ASRock 980DE3/U3S3, FX-8350 and BIOS 2.00 problem

Hi guys again,
I would like to post my full story, after all my investigations, so maybe someone could learn from it.

TL;DR version:
Never mix different RAM sticks - even if they come from same vendor and even if both series are similar / one is successor of the other. No way. Especially, if you find in your motherboard manual, that sticks must be identical. You must fix this even if you have read this after buying incorrect RAM and even if this will cost you additional money. If you keep incorrect mix of RAM sticks, then be prepared for 0x124 system crashes, WHEA 18, WHEA 19 and WHEA 20 errors.

Full version:
My father acquired PC which he used for office work. However, he got it from his friend who is a gamer. So PC had FX-8350 CPU, while mine had just Phenom II 1100T. I asked father and he allowed me for CPU exchange. Then I decided to upgrade amount of RAM from 8 to 16GB. As 8GB is cheaper than 16 + you have to do something with remaining old 8GB, I decided to just buy 8GBs and run on 4x4GB sticks. As I couldn't find Patriot G-Series AMD Black Edition sticks, I just bought Patriot G2-series AMD Edition sticks. They are nearly identical - old ones just have higher and thus more problematic heat spreader.  Then I said - OK new CPU, how about new BIOS - AIDA says my BIOS is so old... And when I have seen "CPU compatibility upgrade" in description of version 2.00 (mine was 1.60) I just immediately upgraded. And that was mistake. I changed 3 different things at the same time - CPU, amount of RAM and BIOS version. My PC started restarting - just sudden black screen and everything exactly if I pressed RESET button on my box. Plus there were many WHEA errors in system logs (Events 18, 19 and 20). At first it turned out that it's BIOS thing - I downgraded and system became much more stable. But still random crashes occurred. Some of them were connected with 0x124 error logged in System log. I know - you will say "overheating". But no - CPU temperatures in stress were not higher than 65*C, normally about 35-38*C and crashes occurred even while there was no stress, so rather cold. Motherboard was normally about 32-33*C and about 39-41*C in stress. So I started thinking that CPU is faulty. But then I said, OK, but how about RAM? Maybe they aren't that compatible as I thought initially? While all 4 sticks were installed system, they worked in Dual Channel mode. When I just removed the newer G2 sticks and left only my old G-Series ones system couldn't work in Dual Channel. Then I recognized problem from the past. 2 years ago I have found out, that you can't install just any stick in any of 2 DC slots (A1+A2 or B1+B2). You must install them in correct order. The truth is than later it turned out that there is no order rule. If DC is not working, it is enough to randomly put out and then put back stick from either B1 or B2 slot. Yes, you read correctly - to have Dual Channel working, you need to just remove the RAM stick and put it back in the same place - no other handling was necessary. Of course on system powered down. So, I did the magic and my old G-Series RAM started working in DC. At first there were no crashes. But then one occurred. So I put all RAM sticks back to mobo. Interesting - for a longer time there were no crashes. So I connected 2 things - when I removed new sticks old ones were not working in DC, so something was bad with them. For the moment, when I started from old sticks working in DC and added 2x new sticks from G2 series everything was quite OK. But of course not so long. So I left only newer sticks (G2 series) in A1+A2 slots. I performed different Prime tests (Blend / Max power / Max heat) for 12 hours without problem. My CPU reached 74*C and my mobo reached 50*C. It last for nearly an hour and no crash! All under topdown cooler (Noctua NH-14C). By the way - it gives higher temperatures than my old NH-U12P, which is NOT a topdown as suggested for 8 cores CPU on 980DE3/U3S3. So topdown cooling suggestion is not 100% correct...
Finally - we ca be now clear - it was all about incorrect mix of RAM sticks. So 0x124 is not only about overheating or about faulty CPU. It is also about incorrect mix of RAM sticks. Same goes to system errors WHEA 18, WHEA 19 and WHEA 20 which were frequently logged in my system shortly before most of crashes. They have all gone on "G2 series only" RAM configuration.

But this is not the end of the story. This is my 3rd motherboard since my old GA-7N400 PRO2 Rev 2.0 died. My first mobo for Phenom II was other Gigabyte - GA-890GPA-UD3H Rev. 2.1. What was really annoying on this mobo was the fact, that I couldn't use hibernation because whatever I did (deleting hiberfile.sys, disk defragmenting, etc.) it was always giving me 0x411 error when going back from hibernation. Always. I thought this was just the mobo problem. But today I am about to say, that this was old G-Series RAM problem and that they are faulty, despite memtest prooved them to be OK. I've read a bit about 0x411 error (which I have also recently found in System log) and it seems it must be RAM issue. This is something that I would like to point out for any user who may read it in future. The past 0x411 errors connected with necessity of trick with reinstalling of RAM sticks in the same positions as they were just to enable Dual Channel tells me, that it must be something with RAM sticks itself.
Plus  there is one more thing about that old Patriot G-Series RAM. They were sold as 1333MHz sticks. But they didn't contain a 667MHz profile - their highest frequency set automatically was 1066MHz (533MHz profile). So they were totally different from Patriot G2 series sticks, which had 667 profile and were detected as 1333MHz automatically. So guys be careful with Patriot RAM.

I will perform more tests with newer BIOS (probably 2.00) in "G2 series RAM only" config and I will post back results here.
But no matter what tests will show I am about to return FX-8350 to my father (where it worked with no problems) and I think I must start saving money for a new PC. Maybe this time it will be my first Intel? Who knows...

Mxyz Sxyz wrote: To have top-down cooling, as mentioned on motherboards description page.

That will help blow air over the VRMs which IMO are what's heating up and causing this. Hence my suggestion of including a separate fan over them.

This board is 4+1 VRM. And even though ASRock states 8-core up to 140W that's a stretch of my imagination being it's 4+1

I'm not Official ASRock but .....

Your 0x124 is most likely heat related. Specifically the VRM section if I were to guess.

Try this to test. Yea, I know. It reqs 2.00. Place a high air flow fan directing it over the VRM area next to the socket and I'll bet you get longer than 0.5hrs runtime.

Too, what processor cooling are you using? Water cooling by chance?


Otherwise I'd highly suggest copying your post into the Support Request form at http://event.asrock.com/tsd.asp