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 I had a screen flicker about a week or so ago but nothing ended up happening: my GPU is really outdated it's a Radeon 6450 passive card with a 140mm fan blowing over it - stays cool.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?

 

[Hitman928]

Has anyone confirmed their failures by running prime95 manually?

 

[Makaveli]

my CO is currently set to -15 on my two best cores and -5 on the rest

 

[BTRY B 529th FA BN]

Has anyone confirmed their failures by running prime95 manually?

Is there a manual option to run SSE only with P95?

 

[Hail The Brain Slug]

Is there a manual option to run SSE only with P95?

I think that's what it uses if you disable AVX/AVX2 in the settings.

 

[BTRY B 529th FA BN]

I think that's what it uses if you disable AVX/AVX2 in the settings.

I disabled all AVX but it doesn't boost to the max while running all cores, hence the since core tester.

 

[Hail The Brain Slug]

I'm playing with manually testing with prime95 by running one thread and hand-setting affinity

 

[BTRY B 529th FA BN]

I've started to manually run 95 too. The error I keep getting is on Core 5 (CPU 10) Yup, manually running P95 and setting the affinity to core 5 will error out

 

[Hail The Brain Slug]

I've started to manually run 95 too. The error I keep getting is on Core 5 (CPU 10) Yup, manually running P95 and setting the affinity to core 5 will error out

I got failures on my own P95, latest version instance as well.

Confirmed failures with AVX disabled. Running it again with AVX to see if it will also fail.

Looks to me like it's going to be RMA time soon.

 

[BTRY B 529th FA BN]

I got failures on my own P95, latest version instance as well.

Confirmed failures with AVX disabled. Running it again with AVX to see if it will also fail.

Looks to me like it's going to be RMA time soon.

Yup, same RMA thoughts. Now I'm trying to figure out what to do in the mean time for the RMA turn around, to have a running system.. I could use my old X58 rig still.. most things are the same on it..

 

[Hail The Brain Slug]

Yup, same RMA thoughts. Now I'm trying to figure out what to do in the mean time for the RMA turn around, to have a running system.. I could use my old X58 rig still.. most things are the same on it..

I might have to buy a new 5950x and sell the replacement after RMA since it sounds like it takes around 2 weeks.

 

[BTRY B 529th FA BN]

Testing my gaming rig before I start any RMA process cause if that one is also bad I'd rather minimize the time I spend in an RMA process.

 

[Hail The Brain Slug]

It's interesting how with AVX enabled it is definitely boosting 50-100 mhz lower and not having any failures. Unfortunately most of my work is not AVX so I would expect to see the higher boost tier that is shown to be unstable.

Edit: boosting vcore by +0.0125 or +0.0250 didn't fix the errors. But applying a +10 positive curve optimizer offset to the two cores appears to have reduced their max boost by 100 MHz or so and prevents them from throwing errors.

 

[VirtualLarry]

It's interesting how with AVX enabled it is definitely boosting 50-100 mhz lower and not having any failures. Unfortunately most of my work is not AVX so I would expect to see the higher boost tier that is shown to be unstable.

New AGESA to fix "unstable SSE boost levels" coming in 3... 2... ....

 

[Hail The Brain Slug]

New AGESA to fix "unstable SSE boost levels" coming in 3... 2... ....

It's really not SSE that is the issue, it just happens to be the instruction set P95 falls back on when AVX is disabled. The CPU boosts lower under AVX load, so this is testing it's stability when it boosts higher under non-AVX workloads. They could still definitely nerf boost, but that would be very questionable. There are clearly processors that are stable under this test at stock settings.

 

[BTRY B 529th FA BN]

Just occurred to me that the failing (brain) i mean core is also one of two fastest cores in the chip. 2nd fastest to be exact which has been core 5.

Running the test on my gaming rig now. It's already tested both fastest cores.. here's hopin

 

[BTRY B 529th FA BN]

Well the gaming rig passed the first iteration, all cores. The gaming rig is running 1:1 1866/3733 so the vSOC is not the same as the daily/main rig. The only thing adjusted with the gaming rig is the vSOC, chip wise. So I've increased the vSOC on the 5800x that's throwing errors from 0.9750 to 0.9850 and am re-testing.

 

[BTRY B 529th FA BN]

How do I nerf the frequency of a specific core? I've increased the vSOC but that seems like a blunt brute force type of work around cause the other cores don't need more vSOC, so the best route would be to nerf the frequency of core 5 for me. I've tried adding a little vSOC and core 5 did pass the first iteration but immediately failed starting the second.. trying vSOC 0.9983

EDIT: increasing the vSOC isn't helping.. failed again

 

[Hail The Brain Slug]

How do I nerf the frequency of a specific core? I've increased the vSOC but that seems like a blunt brute force type of work around cause the other cores don't need more vSOC, so the best route would be to nerf the frequency of core 5 for me. I've tried adding a little vSOC and core 5 did pass the first iteration but immediately failed starting the second.. trying vSOC 0.9983

You have to go into curve optimizer, set it to per core, then while leaving the rest of the cores at 0, set the bad cores to a positive offset. Maybe try 5 to start, then increase to 10 if that isn't enough. 10 seemed to be enough on mine.

 

[BTRY B 529th FA BN]

You have to go into curve optimizer, set it to per core, then while leaving the rest of the cores at 0, set the bad cores to a positive offset. Maybe try 5 to start, then increase to 10 if that isn't enough. 10 seemed to be enough on mine.

Same here. Wow, huh, hmm

 

[thigobr]

It took 1 week for Fedex to return the CPU as they give you a pre-paid label. I believe it will take a couple days processing and then they will send another one back. Probably 2 to 3 weeks total (it just arrived there today afternoon).

I also I tried downloading the latest Prime95 and running manually: 1T, AVX disabled then setting affinity by hand. Exact same results as the Corecycler: in less than 3minutes it showed rounding errors on Core #0.

I hope the new one can boost past 5GHz like the old one but still be stable.

 

[BTRY B 529th FA BN]

Now I'm a little undecided what to do with this chip. Seems to be stable with a positive 10 curve optimizer on core 5 as it passed ~5 iterations with a manual P95 setting. It seems to me this is exactly what AMD intended Curve Optimizer to do to hard limit bad core(s) before being sold, but now the public has access to it.

My 5800X in my gaming rig seems to be ok as it passed a few iterations of corecycler but none of the cores boost past 4800MHz with the majority of them in the low low 4700MHz range. Core 5 on the "bad" chip was boosting to 4850MHz now with the changes it's high 4700MHz range like ~4770MHz. It doesn't seem bad but at the same time it's not the product I assumed I was buying, a rock solid 100% stable chip. On the Micro Center page the Warranty says 3 Years Labor 3 Years Parts. Man I'm so undecided I hope I reach a decision by then, XD

EDIT: Gonna run P95 all AVX disabled on core 5 for most of the day if not all and see what happens.

 

[B-Riz]

5800X, stock settings, ran the cycler for 4 iterations, highest clock was 4841 MHz, no "bad" cores.

 

[Hail The Brain Slug]

Now I'm a little undecided what to do with this chip. Seems to be stable with a positive 10 curve optimizer on core 5 as it passed ~5 iterations with a manual P95 setting. It seems to me this is exactly what AMD intended Curve Optimizer to do to hard limit bad core(s) before being sold, but now the public has access to it.

My 5800X in my gaming rig seems to be ok as it passed a few iterations of corecycler but none of the cores boost past 4800MHz with the majority of them in the low low 4700MHz range. Core 5 on the "bad" chip was boosting to 4850MHz now with the changes it's high 4700MHz range like ~4770MHz. It doesn't seem bad but at the same time it's not the product I assumed I was buying, a rock solid 100% stable chip. On the Micro Center page the Warranty says 3 Years Labor 3 Years Parts. Man I'm so undecided I hope I reach a decision by then, XD

EDIT: Gonna run P95 all AVX disabled on core 5 for most of the day if not all and see what happens.

If your best core(s) are the ones failing like mine it's clearly a failure of the validation process at AMD. The 4.9 GHz boost clock advertised on the 5950x is only achieved on my two best cores, and with curve optimizer set so they do not fail they no longer achieve the advertised boost.

It's clear this chiplet with 8 good cores and the two best cores capable of 4.8 GHz should have been a 5800x probably.

 

[BTRY B 529th FA BN]

@Justinus Yup I'd return it. I can't say the same with my main rig 5800X, all cores boost past the 4.7GHz advertised speed, and actually the "bad" core, core 5, is bouncing + / - 10MHz @4.8GHz in 76*f ambient temps which is still pretty good and if it passes most of the day today (started at 9:19am) It might be wise to just keep it. Who knows what I might end up with, yet it still taints the thought of the chip. And the more I think about it +10 CO on Core 5 to reduce it to ~ 4.8GHz IMO wouldn't take away from it's resale value. And honestly I don't think I'll sell this thing for another 10yrs.

Now let's see what this chip can do! *hopes for an all day stable run*

 

[BTRY B 529th FA BN]

My manual test of Core 5 with +10 CO completed 8960K to 32768K in 3-1/2hrs. I'm thinking it doesn't need to complete another full run.