My computer is burning RAM!

[Feconi]

I got an interesting board and processor from one of my friends, since he did not use them. After assembling the computer, it worked fine.

 

But after a few days it stopped working - at startup it reached post code 10 and went into reboot. By checking all the components, it turned out that two Kingston RAM bars failed at the same time. I thought they were defective and replaced them with new ones from Patriot. A week later, the same situation again - post code 10 and 2 broken RAM bars.

 

Why can this happen? What to do?

 

Link to the video (first I show the operation of the computer with working RAM, and then with the burned-out from 2:03)

google - https://drive.google.com/file/d/1bedZsw7lNUEioDG3W7OrvMrRptAWuJLP/view?usp=sharing

yandex - https://disk.yandex.com.am/i/dPRDbvnMmaaL6Q

 

System:

MB: ASRock Z77 OC Formula

CPU: i5-3570K

GPU: Palit GeForce RTX 4060 DUAL 

SSD: Kingston A400 480 GB 2.5"

PSU: OCZ ZS 650W 

old RAM: Kingston DDR3 8GB (2x4GB)

new RAM: Patriot DDR3 16GB (2x8GB)

[Kilrah]

You don't have Gigabyte RGB Fusion installed by any chance? It's known to "kill" DDR3 RAM (the commands it sends that are supposed to go to LEDs wipe the SPD info on the sticks)

[Feconi]

Just now, Kilrah said:

You don't have Gigabyte RGB Fusion installed by any chance? It's known to "kill" DDR3 RAM (the commands it sends that are supposed to go to LEDs wipe the SPD info on the sticks)

No, with the ASRock board, I use only ASRock programs

[leclod]

I guess you could also kill Ram by applying too much voltage

Edited September 28, 2023 by leclod

[Feconi]

1 minute ago, leclod said:

I guess you could also kill Ram by applying too much voltage

I use the voltage that is indicated on the RAM box - for Patriot it is 1.5 V

[leclod]

1 hour ago, Feconi said:

I use the voltage that is indicated on the RAM box - for Patriot it is 1.5 V

I understand, but I also think that's a lot of V.

By the way, did you set the 1.5V yourself or you just applied the XMP profile ?

 

Edit : ohh, that's ddr3. Voltage drops each generation. So 1.5V is probably normal on DDR3

Edited September 28, 2023 by leclod

[Echothedolpin]

56 minutes ago, leclod said:

I understand, but I also think that's a lot of V.

By the way, did you set the 1.5V yourself or you just applied the XMP profile ?

1.5v is well within spec for DDR3.

[leclod]

1 minute ago, Echothedolpin said:

1.5v is well within spec for DDR3.

Yes, I just noticed it's about DDR3

[Echothedolpin]

11 hours ago, leclod said:

Yes, I just noticed it's about DDR3

I had to double check myself, no worries!

[Beerzerker]

12 hours ago, Echothedolpin said:

1.5v is well within spec for DDR3.

Yes it is - JDEC standards dictate 1.8v's is the "Safe Limit" for DDR3 BUT there are sticks out there that won't tolerate 1.8v's for long, I recall some DDR3 sticks referred to as "Eco" (Can't remember the maker/actual name of these) that it's specs calls for 1.35v's stock.

Just know there is more than just the board that could affect sticks like this, a bad CPU can do it too but usually that happens immediately instead of over time if the CPU is the culprit so I'd have to say move that suspect towards the back of the line at this point.

BTW I have one of those boards too, good board overall and it can volt sticks up like nobody's business BUT you have to use manual (Non-XMP) settings to do that and it's possible the board's XMP profile reference (The way it's programmed in the BIOS per XMP profile detected) is causing it. 

I'm well aware the BIOS is kinda quirky and can do some strange stuff because mine did, ATM mine has a failed component (MOSFET popped) and I'm looking for a parts board to fix it with the correct replacement component so it's been awhile since I've ran it.
Did that running it on Ln2 so just know I'm well versed in running one on the ragged edge and it happens sometimes under the best circumstances.
In the case of this board it's entirely possible a component that regulates RAM voltage is going "Bye-Bye" and this a symptom of it doing so..... But then again maybe not.
If you change boards, run the same chip and it happens again in the other one, you can safely say the CPU itself is the cause of it and it woudn't be the first time I've seen it myself. 

[Feconi]

7 hours ago, Beerzerker said:

Yes it is - JDEC standards dictate 1.8v's is the "Safe Limit" for DDR3 BUT there are sticks out there that won't tolerate 1.8v's for long, I recall some DDR3 sticks referred to as "Eco" (Can't remember the maker/actual name of these) that it's specs calls for 1.35v's stock.

Just know there is more than just the board that could affect sticks like this, a bad CPU can do it too but usually that happens immediately instead of over time if the CPU is the culprit so I'd have to say move that suspect towards the back of the line at this point.

BTW I have one of those boards too, good board overall and it can volt sticks up like nobody's business BUT you have to use manual (Non-XMP) settings to do that and it's possible the board's XMP profile reference (The way it's programmed in the BIOS per XMP profile detected) is causing it. 

I'm well aware the BIOS is kinda quirky and can do some strange stuff because mine did, ATM mine has a failed component (MOSFET popped) and I'm looking for a parts board to fix it with the correct replacement component so it's been awhile since I've ran it.
Did that running it on Ln2 so just know I'm well versed in running one on the ragged edge and it happens sometimes under the best circumstances.
In the case of this board it's entirely possible a component that regulates RAM voltage is going "Bye-Bye" and this a symptom of it doing so..... But then again maybe not.
If you change boards, run the same chip and it happens again in the other one, you can safely say the CPU itself is the cause of it and it woudn't be the first time I've seen it myself. 

I took two more Patriot RAM sticks and managed to lower the voltage from 1.5 V to 1.2 V at a speed of 1600 mhz. MemTest86 had no errors. A very interesting experience. Either I received excellent memory samples or the motherboard greatly increases the voltage relative to the one set in the BIOS

[Beerzerker]

16 hours ago, Feconi said:

I took two more Patriot RAM sticks and managed to lower the voltage from 1.5 V to 1.2 V at a speed of 1600 mhz. MemTest86 had no errors. A very interesting experience. Either I received excellent memory samples or the motherboard greatly increases the voltage relative to the one set in the BIOS

Just keep an eye on it and if it happens again, try a different chip if you have one to use.
I'm glad to see you've has success so far, it's true sometimes an XMP profile can be "Messed up" causing problems like this, whether it's based on what info is in the sticks or board's BIOS.

If you apply an XMP profile it can override any voltage settings you set for it so be aware of that.

Yes, if I'm not mistaken it will tend to skew voltage settings when using an XMP profile after setting RAM voltage and you'll wind up with a different/higher voltage than you intended.

Note you have test points on that board you can use a voltmeter with to check voltages, one test point being for RAM voltage itself.
If you have a meter, use it to see what you get and how different settings changes/affects things by using the test point.