F5F Stay Refreshed Hardware Desktop 2700X identical temperature for single core load, matching all core loads.

2700X identical temperature for single core load, matching all core loads.

2700X identical temperature for single core load, matching all core loads.

X
xAPPLExPIEx
Senior Member
Find
657
05-11-2019, 12:02 AM
#1
2700X equipped with PBO, 150W power budget and a 0.025mV negative voltage offset. Uses an LLC level 2 (or possibly 3) design. Asus ROG STRIX X470-F gaming model, 32GB GSkill Ripjaws Hynnix CJR 3600, running at 3400C18 with a 1.4V thermal right. Includes Kryonaut paste, CPU cold plate is lightly polished, IHS is slightly touched up. All cores operate at 3.85V with single core at 1.2-1.3V and all cores at 1.45V. Recently, I've noticed improved CPU performance after reducing mining power for a year. Previously, I ran it near 190W at 85°C under similar conditions but without power caps, achieving around 1.48V single core and 1.35-1.38V across all cores. Now I cap the CPU at 150W to manage heat and noise, focusing on memory speed. Testing showed that a single-thread Prime95 FFT test runs at about 75°C with roughly 50W power, while all-core FFT peaks near 150W at ~75°C. The temperatures seem consistent regardless of core count. I observed the same readings for single core and all cores instantly, with fans reacting quickly. This isn’t likely due to poor CPU-to-cooler contact since the last time the paste was removed, leaving a uniform layer without gaps. I used a small pea-sized paste blob, pressed the cooler down, repositioned it, and then re-applied fresh paste. The cooler’s flatness is precise enough that when I tried to lift the CPU straight off, it felt like being held by two gauge blocks. Both surfaces were polished to 2000 grit on a tempered glass plate, then lightly buffed with a hand rub for a mirror-like finish. Changing CPU affinity in Task Manager didn’t seem to affect temperatures, which is unusual. I’m puzzled—shouldn’t a single core at 1.45V and 50W be hotter than expected? My memory from over a year ago was unclear, and I wasn’t sure about the voltage specs. It seems normal for Zen+ Ryzen models, but I hadn’t paid much attention to single-core performance before. Swapping coolers would help rule this out, but my board doesn’t have a removable cooler tray, so I had to remove the motherboard for inspection.
Reply
Reply
X
xAPPLExPIEx
05-11-2019, 12:02 AM #1

2700X equipped with PBO, 150W power budget and a 0.025mV negative voltage offset. Uses an LLC level 2 (or possibly 3) design. Asus ROG STRIX X470-F gaming model, 32GB GSkill Ripjaws Hynnix CJR 3600, running at 3400C18 with a 1.4V thermal right. Includes Kryonaut paste, CPU cold plate is lightly polished, IHS is slightly touched up. All cores operate at 3.85V with single core at 1.2-1.3V and all cores at 1.45V. Recently, I've noticed improved CPU performance after reducing mining power for a year. Previously, I ran it near 190W at 85°C under similar conditions but without power caps, achieving around 1.48V single core and 1.35-1.38V across all cores. Now I cap the CPU at 150W to manage heat and noise, focusing on memory speed. Testing showed that a single-thread Prime95 FFT test runs at about 75°C with roughly 50W power, while all-core FFT peaks near 150W at ~75°C. The temperatures seem consistent regardless of core count. I observed the same readings for single core and all cores instantly, with fans reacting quickly. This isn’t likely due to poor CPU-to-cooler contact since the last time the paste was removed, leaving a uniform layer without gaps. I used a small pea-sized paste blob, pressed the cooler down, repositioned it, and then re-applied fresh paste. The cooler’s flatness is precise enough that when I tried to lift the CPU straight off, it felt like being held by two gauge blocks. Both surfaces were polished to 2000 grit on a tempered glass plate, then lightly buffed with a hand rub for a mirror-like finish. Changing CPU affinity in Task Manager didn’t seem to affect temperatures, which is unusual. I’m puzzled—shouldn’t a single core at 1.45V and 50W be hotter than expected? My memory from over a year ago was unclear, and I wasn’t sure about the voltage specs. It seems normal for Zen+ Ryzen models, but I hadn’t paid much attention to single-core performance before. Swapping coolers would help rule this out, but my board doesn’t have a removable cooler tray, so I had to remove the motherboard for inspection.

Reply
Reply