Dram voltage - IMC voltage and processor longevity

[vorsgren]

Back in the day, when Intel released the nehalem architecture with it's imc, it was said that a high dram voltage and memory controller voltage could kill your processor. Running your system with ram voltage - imc voltage >= 0.5v was bad for the longevity of the processor. Is this still true for modern processors, that a high dram voltage and low imc voltage is bad? Tried googling the matter and couldn't find anything on it.

 

[Bouowmx]

Not relevant anymore.

See also:

integralfx/MemTestHelper

C# WPF to automate HCI MemTest. Contribute to integralfx/MemTestHelper development by creating an account on GitHub.

github.com

 

[StrangerGuy]

You still need to watch out for out of the box auto-overvolts especially with enabling XMP, like on my old Z370 mobo.

Lots of Intel mobos with XMP enabled pushes ridiculous VCCIO/VCCSA voltages to the IMC to ensure every XMP RAM under the sun works, without warning the user it can be potentially dangerous to the IMC's longevity in the medium to long term.

 

[DrMrLordX]

Back in the day, when Intel released the nehalem architecture with it's imc, it was said that a high dram voltage and memory controller voltage could kill your processor. Running your system with ram voltage - imc voltage >= 0.5v was bad for the longevity of the processor. Is this still true for modern processors, that a high dram voltage and low imc voltage is bad? Tried googling the matter and couldn't find anything on it.

The problems associated with a voltage delta between IMC and DRAM always confused me, especially on newer platforms where everything should have been running on a separate rail. Pretty sure that is no longer a problem now outside of what @StrangerGuy mentions about not loading up too much VCCIO. You should be able to hit your DRAM with as many volts as you like without necessarily overvolting the IMC to match.

Still, my experience with an over-sensitive IMC on an old x2-3600+ makes me paranoid about that sort of thing. Glad someone else is thinking the same way.

 

[Zucker2k]

You still need to watch out for out of the box auto-overvolts especially with enabling XMP, like on my old Z370 mobo.

Lots of Intel mobos with XMP enabled pushes ridiculous VCCIO/VCCSA voltages to the IMC to ensure every XMP RAM under the sun works, without warning the user it can be potentially dangerous to the IMC's longevity in the medium to long term.

Is XMP not a standard (with predefined settings and voltages) set by Intel?

 

[DrMrLordX]

Is XMP not a standard (with predefined settings and voltages) set by Intel?

It is, but I'm pretty sure that enabling XMP profiles voids your CPU warranty unless you have the performance tuning plan.

 

[JoeRambo]

Enabling XMP is horrible idea for memory tuning anyway. MBs do evil things to make XMP settings work, some are obvious like hilarious IO/SA voltages and some are not, like relaxing timings that are not even exposed for tuning in the BIOS.
Usually advice for manual memory OC goes like this "If you ever enabled XMP, reset BIOS and start from scratch"

 

[StrangerGuy]

Enabling XMP is horrible idea for memory tuning anyway. MBs do evil things to make XMP settings work, some are obvious like hilarious IO/SA voltages and some are not, like relaxing timings that are not even exposed for tuning in the BIOS.
Usually advice for manual memory OC goes like this "If you ever enabled XMP, reset BIOS and start from scratch"

That's why as far as I'm concerned, XMP is borderline false advertising.

 

[vorsgren]

Enabling XMP is horrible idea for memory tuning anyway. MBs do evil things to make XMP settings work, some are obvious like hilarious IO/SA voltages and some are not, like relaxing timings that are not even exposed for tuning in the BIOS.
Usually advice for manual memory OC goes like this "If you ever enabled XMP, reset BIOS and start from scratch"

Interesting. Perhaps I shall have to retune my memory overclock, I just set XMP and tuned from there.