Question 5800X fails single core cycler test

[BTRY B 529th FA BN]

Bought the chip in Nov. 2020 for my main/daily rig and it's been running relatively smooth since then. I had a system crash with the following reboot saying there was no drive. Don't know what fixed it but 3rd or 4th boot it came back to life. I had a screen flicker about a week or so ago but nothing ended up happening: my GPU is really outdated it's a Radeon 6450 passive card with a 140mm fan blowing over it - stays cool. I'm not experiencing reboots nor WHEA errors but I decided to run a P95 core cycler test called CoreCycler-v0.7.9.2 that I found on the Overclock forums. Link: https://github.com/sp00n/corecycler The 5th or 6th core consistently gives an error (only 2 tests) The bios is a full release by Asus 3302 for the Dark Hero. However, it can pass OCCT for an hour without WHEA errors or any errors. Everything is at default, even the memory.




It seems ok but I'm developing a little suspicion. I've already visited the AMD Warranty page. Maybe it's the bios.. I'm waiting for the next release which is fairly soon, there's a beta with the mouse fix AGESA 1.2.0.2 listed on the support page.. I'm hesitant about going back to an older bios but I might look into it if I have to.

What do you think?

 

[Kenmitch]

Although it really seamed like a waste of electricity I decided to try my 5900x and see what the outcome would be. It's looking like mine is fine. Boost looks decent on the 1st CCD, but the second one is lackluster! I guess the lower boosting cores do run cooler at least.



I did let it run for a little while longer, but wound up shutting it down after it got back around to Core 3 again. Don't remember the uEFI cpu tweaks implemented on it currently. I do know the memory is 3600 CL14 with tight timings.

 

[thigobr]

I just got a confirmation they received my 5950X, ran tests, and approved the shipping of a new CPU. It took only one day to process this after they got the CPU back.

I think they have enough stock now to replace the items (it wasn't the case until beginning 2021) so it would be fine to RMA now. If increasing CO offset helps make the system stable then another option is to wait a little bit more to be sure these binning issues are properly sorted out and you get a new CPU that definitely works out of the box.

 

[Hail The Brain Slug]

I just got a confirmation they received my 5950X, ran tests, and approved the shipping of a new CPU. It took only one day to process this after they got the CPU back.

I think they have enough stock now to replace the items (it wasn't the case until beginning 2021) so it would be fine to RMA now. If increasing CO offset helps make the system stable then another option is to wait a little bit more to be sure these binning issues are properly sorted out and you get a new CPU that definitely works out of the box.

I'll be interested to see your corecycler results with your replacement 5950x.

I sent a support ticket to AMD to get the ball rolling, I figured I'd do whatever they asked first instead of going directly to the RMA request.

 

[BTRY B 529th FA BN]

On my gaming rig's 5800X it passed 4 iterations until i manually stopped the test. AMD Cool&Quiet was disabled. So was C-States. I tried testing today with both enabled and my fastest core failed. For most of the uptime of the gaming rig they've both been disabled, and both disabled is usually how I run my gaming pc.

 

[Hail The Brain Slug]

On my gaming rig's 5800X it passed 4 iterations until i manually stopped the test. AMD Cool&Quiet was disabled. So was C-States. I tried testing today with both enabled and my fastest core failed. For most of the uptime of the gaming rig they've both been disabled, and both disabled is usually how I run my gaming pc.

Now that's interesting. I wonder why cool & quiet and/or c-states would have any impact on peak boost frequency stability.

I will try disabling them and seeing what my 5950x does.

 

[BTRY B 529th FA BN]

AMD Cool&Quiet - Disabled
C-States - Disabled
Default CoreCycler-v0.7.9.2 test
Core 0: Fastest, & Core 2: Second Fastest, according to the chip

CORE OPTIMIZER TESTED RESULTS
Core 0 = -0 passes ~4730MHz = max
Core 1 = -15 passes ~4765MHz
Core 2 = -10 passes ~4800MHz
Core 3 = -15 passes ~4765MHz
Core 4 = -15 passes ~4765MHz
Core 5 = -10 passes ~4740MHz = max
Core 6 = -20 passes ~4765MHz
Core 7 = -20 passes ~4765MHz

Gonna run TM5, and then maybe OCCT, to see if any issues come up. When I use Ultimate Performance windows power profile with Processor Idle disable @ disable idle, all cores are at 4800MHz without any load.

 

[BTRY B 529th FA BN]

Here's something interesting. I rolled back my X570 Xtreme bios from F33f to F30 and the Best core, Core 0, sustained ~4780MHz no errors. Cool'nQuiet and C-states disabled, default CoreCycler test, iteration 1. Normally it would fail around that speed on the first iteration.

Gonna enabled CoolnQuiet and C-states after Core 2 finishes the first iteration, and run it again.

EDIT:

Enabling CoolnQuiet + C-States resulted in a higher frequency ~4830MHz on Core 0, however it failed within a minute of the 1st iteration.

Most everything is back to stock including IF and memory so that might be a reason I'm seeing a higher clock on Core 0 even with C&Q n C-states disabled. Either way it seems this bios is a little better atm.

 

[Kenmitch]

Here's something interesting. I rolled back my X570 Xtreme bios from F33f to F30 and the Best core, Core 0, sustained ~4780MHz no errors. Cool'nQuiet and C-states disabled, default CoreCycler test, iteration 1. Normally it would fail around that speed on the first iteration.

Gonna enabled CoolnQuiet and C-states after Core 2 finishes the first iteration, and run it again.

Performance/results will very from uEFI to uEFI. I've gotten a hold of a few that didn't even like my b-die memory. I've never disabled Cool-n-Quiet or the C-states myself. Sometimes the latest is the greatest after all.

 

[scineram]

Here's something interesting. I rolled back my X570 Xtreme bios from F33f to F30 and the Best core, Core 0, sustained ~4780MHz no errors. Cool'nQuiet and C-states disabled, default CoreCycler test, iteration 1. Normally it would fail around that speed on the first iteration.

Gonna enabled CoolnQuiet and C-states after Core 2 finishes the first iteration, and run it again.

EDIT:

Enabling CoolnQuiet + C-States resulted in a higher frequency ~4830MHz on Core 0, however it failed within a minute of the 1st iteration.

Most everything is back to stock including IF and memory so that might be a reason I'm seeing a higher clock on Core 0 even with C&Q n C-states disabled. Either way it seems this bios is a little better atm.

What does failure look like? System lockup?

 

[BTRY B 529th FA BN]

@scineram No lockup, just a rounding errror. No reboot, no WHEA errors.

EDIT:
F30 is a decent bios but it doesn't have Curve Optimizer, oof, lol, so F32 it is.

 

[Hail The Brain Slug]

Just going to throw out here that so far in my interaction with AMD support, they don't care one ounce about this problem. All they care about is if I get WHEA errors and crashes. They want me to disable bios settings that prevent it from boosting and retest and somehow that will determine if it's faulty.

So far, AMD support gets a C-. Erroneous calculations in a very long standing computational and stress testing program like prime95 is unacceptable.

Edit: they literally told me to disable Core Performance Boost, which is what allows boosting beyond the base clock. So I'm supposed to test it with the CPU limited to 3.4 ghz. Of course it's not going to fail, it's not an I/O or fabric problem.

 

[B-Riz]

On my gaming rig's 5800X it passed 4 iterations until i manually stopped the test. AMD Cool&Quiet was disabled. So was C-States. I tried testing today with both enabled and my fastest core failed. For most of the uptime of the gaming rig they've both been disabled, and both disabled is usually how I run my gaming pc.

I turn off CnQ on the X570 Master / 3900X, running the latest beta bios with the USB fix. CPU boosts to 4525 highest single core.

I vaguely recall an MSI rep post something to the effect that CnQ was a legacy thing that is not needed on Ryzen CPU's, especially in the case of the Zen3 that has a lot of power regulation baked into the hardware without need of a Windows power plan.

My 5800X is on an Asus X470 right now, I *think* I disabled CnQ on it, will check later.

 

[thigobr]

...

Edit: they literally told me to disable Core Performance Boost, which is what allows boosting beyond the base clock. So I'm supposed to test it with the CPU limited to 3.4 ghz. Of course it's not going to fail, it's not an I/O or fabric problem.

When I submitted the support ticket they also told me to change some UEFI options, more specifically Power Supply Idle Control but not to disable any boost or power savings feature.

They were still asking so many things that I ended up taking a video of me resetting the UEFI and loading Optimized Defaults, booting Windows, running Prime95 Single Thread and getting the rounding error. I also shared all the Windows logs that had some WHEA 18 fatal errors from when the system reset while playing games (all settings default).

The computer should be 100% stable by just loading defaults after all...

 

[Hail The Brain Slug]

When I submitted the support ticket they also told me to change some UEFI options, more specifically Power Supply Idle Control but not to disable any boost or power savings feature.

They were still asking so many things that I ended up taking a video of me resetting the UEFI and loading Optimized Defaults, booting Windows, running Prime95 Single Thread and getting the rounding error. I also shared all the Windows logs that had some WHEA 18 fatal errors from when the system reset while playing games (all settings default).

The computer should be 100% stable by just loading defaults after all...

They told me to disable Core Performance Boost to test the system to see if it's more stable with less power fluctuations - I went ahead and disabled it and the CPU only ever runs at 3.4 GHz, seems kind of like a self fulfilling prophecy - run the cpu at base clock only to see if it still errors?

That's a good tip. When I have some time I'll see about making a more comprehensive response like that.

 

[BTRY B 529th FA BN]

@thigobr Interesting about the Power Supply Idle option. I do see that under my Advanced CPU settings on my Xtreme.. I wonder if that will help when AMD Cool N Quiet and C-states are enabled and everything else is at default.

 

[thigobr]

Yes, it can help setting it to Typical Current Idle

Fedex just dropped of a new 5950X, now batch 2104PGS... Bad news! After installing, a Clear CMOS and booting I went straight to CoreCycler and let the tests run...Core #5 (second best) is now throwing rounding errors
If I exclude core #5 from testing it goes through cycler without any issues.

No WHEA or reboots though so far... But still not good. I might play with positive offset for Core #5 and keep this CPU for a while then try RMA again in a few months hoping they figure how to properly fix this issue out by then

 

[Hail The Brain Slug]

Yes, it can help setting it to Typical Current Idle

Fedex just dropped of a new 5950X, now batch 2104PGS... Bad news! After installing, a Clear CMOS and booting I went straight to CoreCycler and let the tests run...Core #5 (second best) is now throwing rounding errors
If I exclude core #5 from testing it goes through cycler without any issues.

No WHEA or reboots though so far... But still not good. I might play with positive offset for Core #5 and keep this CPU for a while then try RMA again in a few months hoping they figure how to properly fix this issue out by then

Is core 5 the best or second best core according to Ryzen Master?

I haven't had a real WHEA error/reboot in a long time now (possibly improved due to bios/agesa updates), nor can I really seem to reproduce any instability outside of corecycler. Having a hard time deciding if I should give up on the RMA process for now and see if it worsens/improves.

 

[BTRY B 529th FA BN]

@thigobr Do the cores boost to the advertised speed without failing?

 

[thigobr]

Core #5 is the second best on this new CPU...

So I have PBO disabled, everything default and core #5 is failing when boosting in single thread loads... Either Prime or y-cruncher. Multi core will lead to lower clocks then it's stable for hours... Same behavior as my previous CPU.

I tried playing with CO and added a +5 offset but I still get rounding errors on Core #5. This new CPU is also less forgiven for CO in general as some cores on the CCX2 won't accept -20 as the previous sample and the FCLK won't even run at 1866MHz! Let alone 1900MHz... It's a dud unfortunately

I will get a 5800X next week just to make sure this is not some other weird issue between my system and Zen3 specifically. I already tested a 1700 and 3700X though with exact same system and they were completely stable...

 

[Kenmitch]

Core #5 is the second best on this new CPU...

So I have PBO disabled, everything default and core #5 is failing when boosting in single thread loads... Either Prime or y-cruncher. Multi core will lead to lower clocks then it's stable for hours... Same behavior as my previous CPU.

I tried playing with CO and added a +5 offset but I still get rounding errors on Core #5. This new CPU is also less forgiven for CO in general as some cores on the CCX2 won't accept -20 as the previous sample and the FCLK won't even run at 1866MHz! Let alone 1900MHz... It's a dud unfortunately

I will get a 5800X next week just to make sure this is not some other weird issue between my system and Zen3 specifically. I already tested a 1700 and 3700X though with exact same system and they were completely stable...

Have you tried a different uEFI version to see if you get similar results? It's kind of strange you'd get back to back duds.

 

[killster1]

surprised no one is in here saying this is why they go INTEL i know i stated if you want a more reliable server / workstation intel usually wins. i love amd to death and hope this can be fixed. didn't test any my 5000 series chips but im not happy with RMA anything other than hdd's (even tho i have still have extra 3600 chips to replace them with)

 

[DrMrLordX]

Have you tried a different uEFI version to see if you get similar results? It's kind of strange you'd get back to back duds.

Especially when it's the same core.

 

[thigobr]

I will try another Zen3 before updating BIOS but it seems Core #12 is also bad on this sample... While playing a game I just got a fatal error on it, screen and fans turned off but the MB leds stayed on... I have never seen this before with this MSI MB

 

[B-Riz]

I turn off CnQ on the X570 Master / 3900X, running the latest beta bios with the USB fix. CPU boosts to 4525 highest single core.

I vaguely recall an MSI rep post something to the effect that CnQ was a legacy thing that is not needed on Ryzen CPU's, especially in the case of the Zen3 that has a lot of power regulation baked into the hardware without need of a Windows power plan.

My 5800X is on an Asus X470 right now, I *think* I disabled CnQ on it, will check later.

Could not find anything about CnQ in the Asus BIOS, for Prime X470-Pro.

Maybe someone with an X570 Asus can check?

Possible that Asus did not implement it or is calling it something else.

 

[B-Riz]

I will try another Zen3 before updating BIOS but it seems Core #12 is also bad on this sample... While playing a game I just got a fatal error on it, screen and fans turned off but the MB leds stayed on... I have never seen this before with this MSI MB

View attachment 42083

View attachment 42084

Do you have another board to test with?