5900x crashing

[Saotorii]

Hi, I'm new here but I come seeking help.

I recently completed a new pc build after upgrading from a 4770k so the build is entirely new. Specs are as follows

Mobo: x570 Taichi
CPU: Ryzen 9 5900x|
CPU cooler: Corsair H100i Elite Capellix
RAM: Trident Z Neo 3600mhz CL16-19-19-39 (F4-3600C16D-16GTZNC)
GPU: EVGA FTW3 3080 Hybrid
M.2 SSD: Samsung 970 EVO 1TB
HDD: WD Black 3TB @7200RPM
PSU: Seasonic Prime GX-850W

I before fully completing this build (was waiting on the 5900x) I was borrowing a 3700x from a friend and was running my RAM with DOCP settings at the above clocks with FCLK at 1800mhz. Everything else was set to defaults in bios (auto). At that time my bios version was 3.60, and later upgraded to 3.61 on the same chip. I ran with this set up from 11/7 - 12/30 with absolutely 0 issues. When I put the build together with the 3700x, I stress tested using prime95 as blended and folding at home for gpu, ran rock solid for 12 hours. Just to verify, I ran memtest86 and HC memtest, also with 0 issues.

The issue I'm currently having (which just cropped up 1/4 and has been on-going) is that running the exact same settings on my 5900x is resulting in crashing to black screen, and getting post code 07. This is at completely stock other than enabling the DOCP to be set to 3600mhz for my ram, everything else set to auto. So far the troubleshooting I have done is update bios to both latest and beta bios (3.81 and 3.91) and under bios versions 3.61, 3.81, and 3.91 have tried the following:

running just the ram at DOCP 3600mhz, FCLK downclocked to 1400mhz (i realize this is not efficient for performance, but I'm trying to test for stability primarily)(no crash)
running with DOCP, RAM set to 3200mhz, FCLK set to 1800mhz manually(no crash)
running at completely stock settings with no DOCP(no crash)
running with DOCP, RAM set to 3200mhz, FCLK set to auto (1600mhz)(no crash)
running with DOCP, RAM set to 3400mhz, FLCK set to auto (1700mhz)(this crashes to post code 07, but far less consistently than 1800mhz)

during all testing (running p95, folding at home on gpu) my gpu temps max around 60(set to be my exhaust) and my cpu temps usually settle close to around 65-70 with occasional spikes to 80(which should be fine)

I'm starting to think that my friends 3700x was a great sample, and that I got a super unlucky bin on my 5900x. If anyone could provide any additional insights, other troubleshooting steps, or just wants to break the bad news to me that I did in fact get a bad memory controller that cannot run FCLK at 1800, that would be much appreciated. If anyone would like me to post HWInfo screenshots, or attempt to take a HWInfo log (assuming I can boot to windows) when I crash, I'm more than happy to. Thanks in advance!

 

[thebufenator]

Did you play with any voltages? SOC, Vdimm? If some of those are set to auto, you might be getting different SOC on the 3700x than on the 5900x.


Just a thought

 

[Saotorii]

Did you play with any voltages? SOC, Vdimm? If some of those are set to auto, you might be getting different SOC on the 3700x than on the 5900x.


Just a thought

I've adjusted the SOC voltage from 1.1-1.2 and that doesn't seem to get stability, but between 1.165 and 1.175 soc I can generally get a boot to windows but it doesn't seem stable as I'll eventually crash to black screen. I've also tried this soc voltage range with dram voltage at 1.4 instead of 1.35 with the same results. I'm not familiar enough with ryzen to know what I should be setting the other voltage values to or if I should just be leaving those as is.

 

[DooKey]

Are you running the latest GPU drivers? Also, looks like the RAM doesn't like anything above 3200. I'd increase DDR voltage to 1.4 and see what happens.

 

[pendragon1]

if you havent yet, do a good bios clear and then if its still not stable do that^^^ (1.4v)

 

[Saotorii]

Are you running the latest GPU drivers? Also, looks like the RAM doesn't like anything above 3200. I'd increase DDR voltage to 1.4 and see what happens.

Yeah, I realized today there was a driver update. I used ddu to uninstall graphics drivers and clean installed the new drivers. Bumping voltage to even 1.45(as a friend of mine had advised) resulted in another crash. After extensive testing for the last couple days, including with a 3700x, I've narrowed it down to either a weird ram compatibility issue, or the cpu memory controller not being able to handle more than 1600fclk. I'm planning on getting a different ram kit that is on my mobo qvl and retesting, but I'm thinking I lost the silicon lottery hard on my cpu with a mem controller that won't operate past the "official" spec of 1600fclk supporting 3200mhz ram clock. Hopefully I can get a different ram kit just to make sure. If I can get a ram kit soonish, I'll report back

 

[Saotorii]

if you havent yet, do a good bios clear and then if its still not stable do that^^^ (1.4v)

Yeah, I usually reset everything to bios defaults, reboot, then set settings again to be on the safe side. Tested at both 1.4v and 1.45v with the same results. Referring to my reply to the above comment, I can post my testing methodology of you or anyone else would be interested