Oh dear god, I've been battling this for a few weeks now. I hadn't used my rig much, but now that COVID-19 is here, I've been gaming a lot. Our systems are almost identical, and I have wiped/reloaded/swapped as much hardware as I can and still get the error.
 
What's funny is I only get it in The Division 2, no other game I play or application causes it.
 
CPU Cooling - H100i V2
CPU - i9 7980xe
Memory - G.SKILL TridentZ RGB Series 32GB (4 x 8GB) 288-Pin DDR4 SDRAM DDR4 3200 (PC4 25600) Desktop Memory Model F4-3200C16Q-32GTZR
Graphics - (2) RTX 2080TI Founders Edition
SSD1 - CT500BX100SSD1 (Crucial 500GBSSD) - SATA - OS/config/recordings only.
SSD2 - Intel SSDPEDMW012T4 - PCIe - NvME SSD 1.2tb (Steam Games)
SSD3 - Samsung MZVLB512HAJQ-0000L7 - 500GB M.2 (Port 2) - Ubisoft Titles
SSD4 - Samsung MZVLB512HAJQ-0000L7 - 500GB M.2 (Port 2) - Origin Titles
PSU - Cooler Master Silent Pro 1200watt 80Gold
OS - Windows 10 Version 10.0.18362 Build 18362
 
As far as I know all firmware for all the things are up to date, I update bios yesterday to the motherboard.
 
I'm going to try to load optimized defaults and see if I get the crash and go from there, I will also drop my XMP memory profile down as it's set to 3400mhz now.
 
What's weird is that I ran Intel's Extreme Tuning Memory Benchmark over night while also running CPU Burner and MSI Kombustor and the system was just heating up my room the next day.. like what is it that the Division 2 is calling for that gives me the error!?

negroiso
 
I'm going to try to load optimized defaults and see if I get the crash and go from there, I will also drop my XMP memory profile down as it's set to 3400mhz now.

There you go. the 7980xe is only officially rated for DDR4-2666...  DDR4-3200 should work, but even that isn't a guarantee! You'd have to manually tune your RAM and probably the IMC voltage too to get 3400Mhz stable, don't expect the XMP profile to do that for you! 

Serious question then, my ram is DDR4 3200 (PC4 25600) falling below the official max then right? So just OC'ing the CPU should be fine and not worrying about the XMP profile?
 
Also, with cost in mind (since I was an early adopter of the 7980xe, what would be a decent cpu upgrade that would still allow the PCIe lanes needed for my SLI/NvMe and rig without going bonkers?
 
Also Also, I set everything to optimal defaults. The only thing I really ever change is disabling other boot devices, turn off C3 States, Turn off Virtualization/TurboBoost/HyperThreading, I also turn off whichever NIC I'm not using as well as U.2 and any other port/peripheral not in use. I notice that when gaming recently two CPU cores are always pegged out at 100% while the rest just sit there. I presume that's from games being mostly single threaded at best, would the stutter and jitter be better resolved in like an 8700k equivalent at this range?

The reason I went with the 7980XE at the time was because I was upgrading from a 3970xe and my planned and less expensive path to the 8700k at the time (300$ vs 2,000$) they were back-logged for months and my other rig had already taken a dump.
 
On the last note, I ran PRIME95 for 12+ hours, no issues at optimized defaults, i've played a few hours of game and so far no issues as far as it relates to the WHEA error. Just to be overly cautious, I don't know if I mentioned but I did wipe/reload my OS using the latest Windows Media Creation Tool, removed network card from the back, pre-downloaded all the latest drivers from this web-site and GPU and peripherals that I could. This is as clean of a build of Windows 10 Pro as I can get. I would load up my W10 LTSC build but it hasn't been updated yet to support HDR... (last I checked) otherwise I would be rocking that mofo.

cdc-951
I would say it still has life to it!! Just keep figuring out what part of the chip needs more voltage. This will be a time thing and testing issue. Over clock your system, but keep giving it voltage. Then take voltage from one section. Don’t take voltage from SA, IMC or cores all at once. It will make it too hard for you to figure out then. I’m sure you can figure it out. Have you delided it also to help with temps?


Also have you tried what Sajin posted?

Core voltage: 1.2v
Mesh/Cache/Ring voltage: 1.3
VCCSA/VCCIO voltage: 1.3

Before I go all crazy, can you confirm the screenshot is where those numbers need to be? (just don't wanna blow anything up.)

negroiso
Before I go all crazy, can you confirm the screenshot is where those numbers need to be? (just don't wanna blow anything up.)

 
I use "Adaptive" Vcore which is the way to go imo (and many others) because it only applies additional CPU voltage (over stock voltage) at Turbo/Overclock frequencies..That way the CPU isn't being hit with unnecessary voltage (and consequently more heat) that's not needed when the CPU is operating below Turbo/Overclock frequencies.
 
If you want to run a solid overclock all the time for some reason, with no downclocking, then "Override" is the Vcore setting to go with..Then you need to change the Windows Power Plan to "High performance" in Control Panel -> Power Options..It's more better imo to set your CPU to downclock and use the "Balanced" or "Power saver" plan and then when you want it to run full tilt, just simply/quickly switch to the "High performance" plan.
 
Also know that Vdroop will BIGLY impact Vcore under load..Vcore under load is what matters most..Using the "Auto" Vdroop setting may dump WAY more Vcore (and consequently more heat) on the CPU than what's actually needed, maybe even a dangerous amount, so you really need check it.

Good utilities to have are HWiNFO, as well as the Intel Processor Diagnostic Tool and Intel XTU for checking on things/stress testing/making dynamic changes.

Sajin

negroiso

cdc-951
I would say it still has life to it!! Just keep figuring out what part of the chip needs more voltage. This will be a time thing and testing issue. Over clock your system, but keep giving it voltage. Then take voltage from one section. Don’t take voltage from SA, IMC or cores all at once. It will make it too hard for you to figure out then. I’m sure you can figure it out. Have you delided it also to help with temps?


Also have you tried what Sajin posted?

Core voltage: 1.2v
Mesh/Cache/Ring voltage: 1.3
VCCSA/VCCIO voltage: 1.3

Before I go all crazy, can you confirm the screenshot is where those numbers need to be? (just don't wanna blow anything up.)

Looks good to me. Set your uncore voltage offset to +250 too.

sajin i too have the same issue with 9980xe after changing the voltage to override its fixed so far so good
 

startekselva

Sajin

negroiso

cdc-951
I would say it still has life to it!! Just keep figuring out what part of the chip needs more voltage. This will be a time thing and testing issue. Over clock your system, but keep giving it voltage. Then take voltage from one section. Don’t take voltage from SA, IMC or cores all at once. It will make it too hard for you to figure out then. I’m sure you can figure it out. Have you delided it also to help with temps?


Also have you tried what Sajin posted?

Core voltage: 1.2v
Mesh/Cache/Ring voltage: 1.3
VCCSA/VCCIO voltage: 1.3

Before I go all crazy, can you confirm the screenshot is where those numbers need to be? (just don't wanna blow anything up.)

Looks good to me. Set your uncore voltage offset to +250 too.

sajin i too have the same issue with 9980xe after changing the voltage to override its fixed so far so good
 

I just want to update, after May 29th to July 23, 2020 I haven't had a single WHEA_UNCORRECTABLE_ERROR occur. Not sure if it's Windows 10 updating to 2004, but mostly It's probably adjusting the voltages! Just wanted to report back again and say it's been stable with the suggestions here.

I haven't ventured into OC land yet, but if I do I will remember to slightly increase the numbers mentioned to keep them in line with "base" voltages that were required.
 
 
Thanks to everyone who helped!

negroiso

startekselva

Sajin

negroiso

cdc-951
I would say it still has life to it!! Just keep figuring out what part of the chip needs more voltage. This will be a time thing and testing issue. Over clock your system, but keep giving it voltage. Then take voltage from one section. Don’t take voltage from SA, IMC or cores all at once. It will make it too hard for you to figure out then. I’m sure you can figure it out. Have you delided it also to help with temps?


Also have you tried what Sajin posted?

Core voltage: 1.2v
Mesh/Cache/Ring voltage: 1.3
VCCSA/VCCIO voltage: 1.3

Before I go all crazy, can you confirm the screenshot is where those numbers need to be? (just don't wanna blow anything up.)

Looks good to me. Set your uncore voltage offset to +250 too.

sajin i too have the same issue with 9980xe after changing the voltage to override its fixed so far so good
 

I just want to update, after May 29th to July 23, 2020 I haven't had a single WHEA_UNCORRECTABLE_ERROR occur. Not sure if it's Windows 10 updating to 2004, but mostly It's probably adjusting the voltages! Just wanted to report back again and say it's been stable with the suggestions here.

I haven't ventured into OC land yet, but if I do I will remember to slightly increase the numbers mentioned to keep them in line with "base" voltages that were required.
 
 
Thanks to everyone who helped!