F5F Stay Refreshed Power Users Overclocking Not possible to boost memory and CPU simultaneously

Not possible to boost memory and CPU simultaneously

Not possible to boost memory and CPU simultaneously

W
WildCandy
Senior Member
675
02-25-2016, 05:46 PM
#1
I’m a former Intel MB design engineer who’s overclocked multiple boards, so I know how to ask these questions. Here’s what I’ve found:
- ASUS Rampage VI Extreme Encore
- i9-10940x CPU (14 cores) with two different CPUs
- G.SKILL TridentZ RGB DDR4 4266 (F4-4266C17Q-32GTZR 17-18-18-36)
- CORSAIR AXi Series AX1600i PS
- Cooling: custom water loop
Both CPUs maintain stable performance at 5.0GHz, but temperatures reach a peak of 90°C during Cinebench 20, making 5GHz unsuitable for continuous use. To keep temperatures under 70°C max, I run them at 101.3 * 47 = 4.76GHz across all cores using adaptive mode with an offset of -0.012V and AVX offsets set to 3. I’ve also tried static mode at this frequency. Under heavy load, both CPUs operate at 1.2V, with adaptive mode maintaining stability. At default clock speeds, I can run Cinebench all day at 3800MHz (15-15-15-36) without adjusting voltages.
The issue arises when overclocked to the above settings: enabling XMP at 3800MHz triggers an error in Cinebench within a few minutes. However, I can boot and operate for 10–20 minutes at 3000MHz using the kit’s default 17-18-18-38. This suggests the problem isn’t related to clock speed or latency. Adjusting VCCSA and VCCIO values doesn’t help stability, nor does raising Core voltage.
I’ve re-flashed the BIOS (latest 0603) and cleared CMOS on both CPUs without any effect.
W
WildCandy
02-25-2016, 05:46 PM #1

I’m a former Intel MB design engineer who’s overclocked multiple boards, so I know how to ask these questions. Here’s what I’ve found:
- ASUS Rampage VI Extreme Encore
- i9-10940x CPU (14 cores) with two different CPUs
- G.SKILL TridentZ RGB DDR4 4266 (F4-4266C17Q-32GTZR 17-18-18-36)
- CORSAIR AXi Series AX1600i PS
- Cooling: custom water loop
Both CPUs maintain stable performance at 5.0GHz, but temperatures reach a peak of 90°C during Cinebench 20, making 5GHz unsuitable for continuous use. To keep temperatures under 70°C max, I run them at 101.3 * 47 = 4.76GHz across all cores using adaptive mode with an offset of -0.012V and AVX offsets set to 3. I’ve also tried static mode at this frequency. Under heavy load, both CPUs operate at 1.2V, with adaptive mode maintaining stability. At default clock speeds, I can run Cinebench all day at 3800MHz (15-15-15-36) without adjusting voltages.
The issue arises when overclocked to the above settings: enabling XMP at 3800MHz triggers an error in Cinebench within a few minutes. However, I can boot and operate for 10–20 minutes at 3000MHz using the kit’s default 17-18-18-38. This suggests the problem isn’t related to clock speed or latency. Adjusting VCCSA and VCCIO values doesn’t help stability, nor does raising Core voltage.
I’ve re-flashed the BIOS (latest 0603) and cleared CMOS on both CPUs without any effect.

H
Heyello
Member
196
02-25-2016, 08:03 PM
#2
The error Cinebench is reporting is unclear, but it seems related to power or performance concerns. Possible issues include insufficient power supply capacity or an outdated component nearing end of life. It's recommended to verify the PSU specifications and consider additional wattage availability.
H
Heyello
02-25-2016, 08:03 PM #2

The error Cinebench is reporting is unclear, but it seems related to power or performance concerns. Possible issues include insufficient power supply capacity or an outdated component nearing end of life. It's recommended to verify the PSU specifications and consider additional wattage availability.

E
EpicExplosion
Member
129
02-28-2016, 08:53 PM
#3
PSU appears fine, CPU and RAM seem normal – maybe the motherboard has issues delivering the required current if the CPU and RAM are overclocked? VRMs and other parts that supply voltage and current on the motherboard? Could be worth trying a different PSU. If that doesn’t work, it might be a problem with the motherboard itself, even though your board is high-quality. Are all extra power connectors installed correctly (two 8-pin and one 6-pin)?
E
EpicExplosion
02-28-2016, 08:53 PM #3

PSU appears fine, CPU and RAM seem normal – maybe the motherboard has issues delivering the required current if the CPU and RAM are overclocked? VRMs and other parts that supply voltage and current on the motherboard? Could be worth trying a different PSU. If that doesn’t work, it might be a problem with the motherboard itself, even though your board is high-quality. Are all extra power connectors installed correctly (two 8-pin and one 6-pin)?

V
vinikind08
Member
74
03-15-2016, 05:46 AM
#4
Appreciate your response. I don't remember the exact message, but it mentioned an issue with the system and pointed out where the problem might be. The supply is around 6 weeks old and seems excessive for my current setup. The Corsair iCue lets me monitor voltage, power, current, etc., but not very precisely. I can't completely eliminate the chance of short-term fluctuations, so I'll recheck the voltages with an o-scope.

I just assembled the PC, which also uses only SSDs: two Samsung 970 EVO Plus 1T drives in an iRST RAID 0 setup, plus a 360GB Intel Optane 905P. The graphics card is an MSI Vega 56 – not very power-intensive.

The supply is drawing 643W at its peak from the wall.
Thanks for the advice. With a fresh supply I hadn't thought about checking voltage stability.
V
vinikind08
03-15-2016, 05:46 AM #4

Appreciate your response. I don't remember the exact message, but it mentioned an issue with the system and pointed out where the problem might be. The supply is around 6 weeks old and seems excessive for my current setup. The Corsair iCue lets me monitor voltage, power, current, etc., but not very precisely. I can't completely eliminate the chance of short-term fluctuations, so I'll recheck the voltages with an o-scope.

I just assembled the PC, which also uses only SSDs: two Samsung 970 EVO Plus 1T drives in an iRST RAID 0 setup, plus a 360GB Intel Optane 905P. The graphics card is an MSI Vega 56 – not very power-intensive.

The supply is drawing 643W at its peak from the wall.
Thanks for the advice. With a fresh supply I hadn't thought about checking voltage stability.

B
banshee45
Senior Member
726
03-16-2016, 07:12 AM
#5
I covered the PS in my previous post, but you pointed out another possibility I will address.
I've connected only the 24- and 8-pin headers, which is one manual option. Adding the 6-pin is a no-brainer I hadn't considered, however. Obviously something's not right, so this is a good idea. There is also a 4-pin Molex at the bottom of the board ASUS provides for dual GPU setups, but it certainly can't hurt to connect this header too.
B
banshee45
03-16-2016, 07:12 AM #5

I covered the PS in my previous post, but you pointed out another possibility I will address.
I've connected only the 24- and 8-pin headers, which is one manual option. Adding the 6-pin is a no-brainer I hadn't considered, however. Obviously something's not right, so this is a good idea. There is also a 4-pin Molex at the bottom of the board ASUS provides for dual GPU setups, but it certainly can't hurt to connect this header too.

K
ketman34
Posting Freak
834
03-16-2016, 08:33 AM
#6
It appears the Molex only supply additional current for the PCIe bus. I wonder if incorporating the 6-pin or even the second 8-pin connection could address the issue.
K
ketman34
03-16-2016, 08:33 AM #6

It appears the Molex only supply additional current for the PCIe bus. I wonder if incorporating the 6-pin or even the second 8-pin connection could address the issue.

B
brosukeMakara
Junior Member
9
03-16-2016, 11:18 AM
#7
I began with the 24-pin and 8-pin configurations, then moved to the 6-pin version without any noticeable changes.

I need to note that my overclocking might cause the CPU to consume around 500W and approach that level, although the tests I'm doing now are not close to it.

I've switched the CPU and am currently running tests with it. It seems to perform better, but I haven't confirmed that the issue is fully resolved.
B
brosukeMakara
03-16-2016, 11:18 AM #7

I began with the 24-pin and 8-pin configurations, then moved to the 6-pin version without any noticeable changes.

I need to note that my overclocking might cause the CPU to consume around 500W and approach that level, although the tests I'm doing now are not close to it.

I've switched the CPU and am currently running tests with it. It seems to perform better, but I haven't confirmed that the issue is fully resolved.

D
DeathDark38
Member
211
03-17-2016, 06:33 PM
#8
Thx for info and good luck.
D
DeathDark38
03-17-2016, 06:33 PM #8

Thx for info and good luck.