2 x Dead 9820x CPU's in as many weeks! Thoughts?

[OfficerDD]

TLDR - First 9820x with mild OC worked fine but randomly died with 00 debug code on MSI Tomahawk Arctic x299 board. RMA replacement from Intel. 2 weeks of stable use on new CPU, put computer to sleep. Now same 00 code and dead CPU. What could cause this?!

 

Full Story

So I found what looked to be a good deal on Ebay a couple months back. 9820x in an MSI Tomahawk Arctic X299 board. Bundled with H100i, 250gb EVO Nvme and 32gb Corsair Vengence RGB 3200mhz DDR4 quad channel (4 x 8gb) with 4 x RGB Light Enhancement kit (Fake ram with RGB lights). All a bit more blingy than i care for but for the price (£850) it was a good deal so i took a punt.

 

Fitted it all in a nice new Fractal R6 with a brand new Seasonic 850w Plat Plus PSU and my existing Gigabyte G1 1080. I was coming from a 5960x at a stable 4.2 oc.

 

Having seen the stock setup benchmarking lower than my previous rig on Firestrike. I wanted a mild but stable O.C. I game a reasonable amout so would like good frame rates but I also do rendering and video editing hence why the X series has been a good option. Suffice to say the CPU score was higher but combined was really low. Anyway i digress. I did a simple OC by turning XMP on, changing Ring Ratio and setting the vcore on all cores to 1.185v. System ran stable for about a month with temps maxing out at 80c on full load using an IC Thermal Pad. Idle temps around 30-40c so no concerns at all.

 

One night after some trouble free gaming i put the computer to sleep. I wake it in the morning, fans spool, RGB on.... no post. Try the normal things first of unplug, hold power button etc... nothing. Clear CMOS, nothing. Then i see the 00 code on the board. Look this up and discover a LOT of people having the issue on X299 and X99 too. Many saying 'bricked board or CPU'.

 

So i try all the usual diagnostics including taking it all out of the case, rebuild outside it, swap ram slots, try alternate PSU, alternate RAM.... 00 everytime. So i buy a new Asus TUF X299 on Amazon knowing i can send it back if the CPU is fried. It arrives and sure enough the annoyingly bad Bluetooth dongle/App on the ASUS board says 0x00 code. /basically the board(s) don't register a CPU being on the socket despite 100% perfect pins. Bad news for a man that got it from ebay!

 

Very luckily though I check with Intel and the CPU is under warranty so i raise a case (knowing full well if it is a tray CPU it will get rejected). They accept the case now questions asked and in 7 days a seemingly brand new boxed 9820x arrives. In it goes, no dramas and all back up and running. Only weirdness I posted about on this forum was the UEIF showing vCore at 1.888v .... which was insane as it would have smoked (post of this is here but it turns out it was a false reading confirmed via HWInfo).

 

I OC'd it again to a very safe 4.0 this time and while a little miffed that a 5+ year old 5960x beats it still, i was happy that I can use those extra cores. All stable for 2 weeks with good temps and no BSODs.

 

So this weekend after some 4k video editing with no glitches i put the machine to sleep again and low and behold in the morning i turn on and get no post... i instantly look to the board and nearly start crying when i see 00 code again!! 2 dead 9820x's in as many weeks.

 

So naturally I feel it has to be something other than just the CPU here. Some suggestions go along the lines of there being an issue with MSI Bios with XMP causing the RAM to overvolt and kill the CPU. This was on the lates (Sept 2019) BIOS so if it is a known issue MSI havn't patched it. I can find hardly any info on this at all baring this post on Overclockers where 2 7800x's on MSI X299 died.

 

I am RMA'ing with Intel now of course but I now live in fear that they reject the case. Even if they don't I now can't trust the board and or RAM. Is there anything about the RGB element of the Corsair Vengance RGB that could contribute I wonder or even the Light Enhancement Kits (dummy RGB ram) pulling too much juice I wonder?

 

Any thoughts or suggestions as always GREATLY welcomed.

[Falkentyne]

Buy a different board, or don't put the computer to sleep?  Just power it off.

Both deaths happened in sleep mode.

[OfficerDD]

Both doable but not having sleep mode in fear of bricking the CPU isn't practical for me. From a bit more digging it has been suggested that the Vengance RGB RAM might be causing the CPU to voltage spike when the PC wakes due to it being cached in the RAM. All seems very subjective but then also makes sense. Could simply be that no one component is at fault, just the 3 together causes the problem. Just seems odd that more people on X299 havn't seen the behaviour if that was the case.