Over 1.5v on B-die DDR4/AMD safe?

Mr. Sunshine

Mr. Sunshine

Mr. Sunshine

Associate
Joined
20 Dec 2020
Posts
37
I know there are a lot of older post about this topic but I was wondering what everyones thoughts are on pushing past 1.5v on B-die DDR4. I know that there are some kits out there that (Gskill, Corsair)run 1.5v.

I'm running Patroit Viper 4000mhz 32GB(4x8GB)@ 3800(14-10-15-28-42) 1.53v Memtest ran for over 24hr(2000%) no errors. Long term do you think this will be ok? I know that Adata has some kits that are spec'd @ 1.6v(https://www.xpg.com/us/feature/686/) so I was thinking that I might be ok.
 
The red herring in all this is these XMP profiles people keep referring to in judging if a voltage is safe. They see stuff like "the XMP is 1.5v so it's safe". Yes 1.5v is safe if you are doing automatic motherboard/XMP timings but when you start to lower these, it increases temperatures greatly. The voltage then no longer becomes "safe" unless you have adequate airflow to counter this.

I really like Igor's review here explaning in detail the relationship between timings and temperature, 1 tick completey changed stability in regards to what temperatures the modules could handle: https://www.igorslab.de/en/does-the...r4-3800-cl14-2x-16gb-put-through-its-paces/3/

I would highly recommend reading this. One thing I will say that most people miss and it's covered by Igor above: "I’ve had the case where my RAM was stable overnight while only the TM5 was running, but after 30 minutes of gaming the graphics card generated so much excess heat that the RAM became too warm and thus unstable."

I know you weren't asking about this part of "safe" voltage and more about chip degradation but it's always worth bringing up. If we do want to talk about chip degradation in which temperatures are not the problem due to cooling... let's just ballpark it at 1.6v. I know people run more but I say this for one specific reason. Right now in the pc space with how popular Ryzen is, there is honestly no reason to even go above 3800. Most good B-Die will be able to hit tight 3800 timings with 1.5v anyway (it just comes down to can you cool it well enough to not error).

"BUT INTEL!!!!" I hear you shout. With how bad the CES showing was for Rocket Lake, a tuned Ryzen system will beat that anyway which leads us to Intel's next stab which is... DDR5. It makes all the discussion about b-die voltage basically mute come end of 2021.

Also those XPG kits you linked are Hynix DJR. That's another story. :D
 
Yes it's safe. I'd recommend getting some air flow over the dimms as heat and bdie don't well. Like when you're gaming and your gpu back plate starts cooking the bottom of the dimm's, that'll be an issue for sure.
 
Thanks for the info guys. Heat shouldn't be issue. System is custom looped WC with 360/480 rads with 12*120mm fans. I also have 2*140mm pulling air out the top across the dimms. Great read on Igors lab good stuff.
 
Tightened up timings as much as I could. Not bad for cheap B-die 32GB(4x8) on a x470 board. Getting sub 54ns in Aida64. Might get little better with new BIOS. Will test to see if I can drop voltage on SOC
 
Mr. Sunshine said:
Tightened up timings as much as I could. Not bad for cheap B-die 32GB(4x8) on a x470 board. Getting sub 54ns in Aida64. Might get little better with new BIOS. Will test to see if I can drop voltage on SOC
Click to expand...

Amazing timings there. You got lucky with that Ram. What Voltage did you go with in the end?
 
I know this is really dead. But I still wonder how safe 1.6v really are. I assume they would not ship the kits with those specs if they could not handle it.
I would not worry to much about heat, because you would encounter instablity far before your chips get hot enough to get damaged.

My Patriot Viper 4400cl19s errored out after almost 2h @ 3800mt/s 14-14-14-28. I will try again once they have a waterblock. 4000 cl 16 runs fine maybe I think I can push 4000 cl 14 at 1.55 or 1.6 maybe even 14-14-14-28.
Does anybody know if chips degrade from voltage at 1.6v?
 
I would not run over 1.5 volts and I would keep the temps low below 60c. That just my personal limits. I bought a kit 3600 CL14 that runs at 1.45volts. I have not had a reason to mess with it and it performs very well.
 
rossi94 said:
I know this is really dead. But I still wonder how safe 1.6v really are. I assume they would not ship the kits with those specs if they could not handle it.
I would not worry to much about heat, because you would encounter instablity far before your chips get hot enough to get damaged.

My Patriot Viper 4400cl19s errored out after almost 2h @ 3800mt/s 14-14-14-28. I will try again once they have a waterblock. 4000 cl 16 runs fine maybe I think I can push 4000 cl 14 at 1.55 or 1.6 maybe even 14-14-14-28.
Does anybody know if chips degrade from voltage at 1.6v?
Click to expand...

Temperature is the biggest threat on b-die. If you actively cool the RAM with a fan you can run above 1.5v quite safely, right up to 1.6v. But nearer 1.55~1.57v max would be best.

I'm currently sitting here with my b-die

LtaDPLL.png

This is stable at 1.42v (MemTest5 extreme profile 12~16 cycles).

CL14 at 3800 is easily bootable, but I have to run voltages of like 1.55v~1.6v and even then I've had MemTest5 error out after an hour. This is with active cooling. CL14 at 3800 is just insanely voltage hungry and unless you have A-grade binning, or some of those new retail sticks rated for 3800 CL14, it's a tough and long journey to achieving it.

I've kind of decided instead of spending weeks trying to get CL14 stable at 3800 and 1.55v+, just to tighten as much as I can at CL16 3800. 1.42v won't need active cooling, case airflow is good enough for temps.

I'm still working on the above, it's just slow progress doing one setting at a time and leaving it overnight for 12+ cycles. But with memory you're better slow and stable than trying to cut corners.

But tldr; Yes, above 1.5v will be safe, just have to be actively cooled or make sure you are keeping temps around 40~45 or lower. The lower the better. Question then becomes is it even worth it to push this voltage for CL14 or just go as tight as you can at CL16? CL14 at 3600 is obviously far easier to achieve on b-die. Its 3800+ that is a bigger challenge. Hence why the retail rated sticks at CL14 3800 cost an arm and leg.
 
GregI said:
The red herring in all this is these XMP profiles people keep referring to in judging if a voltage is safe. They see stuff like "the XMP is 1.5v so it's safe". Yes 1.5v is safe if you are doing automatic motherboard/XMP timings but when you start to lower these, it increases temperatures greatly. The voltage then no longer becomes "safe" unless you have adequate airflow to counter this.

I really like Igor's review here explaning in detail the relationship between timings and temperature, 1 tick completey changed stability in regards to what temperatures the modules could handle: https://www.igorslab.de/en/does-the...r4-3800-cl14-2x-16gb-put-through-its-paces/3/

I would highly recommend reading this. One thing I will say that most people miss and it's covered by Igor above: "I’ve had the case where my RAM was stable overnight while only the TM5 was running, but after 30 minutes of gaming the graphics card generated so much excess heat that the RAM became too warm and thus unstable."

I know you weren't asking about this part of "safe" voltage and more about chip degradation but it's always worth bringing up. If we do want to talk about chip degradation in which temperatures are not the problem due to cooling... let's just ballpark it at 1.6v. I know people run more but I say this for one specific reason. Right now in the pc space with how popular Ryzen is, there is honestly no reason to even go above 3800. Most good B-Die will be able to hit tight 3800 timings with 1.5v anyway (it just comes down to can you cool it well enough to not error).

"BUT INTEL!!!!" I hear you shout. With how bad the CES showing was for Rocket Lake, a tuned Ryzen system will beat that anyway which leads us to Intel's next stab which is... DDR5. It makes all the discussion about b-die voltage basically mute come end of 2021.

Also those XPG kits you linked are Hynix DJR. That's another story. :D
Click to expand...


It's really not the memory that's in danger, but your memory controller. So, at 1.5v you're basically, heftily overclocking your CPU. It may break, or not; but your memory is rated for 1.5v, and will usually simply work as advertised, but your CPU is not rated to deliver 1.5v to the DIMM slots. Caveat emptor.
 
Just a quick note, I’ve been running around 1.57 daily on my launch day 9900k that’s left on 24/7 without a hint of deg in any component. Plenty I know daily bdie at 1.55 -1.63 on cfl-r and cml as well.
 
Been having issues with my B-die kit i.e. random reboots without a BSOD (yes, auto restart is turned off in windows settings) whenever a gaming session starts its ok for a few minutes then boom, PC will reboot/restart

The only thing overclocked is the RAM (XMP) so took that off and it was stable for week or two but now its doing it again. Took one stick out so now I'm only 8gb and its stable again.... for how long lord only knows.

The stick I took out, my word it was hot wonder if thats the issue it was only 1.35v or whatever the XMP is. So yeah looks like cooling is an issue for these kits even without 1.5/6v. Gonna have to run memtest and see if its degraded its only a couple of years old I've never had a ram kit deteriorate on me before, ever
 
You must log in or register to reply here.
Back
Top Bottom