F5F Stay Refreshed Hardware Desktop Issues with i9-7980XE OC processors

Issues with i9-7980XE OC processors

Issues with i9-7980XE OC processors

S
SrWaldo_22
Member
239
08-10-2021, 09:59 AM
#1
Hi, I received a used i9-7980XE and attempted an overclock. I adjusted turbo limits for various core counts—1 through 18. To check stability, you should run it continuously for several days. The issue you're facing with crashes and BSODs after just two days at low load suggests instability under real-world conditions. The high current draw from undervolting (-0.02V) likely caused thermal throttling or voltage regulation problems. The PSU supports 32A, which should handle the load if properly cooled, but you might want to verify the actual draw and ensure adequate airflow. Consider monitoring temperatures closely and adjusting voltages gradually to find a safe balance.
S
SrWaldo_22
08-10-2021, 09:59 AM #1

Hi, I received a used i9-7980XE and attempted an overclock. I adjusted turbo limits for various core counts—1 through 18. To check stability, you should run it continuously for several days. The issue you're facing with crashes and BSODs after just two days at low load suggests instability under real-world conditions. The high current draw from undervolting (-0.02V) likely caused thermal throttling or voltage regulation problems. The PSU supports 32A, which should handle the load if properly cooled, but you might want to verify the actual draw and ensure adequate airflow. Consider monitoring temperatures closely and adjusting voltages gradually to find a safe balance.

S
skittles424
Member
61
08-26-2021, 09:17 PM
#2
I believe OCAP is often tweakable or even turned off in the BIOS, but you might rely on elmor evc2 if the board has a poor OCAP cap that’s significantly too low and there’s no way to disable it through the BIOS. Since an 18-core processor acts like a nuclear reactor, solid cooling is essential to manage temperatures. Have you already looked into this? Also, the stock temperature data seems unreliable, and you could potentially achieve up to 20°C lower temps using liquid metal.
S
skittles424
08-26-2021, 09:17 PM #2

I believe OCAP is often tweakable or even turned off in the BIOS, but you might rely on elmor evc2 if the board has a poor OCAP cap that’s significantly too low and there’s no way to disable it through the BIOS. Since an 18-core processor acts like a nuclear reactor, solid cooling is essential to manage temperatures. Have you already looked into this? Also, the stock temperature data seems unreliable, and you could potentially achieve up to 20°C lower temps using liquid metal.

T
taliaxox
Junior Member
11
09-04-2021, 06:48 PM
#3
This refers to the K299 motherboard. You should verify each core separately, running them at lower cooling speeds to simulate full load. Avoid tracking the number of cores under stress—managing 18 cores at the same voltage isn't practical. Lowering the voltage affects stability at higher speeds; a reading below -0.02V indicates issues above 3.9GHz. The default speed is around 3.4GHz across all cores.
T
taliaxox
09-04-2021, 06:48 PM #3

This refers to the K299 motherboard. You should verify each core separately, running them at lower cooling speeds to simulate full load. Avoid tracking the number of cores under stress—managing 18 cores at the same voltage isn't practical. Lowering the voltage affects stability at higher speeds; a reading below -0.02V indicates issues above 3.9GHz. The default speed is around 3.4GHz across all cores.

T
Tomay0
Junior Member
14
09-06-2021, 10:31 AM
#4
The system runs on a Gigabyte UD4 motherboard. I didn’t include the OCP details, thinking it came from the PSU instead. I’m uncertain about the exact procedure and have no confidence in testing it safely. Disabling the CPU in the PSU seems risky. It was purchased used from eBay with a Noctua NH-D14 cooler. Cooling works adequately. I experienced thermal throttling in Cinebench R23, but performance remains significantly below stock speeds. Running at around 100% load for extended periods isn’t feasible; I’m just running short bursts (~1 minute). The only workaround is capping the CPU to 3.9 GHz regardless of workload. That’s below the stock 4.4 GHz (2 cores) spec, though. I’ve reduced turbo frequency by 0.1 GHz across 3-12 cores to observe results. If that doesn’t help, I’ll test single cores or specific core combinations. At 4.4 GHz it seems stable for short bursts, but every core occasionally hits the limit. Adjusting the offset to -0.01 V could help, though it risks triggering the OCP. The difference between stock and -0.02 V is minimal—just a few watts away from the threshold.
T
Tomay0
09-06-2021, 10:31 AM #4

The system runs on a Gigabyte UD4 motherboard. I didn’t include the OCP details, thinking it came from the PSU instead. I’m uncertain about the exact procedure and have no confidence in testing it safely. Disabling the CPU in the PSU seems risky. It was purchased used from eBay with a Noctua NH-D14 cooler. Cooling works adequately. I experienced thermal throttling in Cinebench R23, but performance remains significantly below stock speeds. Running at around 100% load for extended periods isn’t feasible; I’m just running short bursts (~1 minute). The only workaround is capping the CPU to 3.9 GHz regardless of workload. That’s below the stock 4.4 GHz (2 cores) spec, though. I’ve reduced turbo frequency by 0.1 GHz across 3-12 cores to observe results. If that doesn’t help, I’ll test single cores or specific core combinations. At 4.4 GHz it seems stable for short bursts, but every core occasionally hits the limit. Adjusting the offset to -0.01 V could help, though it risks triggering the OCP. The difference between stock and -0.02 V is minimal—just a few watts away from the threshold.

W
willie_killer
Member
187
09-06-2021, 05:15 PM
#5
It might stem from a VCCIN issue (CPU input voltage), which by default sets to 1.8V but I’d maintain at 2V for better VRM cooling and CPU protection during overclocking.
W
willie_killer
09-06-2021, 05:15 PM #5

It might stem from a VCCIN issue (CPU input voltage), which by default sets to 1.8V but I’d maintain at 2V for better VRM cooling and CPU protection during overclocking.