F5F Stay Refreshed Power Users Overclocking Need assistance in running DDR4 at 3000 MHz

Need assistance in running DDR4 at 3000 MHz

Need assistance in running DDR4 at 3000 MHz

I
IIGuiGalouLT
Member
193
12-26-2017, 11:33 AM
#1
I've been overclocking for a long time and it seemed stable at first, but recently I started encountering BSOD Page Faults in some games. I use Prime95 and found that rounding errors were the issue, so I reset my BIOS to its default settings and began from scratch. Here are my specifications:

ASUS Strix 270F Gaming, latest BIOS, Intel i5-7600k Kaby Lake-S Corsair Vengeance LED 2x8Gb DDR4 3000 MHz, 15-17-17-35

I successfully overclocked the CPU to 4600 with these settings:
46 multiplier, BCK 100 MHz Core Voltage: adaptive, 1.20V turbo, offset auto AC/DC load line: 0.01 LLC 4. Everything else set to default.

All tests in Prime95 were stable, and memtest86 confirmed no issues. Even when running games, which previously caused BSODs, everything is now fine.

Now I wanted to upgrade my RAM from 2133 to 3000 MHz. I enabled XMP, and it started BSOD on Windows startup immediately. I manually adjusted timings and voltage, but the problem persisted. I tried increasing VCore and VCCSA gradually from 1V to 1.3V, yet it still crashed at startup. I’m unsure how I used XMP before, as it was stable for three years, but now even with a 4600 MHz CPU and higher voltage settings, it fails to boot Windows without BSOD. The difference between my old configuration (before 3 years) and the current one is significant—previously I ran at 4500 MHz in offset mode, possibly above 1.2V, while now I’m unsure of the exact values. AC/DC was set to auto.

Once again, XMP before resetting the BIOS was quite reliable (only one game caused BSOD), but now I can’t replicate that setup. I need assistance to run my RAM at 3000 MHz safely. I’m uncertain what steps to take next. I’m hesitant to push VCore voltage higher than 1.3, as it might be linked to the previous offset mode settings.
I
IIGuiGalouLT
12-26-2017, 11:33 AM #1

I've been overclocking for a long time and it seemed stable at first, but recently I started encountering BSOD Page Faults in some games. I use Prime95 and found that rounding errors were the issue, so I reset my BIOS to its default settings and began from scratch. Here are my specifications:

ASUS Strix 270F Gaming, latest BIOS, Intel i5-7600k Kaby Lake-S Corsair Vengeance LED 2x8Gb DDR4 3000 MHz, 15-17-17-35

I successfully overclocked the CPU to 4600 with these settings:
46 multiplier, BCK 100 MHz Core Voltage: adaptive, 1.20V turbo, offset auto AC/DC load line: 0.01 LLC 4. Everything else set to default.

All tests in Prime95 were stable, and memtest86 confirmed no issues. Even when running games, which previously caused BSODs, everything is now fine.

Now I wanted to upgrade my RAM from 2133 to 3000 MHz. I enabled XMP, and it started BSOD on Windows startup immediately. I manually adjusted timings and voltage, but the problem persisted. I tried increasing VCore and VCCSA gradually from 1V to 1.3V, yet it still crashed at startup. I’m unsure how I used XMP before, as it was stable for three years, but now even with a 4600 MHz CPU and higher voltage settings, it fails to boot Windows without BSOD. The difference between my old configuration (before 3 years) and the current one is significant—previously I ran at 4500 MHz in offset mode, possibly above 1.2V, while now I’m unsure of the exact values. AC/DC was set to auto.

Once again, XMP before resetting the BIOS was quite reliable (only one game caused BSOD), but now I can’t replicate that setup. I need assistance to run my RAM at 3000 MHz safely. I’m uncertain what steps to take next. I’m hesitant to push VCore voltage higher than 1.3, as it might be linked to the previous offset mode settings.

R
Reuben135
Member
89
12-26-2017, 06:24 PM
#2
Yes, the Vdimm is set to operate at 1.35V.
R
Reuben135
12-26-2017, 06:24 PM #2

Yes, the Vdimm is set to operate at 1.35V.

E
Envil_
Member
126
12-30-2017, 03:34 AM
#3
When attempting to operate at 3000mhz, I adjusted the voltage in BIOS to 1.35. Even a slight increase caused the system to report 1.325v instead. Currently, it runs at 2133Mhz with 1.2v.
E
Envil_
12-30-2017, 03:34 AM #3

When attempting to operate at 3000mhz, I adjusted the voltage in BIOS to 1.35. Even a slight increase caused the system to report 1.325v instead. Currently, it runs at 2133Mhz with 1.2v.

P
petiteaxelle
Member
57
01-15-2018, 04:33 AM
#4
They already passed memtest at 3000MHz with specific parameters, indicating good memory performance. However, the same memory appears to struggle on the board QVL, especially after being overclocked. I suggest adjusting the timings slightly looser, such as 16-18-18-38, to observe any improvements in BSOD frequency before gradually lowering the standard frequencies. It’s unlikely an RMA will be necessary if they pass memtest, but the board might still be the problem.
P
petiteaxelle
01-15-2018, 04:33 AM #4

They already passed memtest at 3000MHz with specific parameters, indicating good memory performance. However, the same memory appears to struggle on the board QVL, especially after being overclocked. I suggest adjusting the timings slightly looser, such as 16-18-18-38, to observe any improvements in BSOD frequency before gradually lowering the standard frequencies. It’s unlikely an RMA will be necessary if they pass memtest, but the board might still be the problem.

M
moffy
Member
59
01-15-2018, 07:54 AM
#5
Using these settings and voltage, I successfully got at least Windows to boot without BSOD. I can even execute Prime95, though it struggles in blend mode. Still, no BSOD is a good step forward. I'm curious about how I managed to run the XMP profile three years ago—it was quite reliable, around 95% of the time. I plan to experiment with frequency now.
M
moffy
01-15-2018, 07:54 AM #5

Using these settings and voltage, I successfully got at least Windows to boot without BSOD. I can even execute Prime95, though it struggles in blend mode. Still, no BSOD is a good step forward. I'm curious about how I managed to run the XMP profile three years ago—it was quite reliable, around 95% of the time. I plan to experiment with frequency now.

M
Moozizzy333
Junior Member
11
01-18-2018, 02:41 AM
#6
Update: i removed second stick of ram and left only one stick in A2 slot and enabled XMP with overclocked CPU and it seems to work, it passes Prime95 now and no bsod. I will run some more tests and then try to install back second stick of ram. Does it mean then when i am installing 2 stick of ram motherboard can't manage two of them with XMP enabled?
M
Moozizzy333
01-18-2018, 02:41 AM #6

Update: i removed second stick of ram and left only one stick in A2 slot and enabled XMP with overclocked CPU and it seems to work, it passes Prime95 now and no bsod. I will run some more tests and then try to install back second stick of ram. Does it mean then when i am installing 2 stick of ram motherboard can't manage two of them with XMP enabled?

M
matluffy84
Junior Member
14
01-22-2018, 10:14 AM
#7
That's what it sounds like.
M
matluffy84
01-22-2018, 10:14 AM #7

That's what it sounds like.

N
nugrr
Member
64
01-22-2018, 10:57 AM
#8
The issue seemed related to adaptive mode. Initially, I tried gradually raising the voltage for vccio and vccsa from 1.0v to 1.25v while testing in memtest86, but without success. I eventually managed a few successful passes at 1.12v for both components in memtest, though 3 attempts always failed. Then I considered adaptive mode, setting voltage only in turbo mode, assuming memory controllers would require more power in both modes. To simplify, I switched the CPU vcore voltage to manual and added some test voltages, setting it to 1.23v for vcore. This time, I incrementally increased the voltage again and achieved good results: at 1.11v for both vccio and vccsa, memory functioned properly. I completed 8 passes in memtest86 with no errors, followed by Windows installation and HCI memtest at 600% load, all without issues. Then I ran prime95 in blend mode, which also remained stable. Now I plan to lower the vcore voltage in manual mode and test switching back to adaptive mode if helpful. Anyone with tips on changing manual to adaptive would be appreciated. In the end, enabling XMP isn't straightforward—selecting it in BIOS is better.)
N
nugrr
01-22-2018, 10:57 AM #8

The issue seemed related to adaptive mode. Initially, I tried gradually raising the voltage for vccio and vccsa from 1.0v to 1.25v while testing in memtest86, but without success. I eventually managed a few successful passes at 1.12v for both components in memtest, though 3 attempts always failed. Then I considered adaptive mode, setting voltage only in turbo mode, assuming memory controllers would require more power in both modes. To simplify, I switched the CPU vcore voltage to manual and added some test voltages, setting it to 1.23v for vcore. This time, I incrementally increased the voltage again and achieved good results: at 1.11v for both vccio and vccsa, memory functioned properly. I completed 8 passes in memtest86 with no errors, followed by Windows installation and HCI memtest at 600% load, all without issues. Then I ran prime95 in blend mode, which also remained stable. Now I plan to lower the vcore voltage in manual mode and test switching back to adaptive mode if helpful. Anyone with tips on changing manual to adaptive would be appreciated. In the end, enabling XMP isn't straightforward—selecting it in BIOS is better.)

W
WhaIe
Junior Member
4
01-26-2018, 01:31 PM
#9
I successfully changed to adaptive mode: 1.2v in turbo, 0.01v offset. Now DDR4 XMP is active, CPU OC is on, and adaptive mode is working perfectly.
W
WhaIe
01-26-2018, 01:31 PM #9

I successfully changed to adaptive mode: 1.2v in turbo, 0.01v offset. Now DDR4 XMP is active, CPU OC is on, and adaptive mode is working perfectly.