Question underclocking a 5800x3d

[maddogmcgee]

I'm back again with another computer related problem 😎

I have instability on my gaming PC after an upgrade. Random blue screens with memory related codes. Chrome instability etc.

5800x3d, b450 mortar titanium, currently 1 16gb kit of 3200 ram at xmp settings and 1.35v

I tried
Bios defaults with cmos clear
Moving to 2 sticks of ram from 4
Moving to 1 stick of ram
ram voltage 1.6,1.7
GPU reinstall with drivers and stock settings
GPU with quite mode
new motherboard (that was laying around)
windows refresh
turning off the cpu boost

The last one worked. Now that my sexy new and beautiful 5800x3d is stuck at an unchanging 3.4GHz, like it's 2010, it has been problem free for a week.

If it was a normal cpu I could increase voltage or maybe just play around with an all core clock. None of that is available on this one. Does anyone have a suggestion for any changes to the voltages I do have access to or a way to reduce the maximum boost speed of the chip?

I guess I could also try and return the cpu but I assume its more of an issue with using an older board rather than a cpu defect?

 

[Shmee]

I'm only running 2 sticks of ram atm, they are 2 seperate kits of 16gb which can cause issues itself. I have been hesitant to up the voltage too much but I will give the 1.4-1.45 ram voltage with manual timings a go. If that doesnt work I will start from scratch with 1 stick at a time at 1.4v 2133MHz. I actually have a brand new 2x4gb kit I can try as well. I did double check the manual to make sure I wasn't doing something stupid with the what ram slots were being used. I guess I also need to make sure I am running each ram test for longer.

You shouldn't need to up voltage on RAM that much, certainly not for working at what it is rated for. If there are issues with the memory, you should return it or RMA. What memory tests are you using other than OCCT? Have you run memtest86?

 

[maddogmcgee]

You shouldn't need to up voltage on RAM that much, certainly not for working at what it is rated for. If there are issues with the memory, you should return it or RMA. What memory tests are you using other than OCCT? Have you run memtest86?

5 minutes of OCCT is enough to get errors so I have just used that for the most part. I did give blended on prime95 a go and windows diagnostic tool. What you are saying about a memory issue would make sense since it's memory related bluescreens.....but I can't imagine 6 sticks of ram (2 brand new) all have an issues so I will test them one by one. If the brand new sticks get an error at 2133Mhz I think its same to assume its the CPU or SSD, as they will be the only parts reused. I guess memtest will also make sure that the crashes arnt due to windows corruption from all this ram stability testing. Appreciate the suggestion.

 

[maddogmcgee]

I’d update your BIOS before doing anything else first.

If you enable XMP but then lower your clocks to 2133 do you still crash? If tRDWR is still not matching, I would fix that in BIOS and then test again.

Defective CPU rare but it does happen. I’d return it asap and try another one.

I'm going to test one by one with 6 sticks OCCT first and if stable memtest86. Is it better to load motherboard defaults and test or do load XMP and lower speed to 2133?

 

[maddogmcgee]

So if ram kits are stable in memtest86 but memory tests are unstable in windows, what does that mean? Does that rule out the cpu as an issue as well? I never replaced the SSD, could that cause errors in Prime 95, OCCT memory tests etc?