Ram trouble, Am I wrong?

[StarsMars]

I have two kits of 2x16gb 3600cl16 trident z neo. So 4 sticks. 64gb total. But it’s two identical kits.

 

its my understanding that getting two kits to work at xmp/dcop is somewhat luck based. 
Is that correct?
 

I had it running for a few months at it’s rated 3600cl16 and 1800 f clock. 
 

today I was playing rainbow and noticed much lower FPS. Usually locked at 144 but was getting 70to 90. 
 

restarted the pc and boot looped. Reset the bios, runs at default 2133 but not dcop rated speed. 
 

Any tips or tricks??  Still trying to get it back at rated speeds. 

[Jurrunio]

7 minutes ago, StarsMars said:

its my understanding that getting two kits to work at xmp/dcop is somewhat luck based. 
Is that correct?

Yes

 

Try running 3466MHz with DOCP enabled

[StarsMars]

Just now, Jurrunio said:

Yes

 

Try running 3466MHz with DOCP enabled

I tried 3200 with dcop earlier with 1600 f clock; still looped. 
 

I’ll give it a try. 

[Jurrunio]

7 minutes ago, StarsMars said:

I tried 3200 with dcop earlier with 1600 f clock; still looped. 
 

I’ll give it a try. 

If slowing it down doesnt work, you may have to dip in to changing the timings manually.

 

Could try 1.1V SOC voltage and pray it's only crashing because of the memory controer

[StarsMars]

3 minutes ago, Jurrunio said:

If slowing it down doesnt work, you may have to dip in to changing the timings manually.

 

Could try 1.1V SOC voltage and pray it's only crashing because of the memory controer

Does it make sense that it would be fine at dcop and randomly stop months later?

[Jurrunio]

6 minutes ago, StarsMars said:

Does it make sense that it would be fine at dcop and randomly stop months later?

It usually doesnt happen in a couple of months but it could, i.e. barely stable from the beginning. This is hard to tell if you havent stress tested it from the beginning.

[Guest]

Could be what we call, degradation. It's simply to the point where it's just become unstable (for whatever reason) and being a memory thing, could be anything from the modules themselves, the board's bios, or even the on chip memory controller. 

 

A diag would include trying a different cpu, different memory or different motherboard each individually to determine which part decided :nope, not happy anymore.  

[StarsMars]

22 minutes ago, Jurrunio said:

It usually doesnt happen in a couple of months but it could, i.e. barely stable from the beginning. This is hard to tell if you havent stress tested it from the beginning.

Seems like it probably wasn’t stable to begin with. It took some patience
 

I ran mem test on it about a month ago. Took forever but no errors. What other way to you stress test ram?  

18 minutes ago, ShrimpBrime said:

Could be what we call, degradation. It's simply to the point where it's just become unstable (for whatever reason) and being a memory thing, could be anything from the modules themselves, the board's bios, or even the on chip memory controller. 

 

A diag would include trying a different cpu, different memory or different motherboard each individually to determine which part decided :nope, not happy anymore.  

The parts are all new as of 2019 December 2019. doubt it’s degradation but anything is possible. 
 

 It runs at 2133. Seems like it’s a lot to ask to get 64gb all running 3600cl16. 

though I do wonder if the motherboard might have an issue. I got it from amazon warehouse. So it had some package scuffs. ASUS x470 crosshair vii

 

i looked over the board pretty well and didn’t see any damage. 

 

[Guest]

Yea 9/10 times you cant physically see the damage.

 

64gb on 4 sticks is asking alot from AMD memory contoller. Yes.

 

Zen+ = 2933mhz past that is OC on memory controller (damage degredation can occure)

Zen2 = 3200mhz past that is OC see above.

 

It is possible to have damaged the memory controller.

Ive lost dual channel in the past running IF .. Ahem... Hyper Transport in excess of 4ghz. Was an FX 9590. Sad times but that sometimes happens when overclocking.

[Jurrunio]

8 minutes ago, StarsMars said:

What other way to you stress test ram?  

memtest by HCI, though more RAM = more time and it will load CPU cores fully.

 

Did you update anything recently? windows? BIOS?

[StarsMars]

2 minutes ago, ShrimpBrime said:

Yea 9/10 times you cant physically see the damage.

 

64gb on 4 sticks is asking alot from AMD memory contoller. Yes.

 

Zen+ = 2933mhz past that is OC on memory controller (damage degredation can occure)

Zen2 = 3200mhz past that is OC see above.

 

It is possible to have damaged the memory controller.

Ive lost dual channel in the past running IF .. Ahem... Hyper Transport in excess of 4ghz. Was an FX 9590. Sad times but that sometimes happens when overclocking.

You could be right.

I just got lucky with letting it boot loop and "memory train". I'm on the PC now at all the desired values.

Fingers crossed.

Just now, Jurrunio said:

memtest by HCI, though more RAM = more time and it will load CPU cores fully.

 

Did you update anything recently? windows? BIOS?

Yea, It was a long night when I did memtest last.

 

No, no updates.

It's running back at 3600cl16 fclock 1800 now.

I'll be running some stress tests.

 

Starting to wonder if the recent storms I went through cause a power outage and reset the bios or something. Idk, the more I learn the less I feel I know.

[Guest]

14 minutes ago, StarsMars said:

the more I learn the less I feel I know.

Tell me about it! lol. 

 

Things are like totally the same but different from past to present. 

Some processors are good running warm. Some not. Memory, chipset... all of it. Most within a general area but some (usually few) are called "golden" or diamond minecart awesome. 

Sometimes more voltage is better... sometimes not. Sometimes Cas 15 can be better than 14, sometimes not. 

 

Could be just the memory controller went unstable suddenly and simply needed a clean stock post up. Now agrees with you. 

 

Some things cannot be explained always. Another gentleman in a recent thread has a beep. Happens 8 minutes on the dot after post. Then again some 2 hours later repeating about the same over and over. Day and day out. Enough to drive a guy mad!!! I haven't the heart to tell yet another person to RMA their motherboard. Still working up to that. (not your's the other gentleman's) 

 

If you are only using 1.11v SOC, you should be ok up to 1.2v and should't create too much additional heat. 

 

For memory, I run OCCT. It gives options for AVX AVX2 AMD AVX (older chips) and no AVX settings. Choose the amount of memory you want to load up. Real time error reporting. Really super user friendly. I will give warning, the Cpu tests can pound some heat. But Not a single one test is tell all. Could be good today, not good tomorrow. That's the risks of overclocking.