cancel
Showing results for 
Search instead for 
Did you mean: 

Corsair DDR4 XMP BSOD

Xopek
Level 7
Long time ago i have overclock and it was pretty stable (at least i thought so), but in some game i started to get BSOD Page Fault.

I run Prime95 in blend mode and realized i have rounding errors, so i reset my bios to default and started from scratch. My specs:

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

So i overclocked CPU to 4600 successfully with this setting:

46 multiplier
BCK 100 MHz
Core Voltage: adaptive, 1.20v turbo, offset auto
AC/DC load line: 0.01
LLC 4
DRAM Voltage 1.2V
VCCIO 1V
System agent voltage 1.024V
Everything else on default

I ran different tests in Prime 95 all stable, then memtest86 all good. Even run game, that previously bsod, it is fine now too.

Now i wanted to run my memory from 2133 to 3000

I enabled XMP and it immediately bsod on windows startup.

I set timings manually and voltage too. Still Bsod.

Then i tried to incrementally increase vccio and system agent voltage from 1 v up to 1.3v !!! Still crashing on startup.
I don’t even know how i was running XMP before, because it was quite stable, after 3 years only the first game encounter bsod. Now i can’t even manage to startup Windows without bsod with xmp as before. The difference between old setting (3 years ago) and current is before i was running CPU at 4500 (4600 now) and voltage was in offset mode, and maybe higher than 1.2v i didn’t remember. Also ac/dc was auto. But with even this setting i can not run xmp now.

Then i tried XMP with one stick of memory and it runs fine, memtest and Prime95 passed , with vccio 1v and system agent voltage 1.024v.

Then i insert 2second stick and again immediate bsod on Windows startup.

So need some help guys. My current queastions:

1) Should i try to increase vccio and system agent even more?
2) Should i try to increase only DRAM voltage?
3) Is it normal, that when i set 1.35V in DRAM voltage, bios shows me only 1.325V? Same with VCCIO AND System agent Voltage
4) Should vccio and system agent voltage be the same? or should i rise them separately?
5) Should i try increase Vcore?
6) What else i can do? Because i know that my system can run this stick in xmp mode, atleast it was like 98% stable, now it is like 5% stable)

Thanks for help beforehand!
3,796 Views
9 REPLIES 9

Silent_Scone
Super Moderator
Hello,

1) VCCIO and VCCSA are signal rails, which means too much voltage can be just as detrimental as too little. You will need to find a value that works well for your CPU sample and memory kit.

2) This shouldn't be necessary if the desired XMP profile voltage is input. However, depending on the amount of guard band put in place by the memory vendor, it may be beneficial to try adding voltage (10-30mv).

3) Those readings are normal as the voltage is tapped off the power plane, so will not always be matched to what is being fed at the DIMM.

4) Each individual CPU is different, as a rule of thumb these values should be within 100mv of one another. VCCSA has the larger impact on stability with memory overclocking.

5) Depending on the course of the instability, you may require more vcore to accommodate for the increased memory frequency due to the increase in I/O.
13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Silent Scone@ASUS wrote:
Hello,

1) VCCIO and VCCSA are signal rails, which means too much voltage can be just as detrimental as too little. You will need to find a value that works well for your CPU sample and memory kit.

2) This shouldn't be necessary if the desired XMP profile voltage is input. However, depending on the amount of guard band put in place by the memory vendor, it may be beneficial to try adding voltage (10-30mv).

3) Those readings are normal as the voltage is tapped off the power plane, so will not always be matched to what is being fed at the DIMM.

4) Each individual CPU is different, as a rule of thumb these values should be within 100mv of one another. VCCSA has the larger impact on stability with memory overclocking.

5) Depending on the course of the instability, you may require more vcore to accommodate for the increased memory frequency due to the increase in I/O.


Thank you very much for your answers!

Can you please elaborate more a lit bit about this " Those readings are normal as the voltage is tapped off the power plane, so will not always be matched to what is being fed at the DIMM." So if i set 1.35V in bios and bios shows me only 1.3250, should i increase voltage till i see that the actual reading is 1.350v in bios? So with VCCSA and VCCIO, i set them the same voltage, but actual reading a very different for them. So should i try to keep them equal in what i set in bios or should i try to keep them equal in actual readings?

Xopek wrote:
Thank you very much for your answers!

Can you please elaborate more a lit bit about this " Those readings are normal as the voltage is tapped off the power plane, so will not always be matched to what is being fed at the DIMM." So if i set 1.35V in bios and bios shows me only 1.3250, should i increase voltage till i see that the actual reading is 1.350v in bios? So with VCCSA and VCCIO, i set them the same voltage, but actual reading a very different for them. So should i try to keep them equal in what i set in bios or should i try to keep them equal in actual readings?



Hello,

As said previously, those readings are normal. The voltages seen in UEFI will not always match what is being received at the socket, thus there's no need to be concerned by variance. If wanting 1.35v, then input 1.35 in the UEFI.

You are better not concerning yourself with those things and focus on dialing in a voltage that works best for your stability and needs in practice.

For VCCIO & VCCSA try within a range of 1.05 and 1.25v methodically using a tried and tested memory stressor such as HCI Memtest or karhu RAM Test.
13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Update:

So the problem was with adaptive mode i think. First i tried to incrementally increase voltage for vccio and vccsa from 1.0v up to 1.25v and testing in memtest86. But with no luck. I managed to get 1 and 2 successful pass on 1.12v for vccio and vccsa in memtest, but 3 pass always failed.

Then i thought about adaptive mode, i set voltage only in turbo mode, so if memory controllers puts more work on cpu, maybe it needs more voltage in both turbo and normal mode. So to get things easy i switch cpu vcore voltage to manual mode and add some voltages, so i set 1.23v in manual mode for vcore. Then started again incrementally increasing voltage for vccio and vccsa. This time i have fortune on my side. On 1.11v for both vccio and vccsa my memory worked. I run 8 passes in memtest86 with no errors. Then loaded windows and run HCI memtest for 600% also with no errors. And finally run prime95 in blend mode. Also stable.

So now i will try to decrease voltage for vcore in manual mode. And then will try to switch manual to adaptive mode again.

If anyone knows some tips for switching manual to adpative in my situation will appreciate it.

So in the end enabling XMP not so easy as select it in bios 😃

Silent_Scone
Super Moderator
Hello,

If using Adaptive Vcore, it's common that more voltage is needed over a fixed/manual voltage due to the change in load transitioning. It is also preferable to increase Vcore if experiencing instability than increasing LLC.

Also, be aware that it is not possible to set a lower Aditional Turbo Voltage than the stock VID for a given ratio.
13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Silent Scone@ASUS wrote:
Hello,

If using Adaptive Vcore, it's common that more voltage is needed over a fixed/manual voltage due to the change in load transitioning. It is also preferable to increase Vcore if experiencing instability than increasing LLC.

Also, be aware that it is not possible to set a lower Aditional Turbo Voltage than the stock VID for a given ratio.


Thanks very much for your help! So i managed to switch to adaptive mode: 1.2v in turbo, 0.01v offset.

So now i have DDR4 xmp enabled, CPU OC and adaptive mode, perfect=)

EXILE157
Level 7
Can anyone here help with my situation?.. I've no luck trying to find a post about the (ROG Strix Z490-G)

I've had nothing but issues from the start with this board and manage to make it work after hours upon hours for 2 days!...

Now here's my situation : For some odd reason DIMM Slot 1 is not allowing me to run the PC, I get that Yellow indicator on the board "Memory Issue" at first I thought it was the memory stick gone bad, but this only happens when I set it on (Slot 1) ??.. I have switched the memories around and its the same problem? - I'm not even running XMP.

- I run 1 Stick on Slot 2 no problem.
- I run 2 Sticks on Slot 2,4 no problem.
- I run 3 Sticks on Slot 2,3 and 4 no problem.

- Slot 1 is giving me problems...

I cant even update to the latest LAN Driver v.1.0.1.4. and I went through the ridiculous steps:
ntel(R) Gigabit Ethernet Driver V1.0.1.4 for Windows 10 64-bit.
Before installing Intel I225-V driver v.1.0.1.4, please ensure that Intel I225-V firmware should be updated to v.1.45 via firmware update tool. After finishing necessary firmware update, please shut down and unplug the power cord then reboot the system normally. For more information, please refer to the firmware update guide.

I cant even update to the latest BIOS v0607 I went through the again the ridiculous steps:
ROG STRIX Z490-G GAMING BIOS 0607
"1.Improve system performance and stability 2.Improve compatibility with the Intel H10 Optane Module 3.Enhance AI Cooling mechanism
Before running the USB BIOS Flashback tool, please rename the BIOS file (SZ490G.CAP) using BIOSRenamer."

When I look into the file and use open the "BIOSRenamer" I see the same (SZ490G.CAP) already there, so I copy the CAP file into the root (USB FAT32) go into BIOS - Tool - "EZ Flash 3" I get a message saying SZ490G.CAP it's the wrong file?????...Something like that and I then go the old way at the back of the Board hit the Flash Button hold it for 3 seconds and nothing happens??... No LED indicator, because it's not doing anything...


Asus have the worst customer service or tech support from my personal experience, especially now with this COVID-19 that's like extreme! - That's why I'm here on the forum.
*Lian Li PC-011D (Black) *i9-10900K *ROG Strix Z490-G *Nzxt Kraken Z63/Corsair QL Series *Corsair Vengeance Pro RGB 64GB (4x16GB) DDR4 *Corsair HX1200i *EVGA GeForce GTX 3080 Ultra FTW3 *Samsung 970 Evo M.2 1TB *Samsung 960 Evo M.2 500GB *Samsung 840 Evo Series 1TB SSD *Asus ROG Swift PG43UQ *Corsair K70 RGB *Corsair Dark Core *Corsair Gaming MM800 *Corsair ST100 *Razer Nari Ultimate.

EXILE157 wrote:
Can anyone here help with my situation?.. I've no luck trying to find a post about the (ROG Strix Z490-G)

I've had nothing but issues from the start with this board and manage to make it work after hours upon hours for 2 days!...

Now here's my situation : For some odd reason DIMM Slot 1 is not allowing me to run the PC, I get that Yellow indicator on the board "Memory Issue" at first I thought it was the memory stick gone bad, but this only happens when I set it on (Slot 1) ??.. I have switched the memories around and its the same problem? - I'm not even running XMP.

- I run 1 Stick on Slot 2 no problem.
- I run 2 Sticks on Slot 2,4 no problem.
- I run 3 Sticks on Slot 2,3 and 4 no problem.

- Slot 1 is giving me problems...

I cant even update to the latest LAN Driver v.1.0.1.4. and I went through the ridiculous steps:
ntel(R) Gigabit Ethernet Driver V1.0.1.4 for Windows 10 64-bit.
Before installing Intel I225-V driver v.1.0.1.4, please ensure that Intel I225-V firmware should be updated to v.1.45 via firmware update tool. After finishing necessary firmware update, please shut down and unplug the power cord then reboot the system normally. For more information, please refer to the firmware update guide.

I cant even update to the latest BIOS v0607 I went through the again the ridiculous steps:
ROG STRIX Z490-G GAMING BIOS 0607
"1.Improve system performance and stability 2.Improve compatibility with the Intel H10 Optane Module 3.Enhance AI Cooling mechanism
Before running the USB BIOS Flashback tool, please rename the BIOS file (SZ490G.CAP) using BIOSRenamer."

When I look into the file and use open the "BIOSRenamer" I see the same (SZ490G.CAP) already there, so I copy the CAP file into the root (USB FAT32) go into BIOS - Tool - "EZ Flash 3" I get a message saying SZ490G.CAP it's the wrong file?????...Something like that and I then go the old way at the back of the Board hit the Flash Button hold it for 3 seconds and nothing happens??... No LED indicator, because it's not doing anything...


Asus have the worst customer service or tech support from my personal experience, especially now with this COVID-19 that's like extreme! - That's why I'm here on the forum.




Hello,

Remove the CPU cooler and CPU. Check for bent pins within the socket, then reattach the CPU and cooler whilst checking for even mounting pressure.

Also check that stand the motherboard standoff screws are not over-torqued.
13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

So after 4 month XMP became unstable again. I disable XMP, and set memory setting to default, but left CPU OC ON, and successfully passed Prime95 for 11 hours, HCI memtest and memtest86. So problem only with XMP, on previous settings it doesnt work any more. What could be a reason for this? And what i can do now to run it again? Just to try to find some new settings again? A lit bit worried, that XMP was stable only for 4 month.