Overclock.net banner
1 - 20 of 163 Posts

·
Registered
Joined
·
799 Posts
Discussion Starter · #1 · (Edited)
The other day when I played some BF2042, I noticed a couple of WHEA errors as reported by HWiNFO64. I went into Windows Event Viewer, and saw that I had gotten a few of those over the past 6 weeks or so. I only play BF2042, so that's when the errors occur. This was with a proven stable and error free system. Things I have tried for troubleshooting :

  • Increasing vcore
  • Running CPU at stock speed
  • Reducing memory speed from 4400C16 down to 4133 Mhz (2 x 16GB dual rank memory)
  • Testing my 4500C18 profile (with looser sub timings)
  • Lowering uncore/ring speed
  • Upgraded from 2301 to the newest 2403 BIOS on my Asus Apex XII motherboard

I still get 1 or 2 errors everytime I play BF2042. I allways get this WHEA error :

HTML:
A corrected hardware error has occurred.

Reported by component: Processor Core
Error Source: Unknown Error Source
Error Type: Internal parity error
Processor APIC ID: 13

The details view of this entry contains further information.
The only line that changes is this : Processor APIC ID. Which I believe is the CPU core thread. Mostly thread 13, 15 and 19. I have not seen any lower.

Is this a sign that my 10900K has degraded, or could it be something else?

Edit : It was SVID Behavior setting in BIOS that was the problem.
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #2 ·
Btw : I have not run my CPU at a very high overclock or vcore. But I have run IO at 1.35v and SA at 1.45v BIOS set.
 

·
9 Cans of Ravioli
Joined
·
23,361 Posts
This was with a proven stable and error free system.
proven stable how? though if you're unstable now even at stock then it's kinda irrelevant.

if you're testing for stability with something that doesn't test AVX instructions then try again with something that does. BF uses AVX.
 

·
Registered
Joined
·
3,804 Posts
proven stable how? though if you're unstable now even at stock then it's kinda irrelevant.

if you're testing for stability with something that doesn't test AVX instructions then try again with something that does. BF uses AVX.
This to the t.
What you thought was stable was definitely stable non avx. Avx is way more intense and bf definitely uses it
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #5 ·
proven stable how? though if you're unstable now even at stock then it's kinda irrelevant.

if you're testing for stability with something that doesn't test AVX instructions then try again with something that does. BF uses AVX.
Maybe I wasn't clear enough. The system is not crashing. I get WHEA errors - like that one in my first post...

Just a few of them after I cleared the log :

Rectangle Font Parallel Screenshot Number
 

·
9 Cans of Ravioli
Joined
·
23,361 Posts
well aware you're getting WHEA errors, you said it in the first sentence.

WHEA errors mean you're unstable. you said you were stable, but if you were stable you wouldn't be getting WHEA errors.

again: what are you using to test for stability?
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #7 ·
again: what are you using to test for stability?
Cinebench R23, Prime95 in-place 112k fft non-avx, Blender Open Data (AVX), MemTestPro 7, TestMem5 anta777 (extreme and ABSOLUT) - and of course Battlefield.
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #8 ·
Like this :
Product Font Computer Line Screenshot
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #9 ·
My point being : Things were good a long time, then suddenly I got WHEA errors. Hence why I am asking if it is a sign of my CPU degrading :)
 

·
Registered
Joined
·
4,304 Posts
My SP110 10700K sprung a leak and suddenly ran 20C hotter out of nowhere. The amount of voltage it required for a given multiplier setting did not change. Have you noticed any change in temperature?
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #11 ·
My SP110 10700K sprung a leak and suddenly ran 20C hotter out of nowhere. The amount of voltage it required for a given multiplier setting did not change. Have you noticed any change in temperature?
Nope. Temps while playing BF2042 is in the 60-70C range during normal gameplay, and I saw max 77C during loading of maps.
 
  • Rep+
Reactions: Dinnzy

·
Facepalm
Joined
·
10,723 Posts
This to the t.
What you thought was stable was definitely stable non avx. Avx is way more intense and bf definitely uses it
Battlefield's AVX instructions are pedestrian. It puts less of a load on the system than Cinebench does. It also doesn't use AVX256, which triggers a downclock on ADL when you use an AVX offset. AVX128 instructions do not. What BF does do, unlike CB, is it stresses the entire system at once, not just CPU cores.

That being said, you guys need to realize he's talking about the infamous PARITY ERROR issue. These errors even happened at stock on some people's systems when Apex Legends first came out, before the code path rework. To this day, no one in the world (outside of Intel) knows precisely what triggers these parity errors.

I think everyone here completely forgot, but you can get Parity Errors in minecraft, which puts a super light load on everything except 2 threads (unless the scheduler was changed in a patch or something), while fully passing prime95 small FFT AVX1. And Minecraft DOES NOT use AVX. So you guys asking him to test AVX isn't helping matters.

Now if he were getting CPU Cache L0 errors, then yes, that's a purely unstable CPU.
 

·
Registered
Joined
·
750 Posts
Cinebench R23, Prime95 in-place 112k fft non-avx, Blender Open Data (AVX), MemTestPro 7, TestMem5 anta777 (extreme and ABSOLUT) - and of course Battlefield.
None of those tests would pick up a low-thread high-clockspeed AVX instability. For that, you need a test like Prime95 in single thread mode, with in-place FFT and AVX2 on. I use Corecycler to automate that testing across all cores.

My point being : Things were good a long time, then suddenly I got WHEA errors. Hence why I am asking if it is a sign of my CPU degrading :)
No, it is more likely that the newer BF update now leverages AVX code more heavily to achieve a higher performance and hence exposed new instabilities. However, I believe a WHEA error 19 is a one that is corrected by the CPU when it appears, so it may just be very borderline.

I had something more extreme happen to me with FS2020's performance update in the past. It started crashing to desktop when I loaded certain combinations of planes and airports. I needed to reduce core clocks by 100 MHz for the same core voltage on my CPU to get it stable again. Every other game and software was fine, but Corecycler did pick up instabilities when I ran it for the first time to figure out why FS2020 was crashing.
 

·
Facepalm
Joined
·
10,723 Posts
None of those tests would pick up a low-thread high-clockspeed AVX instability. For that, you need a test like Prime95 in single thread mode, with in-place FFT and AVX2 on. I use Corecycler to automate that testing across all cores.


No, it is more likely that the newer BF update now leverages AVX code more heavily to achieve a higher performance and hence exposed new instabilities. This happened to me with FS2020's performance update in the past. It started crashing to desktop when I loaded certain combinations of planes and airports. I needed to reduce core clocks by 100 MHz for the same core voltage on my CPU to get it stable again.
Prime95 will cause L0 errors on Skylake cores if unstable. Not parity errors.
 
  • Rep+
Reactions: Ashtrix

·
Registered
Joined
·
14 Posts
parity errors , most likely from degraded cpu cache which nobody takes into consideration, that cache is sensitive to overclocking, and that's where most kill their cpu reliability with.
 

·
Registered
Joined
·
49 Posts
2042 seems to be a thing of its own, stable elsewhere doesn't mean anything really for 2042. If you want to troubleshoot, set EVERYTHING to stock, XMP memory settings only, undo even any GPU OC/undervolt. Not just one or two things, default everything first. Then start adding back OC/settingss one by one until the errors come back. If I had to take a guess - its memory settings or ring/uncore, not the CPU core OC.
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #17 ·
That being said, you guys need to realize he's talking about the infamous PARITY ERROR issue. These errors even happened at stock on some people's systems when Apex Legends first came out, before the code path rework. To this day, no one in the world (outside of Intel) knows precisely what triggers these parity errors.
Thanks for the reply. In your opinion, should I just forget about these errors? I doubt I could get an RMA for this, but I can afford to buy a new 10900K if it becomes necessary.
I did a quick gaming session in BF2042 with HyperThreading set to Off, and no errors. Need to test more to see if it persists, but hmmm...

Edit : Do you overclock uncore/cache these days or just run it at stock 4300 Mhz?
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #18 ·
No, it is more likely that the newer BF update now leverages AVX code more heavily to achieve a higher performance and hence exposed new instabilities. However, I believe a WHEA error 19 is a one that is corrected by the CPU when it appears, so it may just be very borderline.
Yes, but how can it be borderline stable at stock CPU/uncore frequency.... As I said I even dropped the RAM speed from 4400C16 down to 4133 Mhz - and even set RTL and IOL to Auto for the 4133 run so to put less stress on the IMC.
 

·
Registered
Joined
·
799 Posts
Discussion Starter · #19 ·
parity errors , most likely from degraded cpu cache which nobody takes into consideration, that cache is sensitive to overclocking, and that's where most kill their cpu reliability with.
My uncore have been running at only 4700 Mhz, so not that high, but I guess that even that could be enough. 4900 Mhz for a short period when the CPU was new, but that was not stable anyways.
 

·
Registered
Joined
·
3 Posts
parity errors , most likely from degraded cpu cache which nobody takes into consideration, that cache is sensitive to overclocking, and that's where most kill their cpu reliability with.
I saw posts of 8700K, 9900K, 10700K, 10850K &10900K OC and some stock report WHEA corrected errors too, this is clearly a Skylake design side effect that all CPUs exhibit until 10th gen. About the Cache ratio/Ring bus, I think this is one of the reason why Intel limited RKL and ADL Cache ratios to far less multipliers than CML to under 4500MHz.

But the only wierd part to me is how people are fully stable on AVX and multiple bench runs but a damn game gets this. Also BF2042 is a badly coded game just look at how many bugs it has all over the place, pages and pages of changelog, same for FC6. DOOM Eternal is the best optimized game shame we do not get that optimization on many games.
 
1 - 20 of 163 Posts
Top