported temperature from HWinfo, which is normal. Your approach will influence how power-hungry multi-threaded applications behave.
Achieving peak performance requires a direct-die modification. Even with TDP constraints or intense fan settings, the 7800x3D can reach 5.05GHz and hit core temperatures near 70°C or higher. The 3D v-cache acts as a significant temperature buffer compared to standard Zen4 CCDs, regardless of whether they’re direct-die or not. It’s worth noting that not everyone needs this upgrade—it comes with high cost and risk. You can still use a Noctua cooler; I’ve used an NH-D15 for my 7950x3D, which is now outdated. I downgraded to a handheld and eGPU setup after experimenting since 2016. Noctua’s AM5 direct-die spacers are useful, but they cost around $200 for all the necessary components. However, installing a Noctua cooler with just two bolts could create uneven pressure on one CCD CPU due to asymmetry. This is speculative; the 7950x3D likely has dual CPUs and won’t face this problem. Still, it’s possible a faulty unit might have a fake CCD already soldered. The direct-die modules that use socket hardware with four bolts are much more stable.
You're being too sensitive, the 7800X3D is built to be powerful with its stacked 3D cache, which raises the chip temperature by about 20°C compared to a non-3D design. That's fine for gaming at 80°C and normal in benchmarks (my 7950X3D handles it too). A TDP of 65W won't really limit performance much—just run games at around 75W stock, so lowering it isn't necessary. However, keeping the temperature below 80°C is important for smooth performance. You might try a slight undervolt adjustment of -10 to -20 ticks using Curve Optimizer.
When Ryzen entered the scene, it really shifted things, especially with the Ryzen 5000 line. Those who aren’t keeping up often stick to the familiar Intel knowledge from before.
You're not interested in keeping your rig idle all the time, are you? Many folks get frustrated when they think a sudden spike in temperature means something is wrong. Ryzen really heats up quickly—around 70°C+ under load and even higher at full power. The readings I saw on my 7950X3D show the 3D chips at 85°C, while the other cores sit around 88-89°C, and another die around 65-70°C. The display says 89°C, but the actual temps are closer to 75°C with a 150W draw. It's just how it works; the chip is performing as expected, not damaged. No need for arbitrary limits that could hurt it.
I think it would help if we could move more smoothly. With my 7800x3d, I’d prefer opening windows, Excel, and browsers to run as a multi-core task at 0.5Ghz and +10c over ambient temperature. I don’t need a sudden spike reaching 85c. I also don’t want my AI to give me extra headroom for everyday tasks. I definitely don’t want the CPU to do anything it explicitly forbids through its BIOS settings. The truck I rented was limited to 65 mph because the company didn’t want it going faster. How challenging is it for a CPU to stay below a certain threshold that I can set? That’s all. Why is that so difficult? I understand it all. I follow the direct die methods Agall uses, I know about PBO, and I understand how Ryzen is constructed. I also want greater control over my system. I just want it to activate only when I tell it to—like limiting FPS in a game. It’s 2024, not 1993. We should be able to manage nearly everything with precise settings. Or is that still a dream?
I don't understand your logic or it seems to fit an Intel perspective that doesn't apply to AMD. The comparison with a car isn't helpful; speed comes from performance, not just raw numbers. You'd want to manage things like oil pressure or brake fluid temperature—just as long as they stay within the intended range. The "heat" of an AMD CPU is mainly about chip-level limits, not just average temperature. In my experience, half the cores run cooler than the other half, so you can adjust settings in BIOS. It's not useful to just lower performance without reason. If it's about saving power, you can set limits as needed. Cutting power too much usually harms performance, especially in games. AMD processors now require less power than Intel for similar results (except for some low-end models), so reducing it isn't necessary.
I'm not following. Are we arguing over arbitrary core temperatures that, if within safe and expected limits, don't matter? I make sure my system is performing as expected in benchmarks, set my cooling to be as quiet as humanly possible that'll keep it that way, then just forget about temperature. If there is a problem, it'll max the fans. Other than that, I simply could not care less what the temperature is.