RAM overclocking issue

[JackedNerd]

Greetings, I'm new to the forum and pretty green when it comes to overclocking as well. I just build a new system about 6 months ago.

Ryzen 9 7950X3D
Nvidia RTX 4090 FE
Gskill Trident Z5 6400mhz (2 x 32)
Asus X670E-E Wifi mobo

I just moved and my system ran fine until I got to my new place. When I got here I updated my BIOS to the newest version at the time (a newer version was released since I got here which is now 1602). Since that BIOS update my PC keeps restarting aywhere from 10-30 minutes after i start gaming, even games that put little stress on the components (Im currently playing WoW classic). There has been instances where I've played for 6-8 hours and it's fine. My system is water-cooled so my temps are good. After doing some research I'm pretty sure I've nailed it down to a memory issue. My board manufacturer says the RAM I have can run at 6200mhz but whenever I manually OC the ram to that frequency in BIOS it continues to happen. I've dropped it down 5800mhz and I might get an average of a couple hours before it happens again. Like I said I'm pretty green so any diagnostic stuff I'm not too familiar with but it's just weird to me because it was fine because it was fine before I updated my BIOS. One other thing I'll mention that might have 0 significance is when it does happen the computer doesn't instantly restart the monitors turn off first and I can still hear sound coming from my computer for about 15-30 seconds before it actually goes into a restart which I thought was kind of bizarre. Any assistance in this matter would be greatly appreciated and if there was any identifying information that I forgot I apologize.

 

[Zerileous]

Why did you update? Can you go back?

Not knowing fully your bios changes and which revisions do what...but you might look into the soc voltage and if the BIOS changes that. There were problems with some chips overvolting soc with early BIOS and they changed it. My mind first went to that.

Are you using expo or just manually setting? Expo changes a lot.

 

[Niku-Sama]

no dont go back on bios. odds are the bios they upgraded from have the "bug" (or lack of insight more likely) in the AGESA for the X3D cpus.
dont want it pumping V SOC more than it needs to be and killing the chip... which was happening with some of the initial bios versions

so you just moved, so your computer got jostled around. check around and re-seat everything, unlikely but strange things have happened.
also how is the power quality coming out of your wall in the new place?

I had an issue at one place where the power was so dirty it would cause stability issues and my new place, which is in a building thats about 70 years older mind you, i dont have that problem.
it was so bad i could hear it coming out of my speakers, so much interference and loads of stability problems. The lights would have a slight shimmer when they were on and warmed up, all over the house.

I thought it was bad wiring but when they recently installed solar they had some one check it out and say that its just the power coming off the grid thats bad

 

[EarthDog]

I'd second the SOC voltage too... you have A LOT of memory running FAST. I'm GUESSING you moved from a BIOS older than 1301 to newest and the SOC voltage was lowered.

Check what the SOC voltage is currently set at and see if setting it to 1.3V (it's likely lower) stops the problem.

Also, what do you mean 'set manually'? You should just have to set XMP/EXPO (then the voltage bump above).

If I was you, I'd check the hardware and reseat everything. Reset bios to optimized settings, enable XMP/EXPO, save and reboot. Go into bios check SOC voltage, set to 1.3V and save, test.... report back.

If it works, the object is to lower that SOC voltage to use the least needed to be stable.

 

[JackedNerd]

no dont go back on bios. odds are the bios they upgraded from have the "bug" (or lack of insight more likely) in the AGESA for the X3D cpus.
dont want it pumping V SOC more than it needs to be and killing the chip... which was happening with some of the initial bios versions

so you just moved, so your computer got jostled around. check around and re-seat everything, unlikely but strange things have happened.
also how is the power quality coming out of your wall in the new place?

I had an issue at one place where the power was so dirty it would cause stability issues and my new place, which is in a building thats about 70 years older mind you, i dont have that problem.
it was so bad i could hear it coming out of my speakers, so much interference and loads of stability problems. The lights would have a slight shimmer when they were on and warmed up, all over the house.

I thought it was bad wiring but when they recently installed solar they had some one check it out and say that its just the power coming off the grid thats bad

Im at my parents temporarily but I was honestly considering the same thing. I will reseat everything as suggested and check SOC voltage and set to 1.3.

One thing I did notice earlier, the RAM i have isnt on my motherboards QVL list which seemed strange to me (I bought the RAM cheap off a buddy). It is XMP 6400mhz. Not sure if that might be a reason.

 

[Zerileous]

Yeah my understanding is AMD is more comfortable around 6000MHz. Is there a profile for a lower setting like 6000MHz or just the 6400MHz?

GSkill also has their own compatability list on their website, so you can check there if it has been tested on your board by gskill.

I would suggest setting the 6400MHz profile and then manually changing the clock to 6000MHz.

If you just change the clock speed without first loading the profile, then there will likely be issues with memory voltage, timings, and CPU settings.

 

[Niku-Sama]

supposedly bios with AGESA version 1.0.0.7 b (or c?) is supposed to provide better support and i think it was at @Woomack tested it out here and got some better numbers in overclocking although it wasnt with an X3D cpu if i remember correctly.

I have not updated my bios in a while, infact i noticed when i enabled XMP the bios was juicing up some unnecessary voltages and took the time to clamp down on them before it surfaced that this was a problem: https://www.overclockers.com/forums/threads/xmp-voltage-question.803625/

I'm still using that bios with VSOC clamped down to 1.20V as well as some other voltage tweaks, i felt the need to manually set those and wait and see what they did to fix it and its been less than impressive if you ask me. I am however running 6400:


have been since i finished that thread and its been fine. i know i am running the ram voltage manually, i'll have to check later though on the others beside ram and voltage. the timings arent great but its also not the best binned Hynix memory dies either... But its working at its rated XMP speed.

Some other variables to consider, its on an AsRock B650E taichi, Bios version 1.18 with AGESA 1.0.0.5 (the scary version!) and

If i were you i would set it at a safe spot and wait for an update. from what i've been reading 1.0.0.7 (or 1.0.7.0 depending on where your reading it) are still pretty bad.
Theres been rumblings that 1.0.0.8 is getting skipped since they have (supposedly) had so many revisions internally and theres a recent article about 1.0.0.9 at wccf:

AMD AGESA 1.0.9.0 BIOS Firmware Launches In Late August, Includes Support For Phoenix "Ryzen 7000G" APUs

AMD will offer its brand new AGESA 1.0.9.0 BIOS firmware for AM5 motherboards in late August which offers Ryzen 7000G "Phoenix" APU support.

wccftech.com

take it with a grain of salt of course, i imagine they mean 1.0.0.9 instead of 1.0.9.0 but they also mention 1.0.7.0 and both of these would be something like 17 and 19 major iterations off of the intital release and i dont think they have gone that far so i donno. some sketch there.

either way now that they arent killing chips as readily they'll grind to find an actual performance fix

 

[EarthDog]

AMD should work set it and forget it to 6400. For that capacity, 6000 though, right. The issue here is the 2x32gb capacity at that spnow.

It worked before, so there's a reason it isn't working now.

and its been less than impressive if you ask me.

What's been less than impressive? Any of the last 2-4 bios' have the 'fix' for the SOC.

 

[Niku-Sama]

the microcode updates, their "fixes", there were some failed fixes that have been pulled and with 1.0.0.7 being so buggy still. kinda a fail in my mind

 

[EarthDog]

Gotcha. There were some quick fixes of fixes in the past.

Things are solid at this point. Worth updating to, for the most part.

 

[Woomack]

I have no problems with AGESA 1.0.0.7c on ASUS mobos while B650E-I, B650E-E and X670E Gene have the same updates. Maybe it's causing a bit worse overclocking, but I could run 2x32GB kit up to 7600 (Kingston RAM). Problem can be at 1:1 ratio and 6400 as it usually doesn't work stable. 6200 is usually max that is fully stable, while recommended is 6000 and tighter timings like CL32-38-38 1.35-1.40V.

On ASUS mobos, AGESA 1.0.0.7c limits SOC to 1.30V and readings are about 1.28V in BIOS. It shouldn't matter as the motherboard sets as high SOC as it's required under auto settings. Actually, 2x16GB kits go up to 8000 at 1.25V, so I doubt that anything more is required for sub-7k on 2x32GB kits.

You can check Windows logs to find out if it registers anything before the PC restart (some bluescreens take just a second). RAM errors typically have something about memory.

Back to RAM. Load setup defaults, enable XMP/DOCP II and nothing else. Check if it works. If not then manually change memory clock to 6200 ... if not then 6000.

 

[EarthDog]

Problem can be at 1:1 ratio and 6400 as it usually doesn't work stable.

Even though it worked before the BIOS update? Was it common for the updated BIOS' that reach higher speeds to behave this way?

 

[Woomack]

In theory, the new BIOS only limits SOC voltage and adds timing range, so it's possible to set a higher clock with asynchronous ratios. If 6400 1:1 worked before then should now, but maybe something more has changed in the BIOS itself. I don't know what version was before as these X3D updates are already for some time (3 or 4 ASUS BIOS versions already).

I assume that in use are proper slots (2/4) and nothing else has changed.

 

[EarthDog]

RightO... so it "should" work by simply raising the SOC voltage........... in theory, lol.

 

[JackedNerd]

I reverted back to newest version of the bios. Some people on another forum suggested to manually set the SOC voltage to 1.3v and 6400mhz should work fine. Some said that's safe but some said its a little high. Thoughts?

 

[EarthDog]

That's what we said......... it's safe.. it's the limit. As was said earlier, back it down... we're just testing to see if it works at 1.3V/if that was the issue.

 

[JackedNerd]

I'll report back if it dumps on me again!

 

[EarthDog]

So, if it works at 1.30V for a while, back it down to 1.25V... if that works, 1.20V... etc. If it doesn't work, then raise it between the values (for example, 1.25V between 1.2 and 1.3) and it should be good.

 

[JackedNerd]

Was this done correctly? Not sure if extreme over voltage needs to be enabled or not

 

[EarthDog]

That's for Vcore...leave it disabled.

Looks like it's done correctly. Is this what it looks like after you saved and rebooted?