F5F Stay Refreshed Power Users Overclocking High Temps 9700K@5ghz

High Temps 9700K@5ghz

High Temps 9700K@5ghz

Pages (2): Previous 1 2
D
Dun_Schmud
Junior Member
22
01-11-2019, 07:59 PM
#11
my 9700k@5ghz 1.35 model has survived over 10 hours of AIDA64, Cinebench and gaming for about 8 hours, but it still crashes on OCCT within a few minutes. Is this considered a stable OC? Increasing the voltage to 1.37 and 1.38 didn't help much.
D
Dun_Schmud
01-11-2019, 07:59 PM #11

my 9700k@5ghz 1.35 model has survived over 10 hours of AIDA64, Cinebench and gaming for about 8 hours, but it still crashes on OCCT within a few minutes. Is this considered a stable OC? Increasing the voltage to 1.37 and 1.38 didn't help much.

T
Tankboy22
Junior Member
2
01-11-2019, 08:52 PM
#12
If you can run smoothly without issues, I wouldn't have much concern. Even if stability isn't perfect under all loads, a crash seems unlikely to be severe. Ockt might be relying on instructions that aren't typical for your workload. I'm especially concerned about AVX instructions which are very demanding. You could adjust your OC BIOS settings to include an avx offset of 2, which would lower the multiplier by 2 when those instructions appear.
T
Tankboy22
01-11-2019, 08:52 PM #12

If you can run smoothly without issues, I wouldn't have much concern. Even if stability isn't perfect under all loads, a crash seems unlikely to be severe. Ockt might be relying on instructions that aren't typical for your workload. I'm especially concerned about AVX instructions which are very demanding. You could adjust your OC BIOS settings to include an avx offset of 2, which would lower the multiplier by 2 when those instructions appear.

T
The_Zebas
Junior Member
18
01-15-2019, 07:47 PM
#13
Crashes or exit? Occt has a default shutdown at 80°C, so increasing voltages is actually counterproductive.
Occt relies on linpack, which represents a high mix of CPU and RAM usage, much like AVX technologies. This can cause temperatures and CPU activity to exceed typical levels.
For stability, I stick with Asus RealBench, as it tests real-world programs in standard abuse scenarios. I don’t focus on temperatures or stability separately—they’re different metrics.
Regarding temperatures, I run Prime95 with AVX disabled; it’s a full CPU load with minimal RAM usage, offering consistent performance.
Aida64, Occt, and Intel Extreme Burn vary their CPU/RAM demands, making it hard to pinpoint exact temperatures at any moment.
For memory testing, I use memtest86. Surviving 24 hours of error-free stress means it can handle much more.
I haven’t used Cinebench; it was mainly a benchmark tool, though C.20 is becoming a popular alternative to RealBench.
T
The_Zebas
01-15-2019, 07:47 PM #13

Crashes or exit? Occt has a default shutdown at 80°C, so increasing voltages is actually counterproductive.
Occt relies on linpack, which represents a high mix of CPU and RAM usage, much like AVX technologies. This can cause temperatures and CPU activity to exceed typical levels.
For stability, I stick with Asus RealBench, as it tests real-world programs in standard abuse scenarios. I don’t focus on temperatures or stability separately—they’re different metrics.
Regarding temperatures, I run Prime95 with AVX disabled; it’s a full CPU load with minimal RAM usage, offering consistent performance.
Aida64, Occt, and Intel Extreme Burn vary their CPU/RAM demands, making it hard to pinpoint exact temperatures at any moment.
For memory testing, I use memtest86. Surviving 24 hours of error-free stress means it can handle much more.
I haven’t used Cinebench; it was mainly a benchmark tool, though C.20 is becoming a popular alternative to RealBench.

K
KadePlaysMC2
Junior Member
11
01-16-2019, 12:58 AM
#14
I believe you're correct. after adjusting the avx offset from auto to 2, the OCCT stress test completed successfully for 15 minutes without errors or crashes, unlike before when it crashed within 3 minutes prior to the changes. I plan to run another OCCT test in about +4 hours today and hope it passes smoothly. Thanks!
K
KadePlaysMC2
01-16-2019, 12:58 AM #14

I believe you're correct. after adjusting the avx offset from auto to 2, the OCCT stress test completed successfully for 15 minutes without errors or crashes, unlike before when it crashed within 3 minutes prior to the changes. I plan to run another OCCT test in about +4 hours today and hope it passes smoothly. Thanks!

N
NekoKagamine
Member
120
01-16-2019, 01:36 AM
#15
Crashed within 3 minutes when AVX was enabled in auto mode. After switching AVX to 2, I experienced no problems after 15 minutes of stress testing, though I plan to try again in 4 hours. Previous configurations passed AIDA64, realbench, and gaming tests for over 8 hours without any crashes. Regarding the OCCT 80°C default shutdown, I don't believe it exists—I've reached a max temperature of 86°C and the test ran smoothly without interruptions.
N
NekoKagamine
01-16-2019, 01:36 AM #15

Crashed within 3 minutes when AVX was enabled in auto mode. After switching AVX to 2, I experienced no problems after 15 minutes of stress testing, though I plan to try again in 4 hours. Previous configurations passed AIDA64, realbench, and gaming tests for over 8 hours without any crashes. Regarding the OCCT 80°C default shutdown, I don't believe it exists—I've reached a max temperature of 86°C and the test ran smoothly without interruptions.

S
snoffy
Member
119
01-17-2019, 03:12 PM
#16
The temperature at which the test will halt automatically can be adjusted. I assumed the standard setting is 85c.
S
snoffy
01-17-2019, 03:12 PM #16

The temperature at which the test will halt automatically can be adjusted. I assumed the standard setting is 85c.

B
BGStacks
Member
160
01-17-2019, 05:06 PM
#17
it might be around 100c. two of the cores hit 92-93c and the process continued.
B
BGStacks
01-17-2019, 05:06 PM #17

it might be around 100c. two of the cores hit 92-93c and the process continued.

C
ClemCol1
Member
232
01-17-2019, 07:16 PM
#18
During the 5.2ghz 1.38v OCCT test, some cores reached +90c with the highest being 97c. The OCCT stress test ended automatically (I set it to 95c). When I tried 1.375v, the stress froze for 1-2 seconds. This occurred again when I returned to 5.1ghz 1.365v, with slight freezing during the stress test. Now I'm using 5ghz 1.35v, and performance has improved significantly, though millisecond freezes still happen occasionally. Is this typical or could it damage my CPU?
C
ClemCol1
01-17-2019, 07:16 PM #18

During the 5.2ghz 1.38v OCCT test, some cores reached +90c with the highest being 97c. The OCCT stress test ended automatically (I set it to 95c). When I tried 1.375v, the stress froze for 1-2 seconds. This occurred again when I returned to 5.1ghz 1.365v, with slight freezing during the stress test. Now I'm using 5ghz 1.35v, and performance has improved significantly, though millisecond freezes still happen occasionally. Is this typical or could it damage my CPU?

M
MrEpicZocktHD
Junior Member
13
01-17-2019, 08:15 PM
#19
Yes, it is common due to high load, especially on OCCT.
M
MrEpicZocktHD
01-17-2019, 08:15 PM #19

Yes, it is common due to high load, especially on OCCT.

J
JavaCatX
Member
99
01-17-2019, 08:40 PM
#20
Good. 100c is the threshold where the processor will power down. 1.4v corresponds to your maximum output. Set adaptive voltage and speedstep to lower the multiplier and voltage when workload is minimal.
J
JavaCatX
01-17-2019, 08:40 PM #20

Good. 100c is the threshold where the processor will power down. 1.4v corresponds to your maximum output. Set adaptive voltage and speedstep to lower the multiplier and voltage when workload is minimal.

Pages (2): Previous 1 2