F5F Stay Refreshed Hardware Desktop Need assistance with 10850k

Need assistance with 10850k

Need assistance with 10850k

Pages (3): 1 2 3 Next
H
HighSkillsGR
Junior Member
12
09-27-2020, 02:24 PM
#1
After several challenges, I set up a functional system using my 10850k on a RogStrix Z490G gaming Wi-Fi and a 760W PSU from Fractal Ion. When I upgraded to an 850k, I expected a jump to 900k for CEAP and in stock. Now I’m aiming for Core 4.8 at a minus voltage offset of 0.03 and LLC4. During the R23 stress test, my temperatures peaked at 85°C with maximum core heat. I’m new to OC and have reviewed many forum posts about what works. Any suggestions would be appreciated. For testing, I use R23 for 10 minutes, OC for 1 hour, and RealBench for 2 hours. When trying to increase voltage by 49x, I lowered the minus offset to 0.015–0.01, but my readings seem too low or temperatures are spiking to 93–94°C at peak. In Windows, voltage is shown by hwinfo as 1.172–1.234 (sometimes 1.252), and ideally I want a stable x49 or x50 if possible. I’m also experimenting with ASUS AI Auto OC, which handles clock speeds differently—some cores run around 5.2. I think adjusting the Vcore to 1.4 might be safe, but I’m hesitant since my setup feels unstable. Any advice would be really helpful.
H
HighSkillsGR
09-27-2020, 02:24 PM #1

After several challenges, I set up a functional system using my 10850k on a RogStrix Z490G gaming Wi-Fi and a 760W PSU from Fractal Ion. When I upgraded to an 850k, I expected a jump to 900k for CEAP and in stock. Now I’m aiming for Core 4.8 at a minus voltage offset of 0.03 and LLC4. During the R23 stress test, my temperatures peaked at 85°C with maximum core heat. I’m new to OC and have reviewed many forum posts about what works. Any suggestions would be appreciated. For testing, I use R23 for 10 minutes, OC for 1 hour, and RealBench for 2 hours. When trying to increase voltage by 49x, I lowered the minus offset to 0.015–0.01, but my readings seem too low or temperatures are spiking to 93–94°C at peak. In Windows, voltage is shown by hwinfo as 1.172–1.234 (sometimes 1.252), and ideally I want a stable x49 or x50 if possible. I’m also experimenting with ASUS AI Auto OC, which handles clock speeds differently—some cores run around 5.2. I think adjusting the Vcore to 1.4 might be safe, but I’m hesitant since my setup feels unstable. Any advice would be really helpful.

E
Eduardo_GameOn
Posting Freak
921
10-15-2020, 05:53 PM
#2
Getting faster won't come from cutting the v-core. You'll be nearer 1.3v at 5ghz maybe a bit higher. Focus on lowering the temperatures first. You're doing well already! How is your cooling setup looking?
E
Eduardo_GameOn
10-15-2020, 05:53 PM #2

Getting faster won't come from cutting the v-core. You'll be nearer 1.3v at 5ghz maybe a bit higher. Focus on lowering the temperatures first. You're doing well already! How is your cooling setup looking?

P
PisulasRule
Senior Member
676
10-25-2020, 03:27 AM
#3
Yea theres no headroom anymore with these new platforms so id suggest just setting a high static freq like you are doing right now and just running a relatively low volt, diminishing returns zone is at around 5ghz for these cpus so you are good for now Just dont bother ocing the cpu anymore since 5ghz 1.25 is already pretty good, now just move on to overclocking the rams
P
PisulasRule
10-25-2020, 03:27 AM #3

Yea theres no headroom anymore with these new platforms so id suggest just setting a high static freq like you are doing right now and just running a relatively low volt, diminishing returns zone is at around 5ghz for these cpus so you are good for now Just dont bother ocing the cpu anymore since 5ghz 1.25 is already pretty good, now just move on to overclocking the rams

C
connor8c
Member
163
10-26-2020, 03:13 AM
#4
my setup is a meshify 2 compact ui with most options turned off. now i'm running at 4.8ghz, minus voltage offset 0.03, and peak temp reached 85°C during stress test. the max temp was over 90°C—especially with 850k and 4.9ghz at 0.01 voltage. with 90% power, it's fine. thanks for the clarification; this is my first OCP and first PC build ever.
C
connor8c
10-26-2020, 03:13 AM #4

my setup is a meshify 2 compact ui with most options turned off. now i'm running at 4.8ghz, minus voltage offset 0.03, and peak temp reached 85°C during stress test. the max temp was over 90°C—especially with 850k and 4.9ghz at 0.01 voltage. with 90% power, it's fine. thanks for the clarification; this is my first OCP and first PC build ever.

S
Sweet_Dream23
Junior Member
2
10-26-2020, 08:47 PM
#5
It's interesting that someone is attempting to operate a new setup, but newer components seem easier to manage now. The current performance is around 1.25v, which is fine until about 100-105°C. For better longevity and stability, aim for 85°C or below—preferably under 100°C. Running it at 95°C or 105°C is acceptable but not ideal. At higher voltages like 1.4V or above, temperatures should stay at 85°C or less. For maximum safety, consider 1.52V (Intel spec) or the absolute limit of 1.6V, which is around 70°C. If temps exceed that, the system may become unstable.
S
Sweet_Dream23
10-26-2020, 08:47 PM #5

It's interesting that someone is attempting to operate a new setup, but newer components seem easier to manage now. The current performance is around 1.25v, which is fine until about 100-105°C. For better longevity and stability, aim for 85°C or below—preferably under 100°C. Running it at 95°C or 105°C is acceptable but not ideal. At higher voltages like 1.4V or above, temperatures should stay at 85°C or less. For maximum safety, consider 1.52V (Intel spec) or the absolute limit of 1.6V, which is around 70°C. If temps exceed that, the system may become unstable.

M
Mishany174
Member
56
10-27-2020, 01:34 AM
#6
In my experiments with the i9 10900k running at 4.9ghz, all cores performed comparably to those on an i9 9900k or i7 8086k during gaming tasks. When TVB was enabled (below 70c), it outperformed them. A more stable result came from a traditional overclock reaching 5.1ghz, which worked well with a 360mm AIO cooler—no cooling problems there. I don’t use Cinebench much for this kind of testing; at 5.3ghz the CPU stays cool but throttles during tasks like Time Spy. It’s useful for scoring comparisons, but that’s about it.
M
Mishany174
10-27-2020, 01:34 AM #6

In my experiments with the i9 10900k running at 4.9ghz, all cores performed comparably to those on an i9 9900k or i7 8086k during gaming tasks. When TVB was enabled (below 70c), it outperformed them. A more stable result came from a traditional overclock reaching 5.1ghz, which worked well with a 360mm AIO cooler—no cooling problems there. I don’t use Cinebench much for this kind of testing; at 5.3ghz the CPU stays cool but throttles during tasks like Time Spy. It’s useful for scoring comparisons, but that’s about it.

C
Colamile
Junior Member
43
10-27-2020, 03:48 AM
#7
It depends on the context. A score over 90°C during stress testing at less than 1.4V is generally acceptable, but verify with your specific device. The Vcore in HWInfo isn't always the exact value—consider adding a small margin like 0.1 if unsure. If HWInfo shows 1.12, you might expect around 1.22 under similar conditions.
C
Colamile
10-27-2020, 03:48 AM #7

It depends on the context. A score over 90°C during stress testing at less than 1.4V is generally acceptable, but verify with your specific device. The Vcore in HWInfo isn't always the exact value—consider adding a small margin like 0.1 if unsure. If HWInfo shows 1.12, you might expect around 1.22 under similar conditions.

A
Aerithix
Member
182
10-31-2020, 04:41 AM
#8
I’d run Real Bench to simulate real-world stress alongside your actual applications. Cinebench handles all-core CPU work without any GPU heat buildup. With a more powerful modern GPU, under load it will cause the CPU to warm slightly more. Realbench challenges both parts with genuine tasks, offering a more accurate picture of your needs. CB R23 isn’t a true stress test—it just speeds up finding obvious issues before pushing harder. Regarding voltages: I achieved 5.1GHz all-core on my 10900KF at 1.32V without AVX, but needed nearly 1.4V for stable 5.3GHz in lighter loads. Under heavy load it dropped below 1.3V and 5.1GHz. With a 10850k it might be even worse, possibly 100mhz lower at the same settings. Results can differ each time. The Fuma 2 is a good cooler but seems better suited for baseline use on a high-core processor rather than intense stress testing.
A
Aerithix
10-31-2020, 04:41 AM #8

I’d run Real Bench to simulate real-world stress alongside your actual applications. Cinebench handles all-core CPU work without any GPU heat buildup. With a more powerful modern GPU, under load it will cause the CPU to warm slightly more. Realbench challenges both parts with genuine tasks, offering a more accurate picture of your needs. CB R23 isn’t a true stress test—it just speeds up finding obvious issues before pushing harder. Regarding voltages: I achieved 5.1GHz all-core on my 10900KF at 1.32V without AVX, but needed nearly 1.4V for stable 5.3GHz in lighter loads. Under heavy load it dropped below 1.3V and 5.1GHz. With a 10850k it might be even worse, possibly 100mhz lower at the same settings. Results can differ each time. The Fuma 2 is a good cooler but seems better suited for baseline use on a high-core processor rather than intense stress testing.

1
10th_Doctor_
Posting Freak
768
11-09-2020, 12:15 AM
#9
At stock levels you receive 85C, which feels a bit steep. Your 10850k hits around 75C under stress tests. The Fuma 2 performs slightly worse than the NH-D15 but not significantly. I’d look into this first.
1
10th_Doctor_
11-09-2020, 12:15 AM #9

At stock levels you receive 85C, which feels a bit steep. Your 10850k hits around 75C under stress tests. The Fuma 2 performs slightly worse than the NH-D15 but not significantly. I’d look into this first.

H
Hols8888
Member
140
11-13-2020, 07:09 AM
#10
Absolutely, your stress test, Cinebench, and Realbench reached their limits at 85°C with every core capped at 4.8GHz. The voltage drop was noted right away.
H
Hols8888
11-13-2020, 07:09 AM #10

Absolutely, your stress test, Cinebench, and Realbench reached their limits at 85°C with every core capped at 4.8GHz. The voltage drop was noted right away.

Pages (3): 1 2 3 Next