What causes a memory controller on a CPU to go bad?

[Luke]

So i built a computer for a buddy of mine and something weird happened. While he was playing it just shut off and when he went to turn it on the fans would spin but it would continuously restart and wouldn't POST. After diagnosing it turned out that the 1st and 2nd DIMM slots were no longer working. I tried a single stick in all four DIMM slots one at a time and only the 3rd and 4th slot work now.

 

Whats weird though is i tried installing a different CPU into the system and all DIMM slots work fine. I then put the 4790K into a different PC and it has the same issue with the 1st and 2nd slot not working so it seems to be the memory controller on the 4790K went bad. I thoroughly tested every component individually (RAM with Memtest86+, Prime95 for CPU and RAM, Unigine and 3D Mark in Windows etc). So it definitely is the IMC on the CPU.

 

Fortunately he still has two working DIMM slots but unfortunately it only runs in single channel now. Anyway, I just wanted to hear some thoughts on what may have caused this? Do you think it was the 2400MHz at 1.65V that somehow fried part of the IMC? I know that is sort of pushing the speeds on DDR3 but that is the XMP speed that Corsair validated this RAM to run at and it the memory itself is still fine. 

 

Or maybe it was the 4.6GHz OC on the 4790K? The temps were low and i wasn't pushing any more voltage than was absolutely necessary (I stress tested it at it at 1.250V but it was sometimes crash during certain games so i upped the voltage to 1.260V and it never crashed again). These are fairly modest overclocks from my experience so it would surprise me if these caused the problem, especially considering both the 4.6GHz OC and 2400MHz RAM are still completely stable. Maybe it was just bad luck? Anyway i just wanted to hear some thoughts on this. Cheers.

[Turtle Rig]

4 minutes ago, Luke said:

So i built a computer for a buddy of mine and something weird happened. While he was playing it just shut off and when he went to turn it on the fans would spin but it would continuously restart and wouldn't POST. After diagnosing it turned out that the 1st and 2nd DIMM slots were no longer working. I tried a single stick in all four DIMM slots one at a time and only the 3rd and 4th slot work now.

 

Whats weird though is i tried installing a different CPU into the system and all DIMM slots work fine. I then put the 4790K into a different PC and it has the same issue with the 1st and 2nd slot not working so it seems to be the memory controller on the 4790K went bad. I thoroughly tested every component individually (RAM with Memtest86+, Prime95 for CPU and RAM, Unigine and 3D Mark in Windows etc). So it definitely is the IMC on the CPU.

 

Fortunately he still has two working DIMM slots but unfortunately it only runs in single channel now. Anyway, I just wanted to hear some thoughts on what may have caused this? Do you think it was the 2400MHz at 1.65V that somehow fried part of the IMC? I know that is sort of pushing the speeds on DDR3 but that is the XMP speed that Corsair validated this RAM to run at and it the memory itself is still fine. 

 

Or maybe it was the 4.6GHz OC on the 4790K? The temps were low and i wasn't pushing any more voltage than was absolutely necessary (I stress tested it at it at 1.250V but it was sometimes crash during certain games so i upped the voltage to 1.260V and it never crashed again). These are fairly modest overclocks from my experience so it would surprise me if these caused the problem, especially considering both the 4.6GHz OC and 2400MHz RAM are still completely stable. Maybe it was just bad luck? Anyway i just wanted to hear some thoughts on this. Cheers.

Very bad luck.  You can abuse BCLK and your still fine.  It will die when the motherboard dies.

[Jurrunio]

Too much voltage (not sure how much DIMM voltage hurts it, but VCCSA and VCCIO can do a lot of damage), age, bad contact (either the pins or the socket, in this case unlikely to be the socket).

 

I think it's just old. 1.65V DIMM voltage is fine (could go higher imo), 1.26V Core is totally safe (1.35V is still safe imo, and it doesnt affect the memory controller), VCCIN (CPU input voltage) can be anywhere from 1.8V to 2V without problems (as long as you cool the CPU properly), VCCSA imo max at 1.3V and VCCIO at 1.25V.

 

Quote

 You can abuse BCLK and your still fine.  It will die when the motherboard dies.

@Turtle Rig keep your non sense elsewhere, that wont help with dropping memory channels in the slightest.

[Turtle Rig]

2 minutes ago, Jurrunio said:

Too much voltage (not sure how much DIMM voltage hurts it, but VCCSA and VCCIO can do a lot of damage), age, bad contact (either the pins or the socket, in this case unlikely to be the socket).

 

I think it's just old. 1.65V DIMM voltage is fine (could go higher imo), 1.26V Core is totally safe (1.35V is still safe imo, and it doesnt affect the memory controller), VCCIN (CPU input voltage) can be anywhere from 1.8V to 2V without problems (as long as you cool the CPU properly), VCCSA imo max at 1.3V and VCCIO at 1.25V.

 

@Turtle Rig keep your non sense elsewhere, that wont help with dropping memory channels in the slightest.

Why you angry at me....What have I done to you ?  BCLK puts added voltage and abuses everything on the motherboard from VRM to memory controller.  Some people prefered to change BCLK ..... but it draws more wattage even.  Anyhow Jurrunio the only non sense is when someone who has done nothing to you in any way shape or from but you write a nasty mean message like that for no reason.  Learn what BCLK is before you speak.  Thank you

[Luke]

2 minutes ago, Turtle Rig said:

Very bad luck.  You can abuse BCLK and your still fine.  It will die when the motherboard dies.

 

1 minute ago, Jurrunio said:

Too much voltage (not sure how much DIMM voltage hurts it, but VCCSA and VCCIO can do a lot of damage), age, bad contact (either the pins or the socket, in this case unlikely to be the socket).

 

I think it's just old. 1.65V DIMM voltage is fine (could go higher imo), 1.26V Core is totally safe (1.35V is still safe imo), VCCIN (CPU input voltage) can be anywhere from 1.8V to 2V without problems (as long as you cool the CPU properly), VCCSA imo max at 1.3V and VCCIO at 1.25V.

Thank you both! It's a shame but this type of stuff just happens i guess. At least the system still works (Albeit with a slightly lower frame rate) and there are really good upgrade options nowadays 

[Jurrunio]

@Luke btw Does C-states work on your system? On my system (same board as yours, with 2600k) running C-states will randomly crash the system, and no amount of voltage or reducing frequency helps. Disabling it is the only option, which means the CPU is stuck at max clock whenever it's running.

 

23 minutes ago, Turtle Rig said:

Why you angry at me....What have I done to you ?  BCLK puts added voltage and abuses everything on the motherboard from VRM to memory controller.  Some people prefered to change BCLK ..... but it draws more wattage even.  Anyhow Jurrunio the only non sense is when someone who has done nothing to you in any way shape or from but you write a nasty mean message like that for no reason.  Learn what BCLK is before you speak.  Thank you

You dont have to provoke me, spreading false info is already a good enough reason for me to target anyone.

 

the effects you mentioned about touching BCLK also applies to multiplier overclocking, only there are more disadvantages. the PCIe bus gets overclocked unless your board has a clock gen, you can corrupt if not break NVMe SSDs and auto settings can get screwy (as they usually cant read BCLK changes). You just dont touch BCLK, ever, when there's a problem.

 

Yeah you'd better learn BCLK and multipler.

[Zando_]

15 hours ago, Turtle Rig said:

Why you angry at me....What have I done to you ?  BCLK puts added voltage and abuses everything on the motherboard from VRM to memory controller.  Some people prefered to change BCLK ..... but it draws more wattage even.  Anyhow Jurrunio the only non sense is when someone who has done nothing to you in any way shape or from but you write a nasty mean message like that for no reason.  Learn what BCLK is before you speak.  Thank you

What? Any type of overclocking draws more wattage, and more voltage if you leave it on auto or have LLC on. 

14 hours ago, Jurrunio said:

You dont have to provoke me, spreading false info is already a good enough reason for me to target anyone.

That's about 90% of what he does, sadly. States wrong info as if he's an authority on the subject. 

14 hours ago, Jurrunio said:

@Lukethe effects you mentioned about touching BCLK also applies to multiplier overclocking, only there are more disadvantages. the PCIe bus gets overclocked unless your board has a clock gen, you can corrupt if not break NVMe SSDs and auto settings can get screwy (as they usually cant read BCLK changes). You just dont touch BCLK, ever, when there's a problem.

 

Yeah you'd better learn BCLK and multipler.

^^^ AFAIK X58 was the last gen where BLCK doesn't effect PCIe (it changes CPU, RAM, and Uncore clocks on X58, maybe a couple others but I haven't really been in the BIOS on my X58 board in a bit, it currently runs GPUs for folding so the CPU is at stock). On most newer stuff, upping the BLCK is just a "fuck my rig" button. Apparently peeps still did BLCK overclocking on X99, but there's no real advantages over just using the normal slap the CPU multiplier up and add more voltage method. On many other platforms anything above 101 BLCK makes everything wack, heck even that tiny of an increase can cause all sorts of instability.