WHEA Uncorrectable Error on Ryzen 3000

[AnthonyPaull]

Hey guys, I just upgraded from a 1700 to a 3700X on my X370 Taichi. I've started experiencing BSOD's  every few hours with the error reported in windows as:

0x124_AuthenticAMD_PCIEXPRESS_UNSUPPORTED_REQUEST_UNEXPECTED_COMPLETION_COMPLETION_TIMEOUT

And the BSOD reporting:

WHEA_Uncorrectable_Error.

I've tried disabling PBO and everything else in the bios bar XMP (I still experience crashes with it disabled). I've just disconnected 2 additonal drives that experience little use but I feel as though the issue is between the GPU and the CPU and possibly driver related due to the error.

 

Here's a non exhaustive but fairly complete list of things I've tried so far:

-Reinstall pretty much every driver

-Clear all junk files from OS

-SFC /Scannow

-Memcheck

-Diskcheck (On all disks incl C:)

-Reinstall windows

-Disable PBO/Auto OC etc and remove extra HDD's

-Disable GPU OC

-Few other things but I can't remember off the top of my head.

 

My system is as follows:

 

X370 Taichi running Bios version 5.60 AGESA 1.0.0.1

3700X

2X8GB Flare X B-Die 3200

EVGA 1080ti

Corsair RM 750i

 

There seems to be no indication that the PC is about to crash it just happens, both at idle and at load.

I'm running the latest Nvidia drivers, haven't attempted to rollback yet but that is on my list so I'm looking for any other suggestions. I emailed Nvidia's driver support team as well as AMD's support email about 36 hours ago but I haven't had a reply yet likely due to the weekend.

 

Any help is appreciated, I love this CPU the performance is awesome and Precision Boost is finally a real answer to Turbo Boost, I want to get this fixed us so I can enjoy my PC. Please let me know if there's any additional info that might help and i can provide it.

[steelo]

This is usually due to a hardware malfunction in your system. I would run hardware diagnostic tools to test your memory and hard drive. If they pass, try a stress tester, such as Prime95 or FurMark.

[wojtepanik]

x370 might not like the bios you updated, try reaching support of your mobo manufacturer, they will tell you more.

[AnthonyPaull]

7 minutes ago, steelo said:

This is usually due to a hardware malfunction in your system. I would run hardware diagnostic tools to test your memory and hard drive. If they pass, try a stress tester, such as Prime95 or FurMark.

As I said I've run a diskcheck and memtest, I know that this is *usually* to do with a hardware error but it doesn't occur with my 1700, and it's paired specifically with a PCIE error pointing to a driver issue between the CPU and GPU. I've run plenty of stress tests and nothing happens, it crashes at idle dude. In fact I'd say it's almost more likely to crash at idle/low load as it hasn't crashed once in cinebench, Aida 64, IBT or Prime 95.

8 minutes ago, starcoaster said:

WHEA_UNCORRECTABLE generally happens when the system isn't happy with an overclock. are you running one? i can see you've tried disabling XMP but usually memory problems come up as something like the MEMORY_MANAGEMENT BSOD

 

has it failed CLOCK_WATCHDOG_TIMEOUT at all?

No OC is being run on the CPU and as I said it crashed with the memory at stock a few times with the same error so I just re-enabled XMP

[AnthonyPaull]

1 minute ago, starcoaster said:

ok, try installing w10 fresh on another disk, you might not've seen it as i edited it into my post

I've already reinstalled windows, happened both before and after the reinstall. First install was on my old 250GB 840 Pro and the fresh install is onto a new 970 Evo Plus. Right now the only things plugged in are my 4TB games drive and my NVME ssd. I could try a fresh install with the NVME drive and HDD removed onto the old SSD but I doubt that would chance much due to the fact the reinstall didn't work the first time.

[AnthonyPaull]

14 minutes ago, starcoaster said:

has it failed CLOCK_WATCHDOG_TIMEOUT at all?

Come to think of it it may have failed just like that prior to the reinstall and WHEA errors but it hasn't happened since, and was right after I installed the CPU (Windows just had a heart attack and took 5 or 6 reboots with various settings and drivers reinstalled before working so it may have been an isolated incident related to that. With the fresh install it's been flawless aside from the crashing every 2-4 hours.

[AnthonyPaull]

1 minute ago, starcoaster said:

Have a check in the bios that the cpu isn't being run at a wierd voltage for ryzen 3 perhaps? i'm out of options other than that

Yeah dude I'm pretty out of ideas too and I consider myself reasonably tech savvy. The CPU is running flawlessly, will boost up to 4.4ghz on a few cores and the voltage will be around 1.4-1.5 but I'm under the impression that's normal, all core voltage comes well down to 1.25v at 4.2 or so. I have HWInfo constantly dumping monitoring data to a file that I think with the way I've set it up is recording right up until a few milliseconds before the crash and there's been no odd behaviour there.

[Caennanu]

Maybe a shot in the dark, but it seems you've exhausted just about every option i would think off.

Have you tried downclocking your cpu and see what happens? (if this does work stable, it is probably a bios thing, or powerdelivery)

[AnthonyPaull]

Just now, Caennanu said:

Maybe a shot in the dark, but it seems you've exhausted just about every option i would think off.

Have you tried downclocking your cpu and see what happens? (if this does work stable, it is probably a bios thing)

I could just try a dumb 'overclock' to 4Ghz all core at 1.35v or something to just stabilise the heck out of it I suppose, that would rule out any weird spiking behaviour with precision boost. I'll give that a shot tomorrow

[Caennanu]

4 minutes ago, starcoaster said:

yeah that's something i forgot to consider, the cpu might be keeping up but after a while if the VRM starts throttling back too far it will cause power delivery issues and crashes. i remember having this issue on my z97 board when i was trying to push for 5ghz and it kept falling apart earlier and earlier into the test runs the longer i went at it for

If it is caused by power delivery due to VRM's throttling. it should be fairly easy to 'add' cooling to them.

Don't know what cpu cooler you're using @AnthonyPaull but if its an AIO, i would try aiming a fan at them  Should be a fairly easy test.

[AnthonyPaull]

9 minutes ago, starcoaster said:

yeah that's something i forgot to consider, the cpu might be keeping up but after a while if the VRM starts throttling back too far it will cause power delivery issues and crashes. i remember having this issue on my z97 board when i was trying to push for 5ghz and it kept falling apart earlier and earlier into the test runs the longer i went at it for

I should probably point out that this mobo was able to run my 1700 at 4.35 pulling 230W at 1.58V with chilled water so I doubt the VRM is at fault here, and there's nothing in the hwinfo data to suggest throttling. It's also happening predominantly when the system is idle/under low load

 

I'm pretty confident in saying it's either an issue with the Pcie communication between Ryzen 3000 and the 1080ti, or it's some kind of bios issue. I'm not seeing many other possibilities at this point

[Bananasplit_00]

lowering clock speed would be the first thing i would do, WHEA_UNCORECTABLE_ERROR is from my experience only found when pushing the edge of stability on your CPU cores. Look at what the base clock is set too btw, if you are having PCIe errors the base clock might have somehow skewed away from 100MHz. I would also check every menu for PCIe related options to see if changing anything there helps. Also this is probably quite a longshot and not related but are you using two separate cables from the PSU to the graphics card or are you using a single cable with a daisy chain?

[Caennanu]

I didn't notice in start, but you're using B-DIE 3200 rated memory.

You didn't by chance overclock those beyond 3600 causing a 2:1 ratio, that might cause issues, did you?

[AnthonyPaull]

7 minutes ago, Bananasplit_00 said:

lowering clock speed would be the first thing i would do, WHEA_UNCORECTABLE_ERROR is from my experience only found when pushing the edge of stability on your CPU cores. Look at what the base clock is set too btw, if you are having PCIe errors the base clock might have somehow skewed away from 100MHz. I would also check every menu for PCIe related options to see if changing anything there helps. Also this is probably quite a longshot and not related but are you using two separate cables from the PSU to the graphics card or are you using a single cable with a daisy chain?

Single with daisy chain but this card and PSU have been running together for years with no issues now. 

 

I just set everything to stock in the bios and then set the CPU to 4ghz at 1.35v, I guess we'll see if it's the actual chip then... 

[AnthonyPaull]

1 minute ago, Caennanu said:

I didn't notice in start, but you're using B-DIE 3200 rated memory.

You didn't by chance overclock those beyond 3600 causing a 2:1 ratio, that might cause issues, did you?

No memory was either ar 2133/2400 or 3200 during all previous crashes

[Emanbaird]

Hey OP I just upgraded my PC I have an x470 Taichi, I also bought a 3700x and an EVO 970 1tb

I had this WHEA BSOD when I first set it up, I would try mucking about with the PBO options a bit, do you have LLC enabled? or any voltage offsets?

 

 Now, I am having another issue. Without fail my PC will restart itself after leaving it completely idle while updating FFXIV, If i'm updating FFXIV in the background I can play games all day, I wonder if this is some kind of power thing. I did notice that setting a negative voltage off-set caused some serious instability and also that my ram did not OC nearly as well as it used to when I was on the 2700x, I might try running a manual OC tonight to see if that fixes stability issues. 

 

 

EDIT: I found this link, 

 

 

 

[AnthonyPaull]

1 hour ago, Emanbaird said:

Hey OP I just upgraded my PC I have an x470 Taichi, I also bought a 3700x and an EVO 970 1tb

I had this WHEA BSOD when I first set it up, I would try mucking about with the PBO options a bit, do you have LLC enabled? or any voltage offsets?

 

 Now, I am having another issue. Without fail my PC will restart itself after leaving it completely idle while updating FFXIV, If i'm updating FFXIV in the background I can play games all day, I wonder if this is some kind of power thing. I did notice that setting a negative voltage off-set caused some serious instability and also that my ram did not OC nearly as well as it used to when I was on the 2700x, I might try running a manual OC tonight to see if that fixes stability issues. 

 

 

EDIT: I found this link, 

 

 

 

Thanks for the reddit tip, I've had a look and it seems like I've already sent most of the relevant info to them via email.

 

I just enabled LLC to level 3 to see if that potentially helped with stability, and as for PBO I've tried with it both on and off and it's still happening. Hoping I just get an Nvidia driver update soon that puts this bug to bed.

[Emanbaird]

9 minutes ago, AnthonyPaull said:

Thanks for the reddit tip, I've had a look and it seems like I've already sent most of the relevant info to them via email.

 

I just enabled LLC to level 3 to see if that potentially helped with stability, and as for PBO I've tried with it both on and off and it's still happening. Hoping I just get an Nvidia driver update soon that puts this bug to bed.

I'm going to try rolling back drivers later to see if that does anything for me.

[AnthonyPaull]

1 minute ago, Emanbaird said:

I'm going to try rolling back drivers later to see if that does anything for me.

Thanks dude, please let me know, I'll give this a shot tomorrow when I have time too and let you know if you haven't done it by then

[thefatman]

3800x after upgrading from my x370 asrock fatality gaming x board to my x570 gigabyte aorus pro wifi... i havnt had any bsod's but i am still getting "correctable" whea errors in my event viewer ...i do hate i got higher MHZ and LOWER voltages/heat on my x370 still a 10 point gap in my cpu-z score from board to board, but thats also a bios difference as the x370 only has 1.0.0.1 and the x570 has 1.0.0.3ab man i should have just waited 680 bucks just to WAIT for full performance

[thefatman]

i also tried everything but couldnt stop the issue on my x370, and its sad that i am still getting them on my x570 just no bsod's  since they are now "correctable" in the event viewer, its a bios issue plain and simple your still on 1.0.0.1and u got to wait till at least 1.0.0.2 or 3 to stop the blue screens, but u are STILL going to get the errors... u just wont notice them till u enter the event veiwer, as of 1.0.0.3ab

[Emanbaird]

It does appear that manual overclocks are more stable for me,  I have been running 4225 @1.35v for a while now and no stablitiy issues like I did recently

[AnthonyPaull]

9 hours ago, thefatman said:

i also tried everything but couldnt stop the issue on my x370, and its sad that i am still getting them on my x570 just no bsod's  since they are now "correctable" in the event viewer, its a bios issue plain and simple your still on 1.0.0.1and u got to wait till at least 1.0.0.2 or 3 to stop the blue screens, but u are STILL going to get the errors... u just wont notice them till u enter the event veiwer, as of 1.0.0.3ab

Idc about the errors as long as they don't crash my pc. Are you using an Nvidia GPU? And getting the Whea uncorrectable error BSOD specifically? 

[thefatman]

yes on my x370 board i was 1070ti

 

i broke down and got a x570 board no more blue screens, but still whea errors they are just "correctable now"

 

most not all x370's are on agesa 1.0.0.1, the x570 is agesa 1.0.0.3ab so i think u just got to wait for a new bios update for the fix, i know of other people getting the errors on amd video cards

 

but x570 still isnt perfect, voltage and temps are higher then when my 3800x was in my x370 board, i got better cpu bench scores in my x370 board also, but i think amd just rushed this launch and we have to wait for a better bios update .... also i tried alot of fix's with my x370 the only 1 that worked the longest was lowering the infinity fabric speed below 1500, but i still got a bsod like 3 days in , so i bought the x570 board no blue screens since

[AnthonyPaull]

5 hours ago, thefatman said:

yes on my x370 board i was 1070ti

 

i broke down and got a x570 board no more blue screens, but still whea errors they are just "correctable now"

 

most not all x370's are on agesa 1.0.0.1, the x570 is agesa 1.0.0.3ab so i think u just got to wait for a new bios update for the fix, i know of other people getting the errors on amd video cards

 

but x570 still isnt perfect, voltage and temps are higher then when my 3800x was in my x370 board, i got better cpu bench scores in my x370 board also, but i think amd just rushed this launch and we have to wait for a better bios update .... also i tried alot of fix's with my x370 the only 1 that worked the longest was lowering the infinity fabric speed below 1500, but i still got a bsod like 3 days in , so i bought the x570 board no blue screens since

Honestly, I was shocked when at stock I did a User benchmark run and got in the 96th percentile. The X370 boards may be running the chips faster but less stable? I'm pretty confident this will be fixed in the long run as my 1700 went from super flaky in the Taichi at launch to rock solid in Less than 6 months or so, especially with the flareX