Question Ryzen 5 1400 - BIOS resets to base clock even though it keeps my Core Ratio Input ?

[AlexTatgjonaj]

I was overclocking my cpu and got to a stable 3.7ghz on 1.4v and 2666 at 14-16-16-16-34 for my ram (MSI Try It! Setting) .
After going into bios again to try 3.75 it shut down the moment I pressed save and reboot. I repowered my pc on and in the bios it kept all my settings but 1 thing stood out.My core ratio was still at 3.75 but my CPU was running at the base clock of 3.2.
So i dialed it back to the stable OC i had before ( 3.7ghz on 1.4v and 2666 at 14-16-16-16-34 ) and follows a weird procedure. After pressing save and reboot it shuts down after a bit,then boots up for 5s then again shut down,5s delay then boots back up into windows at 3.2ghz while still keeping my 3.7ghz Core Ratio.Anyone has any suggestions or Ideas?
Picture of the BIOS keeping the Core Ratio but running at Base Clocks
Video of what happens after Save and Reboot
~~~
My Theory:
1.I believe that the 5s Shut Down,5s standby then Bootup might be MSI's Memory Try which i have set to the default ( Enabled-2 tries ) But i have no clue why that would happen even with the stock settings.
2.Maybe BIOS is bugged?Will try reinstall a previous version of the BIOS then switching back to the newest again.But 1st want to see if there is another solution,if it shuts down during the update i am toast!
3. RGB was meant to be blue not red,but gets reset after it reboots.Didn't do that before so that tells me it loses power completely.Maybe MOBO battery got fried during the OC? Keeps the date and settings tho.

My System:
CPU: AMD Ryzen 5 1400 ( Cooler: Cryorig H7 ,CPU temps were under 78C at all times)
RAM: 4 Sticks (4x4GBs 2666) Ballistix Sport LT Red
GPU: Gigabyte GTX 1660 Super
MOBO: MSI B350 Gaming Plus (BIOS version:E7A34AMS.MHR 07/02/2020 )
PSU: BeQuiet PurePower 11 600W 80+Gold

 

[CountMike]

I was overclocking my cpu and got to a stable 3.7ghz on 1.4v and 2666 at 14-16-16-16-34 for my ram (MSI Try It! Setting) .
After going into bios again to try 3.75 it shut down the moment I pressed save and reboot. I repowered my pc on and in the bios it kept all my settings but 1 thing stood out.My core ratio was still at 3.75 but my CPU was running at the base clock of 3.2.
So i dialed it back to the stable OC i had before ( 3.7ghz on 1.4v and 2666 at 14-16-16-16-34 ) and follows a weird procedure. After pressing save and reboot it shuts down after a bit,then boots up for 5s then again shut down,5s delay then boots back up into windows at 3.2ghz while still keeping my 3.7ghz Core Ratio.Anyone has any suggestions or Ideas?
Picture of the BIOS keeping the Core Ratio but running at Base Clocks
Video of what happens after Save and Reboot
~~~
My Theory:
1.I believe that the 5s Shut Down,5s standby then Bootup might be MSI's Memory Try which i have set to the default ( Enabled-2 tries ) But i have no clue why that would happen even with the stock settings.
2.Maybe BIOS is bugged?Will try reinstall a previous version of the BIOS then switching back to the newest again.But 1st want to see if there is another solution,if it shuts down during the update i am toast!
3. RGB was meant to be blue not red,but gets reset after it reboots.Didn't do that before so that tells me it loses power completely.Maybe MOBO battery got fried during the OC? Keeps the date and settings tho.

My System:
CPU: AMD Ryzen 5 1400 ( Cooler: Cryorig H7 ,CPU temps were under 78C at all times)
RAM: 4 Sticks (4x4GBs 2666) Ballistix Sport LT Red
GPU: Gigabyte GTX 1660 Super
MOBO: MSI B350 Gaming Plus (BIOS version:E7A34AMS.MHR 07/02/2020 )
PSU: BeQuiet PurePower 11 600W 80+Gold

1. Yes, it's called "Memory training".
Also does frequency go up under load ?
Did you check Windows power plan ? Maximum processor state should be 100%.

 

[drea.drechsler]

I was overclocking my cpu and got to a stable 3.7ghz on 1.4v and 2666 at 14-16-16-16-34 for my ram (MSI Try It! Setting) .
After going into bios again to try 3.75 it shut down the moment I pressed save and reboot. I repowered my pc on and in the bios it kept all my settings but 1 thing stood out.My core ratio was still at 3.75 but my CPU was running at the base clock of 3.2.
So i dialed it back to the stable OC i had before ( 3.7ghz on 1.4v and 2666 at 14-16-16-16-34 ) and follows a weird procedure. After pressing save and reboot it shuts down after a bit,then boots up for 5s then again shut down,5s delay then boots back up into windows at 3.2ghz while still keeping my 3.7ghz Core Ratio.Anyone has any suggestions or Ideas?
Picture of the BIOS keeping the Core Ratio but running at Base Clocks
Video of what happens after Save and Reboot
~~~
My Theory:
1.I believe that the 5s Shut Down,5s standby then Bootup might be MSI's Memory Try which i have set to the default ( Enabled-2 tries ) But i have no clue why that would happen even with the stock settings.
2.Maybe BIOS is bugged?Will try reinstall a previous version of the BIOS then switching back to the newest again.But 1st want to see if there is another solution,if it shuts down during the update i am toast!
3. RGB was meant to be blue not red,but gets reset after it reboots.Didn't do that before so that tells me it loses power completely.Maybe MOBO battery got fried during the OC? Keeps the date and settings tho.

My System:
CPU: AMD Ryzen 5 1400 ( Cooler: Cryorig H7 ,CPU temps were under 78C at all times)
RAM: 4 Sticks (4x4GBs 2666) Ballistix Sport LT Red
GPU: Gigabyte GTX 1660 Super
MOBO: MSI B350 Gaming Plus (BIOS version:E7A34AMS.MHR 07/02/2020 )
PSU: BeQuiet PurePower 11 600W 80+Gold

The MHR BIOS is a BETA built to provide Ryzen 3000 support on the board. Not that it's incompatible with a 1st gen CPU but I do believe an earlier revision, MF, MD or even ME, might be better for pushing an overclock with a 1400. One problem, though, is MSI frequently makes it hard to revert BIOS. If you run into that there's a tool on the MSI forums that will do it.

The boot cycling is indeed memory training. When that happens after a crash it's frequently a good idea to shut down and do a CMOS reset that includes pulling battery. That completely clears out any residual memory settings that can mess up the settings you're making and cause instability.

Memory Try-It can work, sometimes, but it uses some canned timings MSI 'thinks' might be loose enough to work. I'd use XMP as that uses timings the RAM manufacturer has certified to work, so it's both safer and probably better latency at the same time. But in either case manually load the DIMM voltage...probably 1.35V...that's listed on the package they came in. That frequently helps with the boot cycling.

If core clocks in Window are stuck at 3.2 (when you set the ratio to 3.75, for instance) it can be because Cool-n-Quiet was not disabled. So try that and see if it works now.

And lastly when it shuts down, or crashes, as it did and then reboots back into BIOS it usually does retain settings (it does on my B350m Mortar). I believe the premise is if experimenting with an overclock you'll want to keep most of the settings but make a change to tweak in stability.

 

[AlexTatgjonaj]

1. Yes, it's called "Memory training".
Also does frequency go up under load ?
Did you check Windows power plan ? Maximum processor state should be 100%.

Frequency stayed at 3.2 during load after the incident and i have my Win10 powerplan on Ultimate Performance(enabled through the CMD) I did try everything but after hours of troubleshooting i found what caused this bizarre behavior.I removed all my Ram Modules and reinstalled them in pairs using the exact same settings.1st Pair worked fine at 3.7ghz and so did the 2nd pair when run seperately.For some reason after that when i reinstalled all 4 everything worked liked before and even got a 10 point boost in Cinebench.I have no idea how that simple re-seat of the memory fixed it since the problem occurred without any physical change to the system but i am glad it is up and running now.

 

[AlexTatgjonaj]

The MHR BIOS is a BETA built to provide Ryzen 3000 support on the board. Not that it's incompatible with a 1st gen CPU but I do believe an earlier revision, MF, MD or even ME, might be better for pushing an overclock with a 1400. One problem, though, is MSI frequently makes it hard to revert BIOS. If you run into that there's a tool on the MSI forums that will do it.

The boot cycling is indeed memory training. When that happens after a crash it's frequently a good idea to shut down and do a CMOS reset that includes pulling battery. That completely clears out any residual memory settings that can mess up the settings you're making and cause instability.

Memory Try-It can work, sometimes, but it uses some canned timings MSI 'thinks' might be loose enough to work. I'd use XMP as that uses timings the RAM manufacturer has certified to work, so it's both safer and probably better latency at the same time. But in either case manually load the DIMM voltage...probably 1.35V...that's listed on the package they came in. That frequently helps with the boot cycling.

If core clocks in Window are stuck at 3.2 (when you set the ratio to 3.75, for instance) it can be because Cool-n-Quiet was not disabled. So try that and see if it works now.

And lastly when it shuts down, or crashes, as it did and then reboots back into BIOS it usually does retain settings (it does on my B350m Mortar). I believe the premise is if experimenting with an overclock you'll want to keep most of the settings but make a change to tweak in stability.

The thing is that the core clock was showing 3.2 in the BIOS too.
I did try everything but after hours of troubleshooting i found what caused this bizarre behavior.I removed all my Ram Modules and reinstalled them in pairs using the exact same settings.1st Pair worked fine at 3.7ghz and so did the 2nd pair when run seperately.For some reason after that when i reinstalled all 4 everything worked liked before and even got a 10 point boost in Cinebench.I have no idea how that simple re-seat of the memory fixed it since the problem occurred without any physical change to the system but i am glad it is up and running now.I will give your suggestion of switching to a more stable BIOS version to OC and try 3.8ghz (currently 3.8ghz at 1.4v is not stable at all and whole system crashes).I did update the bios cause i was planning on buying a ryzen 5 3600 but decided to hold off since i might be moving soon for studies.Thanks for that heads up about the MSI forum,will head there if reverting the BIOS seems tedious

 

[drea.drechsler]

...I have no idea how that simple re-seat of the memory fixed it since the problem occurred without any physical change to the system but i am glad it is up and running now....

So... you have a 4 DIMM setup I see now. You do know 4 DIMM's are very sketch to run successfully in high clocks, especially so with 1st gen on a low-end B350.

But I believe I have an idea as to what's happening (if I understand what you did right) and it has to do with how the processor 'trains' memory. As I understand it when it first boots with a fresh bios (as after doing a BIOS reset with a battery pull) the processor makes certain tests on memory to find basic settings so it can proceed into the rest of the POST. When it 'trains' successfully I think it remembers the results and uses it the next time so it just goes straight to POST. So with all 4 modules in it fails to train, but with two in it succeeds (not uncommon with Zen1/B350 boards).

So when you subsequently swap in the next two it just goes straight to POST; you're essentially tricking it. Since you're operating at fairly low settings the trick works, but it might be unstable. I hope you've tested your setup thoroughly using HCI MemTest. I'd also suggest setting a manual DIMM voltage up around 1.4V might be in order since you have 4 DIMM's total. DDR4 is rated capable of 1.5V operation, but 1.4V is the smart limit for 24/7 useage.

 

[AlexTatgjonaj]

So... you have a 4 DIMM setup I see now. You do know 4 DIMM's are very sketch to run successfully in high clocks, especially so with 1st gen on a low-end B350.

But I believe I have an idea as to what's happening (if I understand what you did right) and it has to do with how the processor 'trains' memory. As I understand it when it first boots with a fresh bios (as after doing a BIOS reset with a battery pull) the processor makes certain tests on memory to find basic settings so it can proceed into the rest of the POST. When it 'trains' successfully I think it remembers the results and uses it the next time so it just goes straight to POST. So with all 4 modules in it fails to train, but with two in it succeeds (not uncommon with Zen1/B350 boards).

So when you subsequently swap in the next two it just goes straight to POST; you're essentially tricking it. Since you're operating at fairly low settings the trick works, but it might be unstable. I hope you've tested your setup thoroughly using HCI MemTest. I'd also suggest setting a manual DIMM voltage up around 1.4V might be in order since you have 4 DIMM's total. DDR4 is rated capable of 1.5V operation, but 1.4V is the smart limit for 24/7 useage.

yes i knew what i was getting into when i decided to go with a 4 DIMM setup.I thought of the same things with the sequential training of the 2 pairs but i thought it retrains if you install all 4.I know that DDR4 is capable of 1.4V on a daily basis but on the ram itself it says 1.2v so i got pushed away from that (even tho the Memory try it sets it at 1.360).I havent tried Memtest yet,i just did a quick 20 min Aida64 Extreme Stress Test while looking at temps and volts from HWmonitor.But during gaming or some casual editing i did i havent seen any instabilities.Would you suggest setting the 1.4V on the ram would give it some more safe headroom fom when i manually precisely OC it?(Despite the 1.2V sticker on the ram stick itself)

Edit: I am running 2.666ghz on the sticks at 14-16-16-16-36 ,1.36V Would that be considered a "high clock" even tho those are the manufacturer rated speeds?

 

[drea.drechsler]

yes i knew what i was getting into when i decided to go with a 4 DIMM setup.I thought of the same things with the sequential training of the 2 pairs but i thought it retrains if you install all 4.I know that DDR4 is capable of 1.4V on a daily basis but on the ram itself it says 1.2v so i got pushed away from that (even tho the Memory try it sets it at 1.360).I havent tried Memtest yet,i just did a quick 20 min Aida64 Extreme Stress Test while looking at temps and volts from HWmonitor.But during gaming or some casual editing i did i havent seen any instabilities.Would you suggest setting the 1.4V on the ram would give it some more safe headroom fom when i manually precisely OC it?(Despite the 1.2V sticker on the ram stick itself)

Edit: I am running 2.666ghz on the sticks at 14-16-16-16-36 ,1.36V Would that be considered a "high clock" even tho those are the manufacturer rated speeds?

It may be more than needed but I think 1.4V would be safe and give margin for stability if you don't want to test to find the lowest. I'm not sure how well the Aida Extreme test exercises memory but casual useage doesn't really put it to the test at all. You want something that will hit ALL of the memory and you can get HCI MemTest to do that. It's fairly easy, but does take quite a while to run through twice.

2666 is only considered 'high clock' since you're running a 1st gen CPU which is specc'd for only 2666 and I'm pretty sure that was for running 2 DIMM's. 4 DIMM's is very challenging on the memory controller in the CPU. Add in a low-end B350 motherboard which are well known for poor routing to the DIMM sockets and I have to think you're actually lucky to get it to boot and run at all with the default settings Memory Try-It gives you.