Memory OC increases CPU power consumption by 70%.
Memory OC increases CPU power consumption by 70%.
My essential system details are listed first: a Ryzen 2700X with an Asus Crosshair Hero VII 2x8GB, G.Skill FlareX 3200, CL14, EVGA GeForce GTX 1070, and Corsair H100iV2. I’m experiencing unusual behavior recently after overclocking my memory. After checking the timing with a DRAM calculator (with the attached Ryzen Timing Checker image), I set Performance Enhancement to level 2 in the BIOS and stopped further changes. Recently, I attempted an all-core overclock for tasks like Rainbow Six Siege or streaming, aiming for at least 4.1Ghz. I loaded into BIOS with optimized defaults and began increasing clocks. At around 4.15Ghz, voltage stabilized at 1.3375v, and temperatures dropped noticeably. During the OCCT burn-in test, power consumption reached about 100-110w. After adjusting the CPU settings, I re-enabled memory overclocking. The results were surprising: under 185w of total package power during a short stress test. This indicates a significant drop in power draw despite higher performance. At 3.85Ghz, temperatures rose by roughly 17°C. When I tried to fine-tune again, the system crashed within two minutes. It seems faster memory and tighter timings boost performance but also increase power consumption. The voltage stayed stable at 1.0125v. This behavior could be related to Ryzen+ and IF settings or possibly an Asus CH7 quirk. Have you encountered this issue before?
The SOC voltage increases when you change the memory, which explains why the higher package draws more power. It’s not a big problem—you can exceed these limits. I’m not sure what’s causing the issue. I skimmed through this a bit, but it seems we were comparing wattage numbers at specific clock speeds without mentioning voltage. There are missing details, and it’s hard to compare screens side by side with inconsistent placement. Could you run your 4.1GHz OC with the memory at lower speeds, then test the 3850MHz version after memory optimization but with both tests using screen shots? Also, could you add CPU-Z and voltage monitoring? I think we can push closer to 200W. The Flagship Zen+ is powerful, and it tends to get hot by design.
Details are still unclear here. The information provided covers all relevant aspects. Voltages were given when not auto-calculated. The SOC voltage remains fixed at 1.0125v with the memory OC enabled. The SOC voltage isn’t automatically adjusted; it stays at 1.0125v unless changed manually. The issue lies in balancing high CPU frequency and memory operation without exceeding safe limits. Specific values for voltages, temperatures, and power consumption are available when requested.
Screenshot of HWiNFO64 during OCCT execution. Medium data size, no AVX support. Memory usage reached 3533CL14 with quick subs and fast subs. Performance Enhancement level 2. More images coming soon—I'll wait for a F@H CPU unit before adjusting settings.
Here are some updated screenshots along with the text labels. BTW, I reset the CPU clock speed back to 4.1Ghz because AVX tasks (like Blender) exceeded that limit. F@H also supports AVX. I included both CPU and SOC power consumption in my HWiNFO setup, which wasn’t visible before. From the results, simply increasing the memory configuration noticeably boosts core power draw. Even when running at optimized defaults without PBO, the extra power isn’t coming from the SOC. I also tested Blender (the BMW demo) to compare performance across the three settings. Optimized defaults – 3:54 CPU default, mem OC – 3:49 CPU 4.1Ghz, mem default – 3:41 CPU PE2 + PBO, mem OC – 3:41 (CPU ran about 3.9Ghz when loaded). I plan to experiment with LLC to see if I can stabilize at 4.1Ghz with the memory OC.