Memtest no errors, Stop Test causes BSOD

[just_dave]

Hello, i ve tried to change timings on my ram (Rated 3200 @1.35V 16-18-18-36) from rated ones to C14 14-14-14-34 (freq 2933 with both), system and everything feels stable and bit faster than before, but when i run memtest64, it shows no errors, tried for 20 cycles and 10 minutes but when it stops the test or i stop it manually, it goes into BSOD with error "attempted write to readonly memory". Should i keep the timings or not?

Thanks.

[3rrant]

20 minutes ago, dave_k said:

Hello, i ve tried to change timings on my ram (Rated 3200 @1.35V 16-18-18-36) from rated ones to C14 14-14-14-34 (freq 2933 with both), system and everything feels stable and bit faster than before, but when i run memtest64, it shows no errors, tried for 20 cycles and 10 minutes but when it stops the test or i stop it manually, it goes into BSOD with error "attempted write to readonly memory". Should i keep the timings or not?

Thanks.

As long as they run fine once you boot into the OS, it should be stable. The fact that memtest fails might be a lot of things, even just a bad "install" of memtest on the USB you're using. 

 

Use some windows based software to test them in real application, like Prime95 memory stress test, or Aida

[just_dave]

3 hours ago, ErrantNyles said:

As long as they run fine once you boot into the OS, it should be stable. The fact that memtest fails might be a lot of things, even just a bad "install" of memtest on the USB you're using. 

 

Use some windows based software to test them in real application, like Prime95 memory stress test, or Aida

@ErrantNyles It just crashed on me while playing War Thunder with IRQL NOT LESS OR EQUAL and then even didnt want to boot to windows so i went back to stock timings, well, i am running the windows based Memtest64, is there any other one?

[dexT]

Use more dram voltage 1.4, 1.45.

[just_dave]

1 minute ago, dexT said:

Use more dram voltage 1.4, 1.45.

Wouldnt that be too much?

[dexT]

1 minute ago, dave_k said:

Wouldnt that be too much?

No try either of those and then back off the voltage until unstable again and use last stable setting. Or keep going and clock the memory back to 3200 and use the tighter timings as well.

 

1.5v is the max Intel allows for XMP certification so that's most likely the highest safe 24/7 volts.

[just_dave]

7 minutes ago, dexT said:

No try either of those and then back off the voltage until unstable again and use last stable setting. Or keep going and clock the memory back to 3200 and use the tighter timings as well.

 

1.5v is the max Intel allows for XMP certification so that's most likely the highest safe 24/7 volts.

Mobo and most likely CPU wont let me 3200, only 2933, i am waiting for AGESA update, tried 14-14-14-34 1.4 volts and doesn't work, so i'll stick with the stock timings

[just_dave]

13 minutes ago, dexT said:

No try either of those and then back off the voltage until unstable again and use last stable setting. Or keep going and clock the memory back to 3200 and use the tighter timings as well.

 

1.5v is the max Intel allows for XMP certification so that's most likely the highest safe 24/7 volts.

So i tried memtest64 and it didnt crash when stopping test on stock, so yeah, ill stick with that.

I read on the forums that 1.4 volts is slowly going to be too much, so i will see what i can do at 1.35V, gonna try 15-15-15-34

[dexT]

1 hour ago, dave_k said:

So i tried memtest64 and it didnt crash when stopping test on stock, so yeah, ill stick with that.

I read on the forums that 1.4 volts is slowly going to be too much, so i will see what i can do at 1.35V, gonna try 15-15-15-34

Lower one timing at a time. You might find 14-16-16 working. 

 

I ran 1.45v dram on X99 24/7 so entertain raising the volts.

[just_dave]

4 minutes ago, dexT said:

Lower one timing at a time. You might find 14-16-16 working. 

 

I ran 1.45v dram on X99 24/7 so entertain raising the volts.

Well, i tried and nothing at CL14 works, also motherboard wont let me change to 2T and cant use CAS Latency 15, so i guess ill have to wait for new bios  Thanks for your help

[just_dave]

3 hours ago, dexT said:

Lower one timing at a time. You might find 14-16-16 working. 

 

I ran 1.45v dram on X99 24/7 so entertain raising the volts.

Okay, i tried to clear CMOS and followed your guide and got C14 14-15-15-35 at 1.35V rock stable, thanks for your guide!

[dexT]

6 minutes ago, dave_k said:

Okay, i tried to clear CMOS and followed your guide and got C14 14-15-15-35 at 1.35V rock stable, thanks for your guide!

Noice!

[just_dave]

22 hours ago, dexT said:

Noice!

@dexT Hey man, i tried and got stable at 14-14-15-35, is it better than 14-15-15-35 or should i revert back?

[dexT]

1 minute ago, dave_k said:

@dexT Hey man, i tried and got stable at 14-14-15-35, is it better than 14-15-15-35 or should i revert back?

Lower is always better so good job. 

 

I'd bet with 1.45v you could run the 3200 but I can see why you'd want to not risk it.

[just_dave]

2 minutes ago, dexT said:

Lower is always better so good job. 

 

I'd bet with 1.45v you could run the 3200 but I can see why you'd want to not risk it.

This RAM kit is kind of certified for Ryzen at 2933, the ryzen certified one is most likely the same (same Hynix chips) but tested at lower speed/cheaper. The AMD engineers should be working with this exact memory kit and include support in next AGESA.