Question FIVE dead Ryzen CPUs? Greg Salazar on YT.

[VirtualLarry]

Tech Yes City on YT recently did a vid about dead 3600 CPUs shpwing up.

Are we seeing results of "early aging" issues with Ryzen?

 

[In2Photos]

It would be interesting to see the serial numbers and manufacturer dates for these to see if there's any correlation.

 

[DAPUNISHER]

As Greg points out, these are viewers CPUs and there is no way to know their history. PEBKAC/PICNIC is always the prime suspect. Anything could have happened to them, including overclocking related damaged.

Many of the top comments offer good hypothesis concerning SoC being damaged by overvolting, and IHS issues. I have had a few boards that were concerningly aggressive OOB. By Zen+ I had taken to setting everything manually in the UEFI because of that.

My longest experience with a Ryzen CPU is a Ryzen 2600 that had seen heavy use for over 4yrs. It was still rock solid when I sold it.

 

[Ertaz]

Salazar is the last tech-tuber I would listen to.

 

[alcoholbob]

Salazar is the last tech-tuber I would listen to.

Not a fan of Florida Man I take it?

 

[A///]

Can't stand the wanker. Can't think of any troublesome Ryzen processors apart from 1st gen with the segfault bug.

 

[nurturedhate]

Are we really doing this again?

Question - KrisFix Germany video on 6800 and 6900 cards dying while running Adrenaline 22.11.2

Hey guys I ran across his video tonight and was wondering if anybody has encountered this issue with the latest whql adrenaline driver? I have an AIB 6900 XT and run it but haven't seen anything weird apart from the normal stuff like the card randomly drops signal to my 2nd UW monitor. He's...

forums.anandtech.com

 

[coercitiv]

Are we really doing this again?

Nah, we're doing this again!

Ryzen 5000 failure rates: We reality-check the claims

One system vendor reported that Ryzen 5000 chips are failing at a high rate. Our sources and others suggest the problem is isolated.

www.pcworld.com

 

[A///]

Nah, we're doing this again!

Ryzen 5000 failure rates: We reality-check the claims

One system vendor reported that Ryzen 5000 chips are failing at a high rate. Our sources and others suggest the problem is isolated.

www.pcworld.com

I would go further and question those very low failure rates and examine whether they were caused by autooc functions on motherboards. Ai tuning is another term for it.

 

[aigomorla]

Can't stand the wanker. Can't think of any troublesome Ryzen processors apart from 1st gen with the segfault bug.

Raises hand..
Ive lost 2 5600G in a span of less then 2 months...

Question - I have lost 2 x 5600g in a span of 2 months.

I swear there is something seriously wrong with the 5600G. I built both machines. CPU's came sealed. First cpu died in 3 weeks from my niece gaming on it probably 6 hours every day. Second cpu died from my Cousin watching movies a little under 2 months. Both systems were built near the same...

forums.anandtech.com

i thought i should go out and get a lottery ticket that night tho, seeing how i broke odds.
They were probably from the same batch, as i bought them from the same vendor.

But later after some googling, its not just me that lost 5600G. Something overall with that batch i think was FUBARD.

I did replace it to a 5600X which has been running flawlessly since tho, which makes me think something really was wrong with a batch of these cpus.

I would go further and question those very low failure rates and examine whether they were caused by autooc functions on motherboards. Ai tuning is another term for it.

And how would you go about doing that on a dead processor?
Without ripping the IHS and taking it under a microscope, there is no way to find out what killed the CPU.

Maybe AUTO voltages did kill the CPU, but again, that's not the end users fault.
The board makers would need to be addressed, but if it was the board makers, you would hear a LOT more CPU's failing them a small percentages, unlike if a batch of waffers were bad to begin with, it would be very localized, and the qty would be small.

AMD tho replaced one of them really quickly, and the other was sent back to AMAZON for an exchange to a 5600X.

I gave away the RMA'd 5600G to a friend, i haven't heard any issues from it so far, so i really think its a bad batch they had.

Many of the top comments offer good hypothesis concerning SoC being damaged by overvolting, and IHS issues. I have had a few boards that were concerningly aggressive OOB. By Zen+ I had taken to setting everything manually in the UEFI because of that.

Possibly, but then fault would lie on the board makers.
Board makers will never admit fault tho.
But again, the replacement i received and the 5600X i swapped out, has been running fine so far since the time i wrote that thread.

Im still leaning towards a bad batch, especially with how TSMC was probably overstocked with orders, and where i think they were more focused on pumping out silicon then QoC.

 

[DAPUNISHER]

It could be all of the above. Manufacturing issues from the silicon level to the IHS being installed. Boards being overly aggressive with voltages exposing such issues. PICNIC/PEBKAC particularly overclocking.

Most of us that have used Ryzens from OG Zen have not had any deal breaking issues. Which would lend itself to this not being a prevalent defect or whatever it is. Remembering buildzoid intentionally degrading his 3700x certainly doesn't get the users off the hook either. PICNIC and the DIY community go together like chocolate&peanut butter, peas&carrots, strippers&blow.

Another problem I have with this raising the red flag, is the two doing it are friends. Greg and Bryan of TYC both doing content on this, well, I'll let Mr. Krabs take in from here -

 

[Ranulf]

I don't know much about Salazar/SciStudio (I remembered his name change) but TYC seems legit overall.

 

[A///]

And how would you go about doing that on a dead processor?

Pop in a similar processor and view the bios settings. All aib are pushing stupid autooc aioc now that's constantly bumping the voltage up and down, thermal action causes more aggressive behavior. Can't say much about your 5600g though. You're maybe the 2nd person I've spoken to who had them.

 

[coercitiv]

but TYC seems legit overall.

I thought that as well, after seeing his video I'm not so sure anymore. The title of the video is This is Why Ryzen 5 3600s Are FAILING! yet the content in that video does not provide any serious evidence for either failure rate or failure reason. He even goes on to say that high 90C+ temps will likely lead to Ryzen 5000 failed chips, in spite of the fact that AMD themselves say 95C is typical and despite the fact that most Intel CPUs are currently configured to run close to 100C as well.

For the record, I think lowering the thermal limit in the 75-85C range on modern CPUs is a good idea for both AMD and Intel, but I encourage people to do this as part of a philosophy that limits pointless waste and potential risk while tinkering (undervolting, overclocking etc). It also happens to be the easisest way to configure power / cooling / noise balance. That being said, challenging both Intel and AMD and arguing that high temps are killing CPUs should be done with better methodology and journalistic investigation. Opening up with an all-caps "FAILING" in the title isn't helping at all, makes me think the entire piece was optimized for revenue instead.

 

[aigomorla]

Pop in a similar processor and view the bios settings. All aib are pushing stupid autooc aioc now that's constantly bumping the voltage up and down, thermal action causes more aggressive behavior. Can't say much about your 5600g though. You're maybe the 2nd person I've spoken to who had them.

You see i did that.
I replaced the first one right away with a 5600G for about 2 weeks ran fine, and then did a return on it to a 5600X because i wanted the PCI-E 4.0 and she needed a dedicated video card.

The other 5600G which got RMA'd was swapped out to my older 3600X and my cousin refused to give it back saying it felt snapper then the 5600G, which of course i also loaned / gave him a cheap dedicated card.

The replacement 5600G has been working in my friends system so far without complaint when i sold him that cpu.

The two faulty CPU's were bought at near time from AMAZON. I got them earily and was intending on holding onto them for office machines, until family members wanted them. The CPU's i believe were from the initial launch.

Im pretty sure they were most likely same BIN as well, or at the very least same FAB machine.
Both CPU's were on a B550 chipset, so it was a newer chipset.
Both systems are still running perfectly fine on the same B550.

This lead me to the conclusion, that i had a bad batch, or there is a bad batch of 5600G, as AMD immediately advance RMA'd me the moment i said i lost a 5600G and told them exactly what issues i had.

IF the issues were board side fault, there would of been bios revisions to address that issues but i do not see any critical updates.

I just think the fault lies at TSMC during that time, and probably one of there FAB machines.
They either probably validated the CPU very poorly, or probably didn't have enough time to validate it properly, and hence it sneaked by the QC, and made it to circulation.

 

[Leeea]

You're maybe the 2nd person I've spoken to who had them.

I made a 5600g system for a friend, it is all working very nicely five months later.

 

[DAPUNISHER]

I made a 5600g system for a friend, it is all working very nicely five months later.

I have had a 5700G for well over a year now. PBO with +200MHz boost - iGPU@2400MHz and with 32GB of 3600MT/s - SAM enabled, I just got a legendary score in 3DMark Night Raid. BTW with 32GB on the ASRock Fatal1ty B450 GAMING-ITX/AC, it auto allocates it as a 4GB GPU. Which is good, because that setting is missing in the UEFI.

 

[A///]

You see i did that.
I replaced the first one right away with a 5600G for about 2 weeks ran fine, and then did a return on it to a 5600X because i wanted the PCI-E 4.0 and she needed a dedicated video card.

The other 5600G which got RMA'd was swapped out to my older 3600X and my cousin refused to give it back saying it felt snapper then the 5600G, which of course i also loaned / gave him a cheap dedicated card.

The replacement 5600G has been working in my friends system so far without complaint when i sold him that cpu.

The two faulty CPU's were bought at near time from AMAZON. I got them earily and was intending on holding onto them for office machines, until family members wanted them. The CPU's i believe were from the initial launch.

Im pretty sure they were most likely same BIN as well, or at the very least same FAB machine.
Both CPU's were on a B550 chipset, so it was a newer chipset.
Both systems are still running perfectly fine on the same B550.

This lead me to the conclusion, that i had a bad batch, or there is a bad batch of 5600G, as AMD immediately advance RMA'd me the moment i said i lost a 5600G and told them exactly what issues i had.

IF the issues were board side fault, there would of been bios revisions to address that issues but i do not see any critical updates.

I just think the fault lies at TSMC during that time, and probably one of there FAB machines.
They either probably validated the CPU very poorly, or probably didn't have enough time to validate it properly, and hence it sneaked by the QC, and made it to circulation.

I usually tell people to avoid launch period hardware because even if it passes the smell test it may not work out. It's a recommendation spawned from days old where if you didn't say a few prayers your hardware would catch on fire if you looked at ir wrong.

 

[blckgrffn]

The last two CPUs I have gotten from eBay have been dead.

A 12500t and a 3700X.

After years of getting fully functional and cheap cpus I am a bit miffed. I mean, they don't even have pins any more!

 

[Ranulf]

I usually tell people to avoid launch period hardware because even if it passes the smell test it may not work out. It's a recommendation spawned from days old where if you didn't say a few prayers your hardware would catch on fire if you looked at ir wrong.

Ah, the days of ye old raid arrays on consumer boards.

 

[DAPUNISHER]

The last two CPUs I have gotten from eBay have been dead.

A 12500t and a 3700X.

After years of getting fully functional and cheap cpus I am a bit miffed. I mean, they don't even have pins any more!

Last 3 PC components I bought from Ebay were all defective or DOA. And 4 of the last 6. I am over the used market, outside of forums. Not worth my time and effort to go through the return process. Never gotten anything that needed more than TLC, from tech forums.

 

[blckgrffn]

Last 3 PC components I bought from Ebay were all defective or DOA. And 4 of the last 6. I am over the used market, outside of forums. Not worth my time and effort to go through the return process. Never gotten anything that needed more than TLC, from tech forums.

FWIW, I’ve had zero issues with eBay return policy. Even items marked no returns - eBay approved it and refunded me when the package was delivered. That’s the only reason I’ll keep shopping there, but it’s been a bad run after years of zero issues for me.

I had dm’d you that guy who just put the RX5500 in a big box - not even a static bag! - and it was dead too. Pretty sure this is a user problem in many cases. The new/opened 2060 12GB I bought was plausibly new with all plastic intact, but the dude shipped it in just the EVGA box. 18 months ago zero percent chance that just gets delivered without being in another box 😂

 

[Shmee]

Yikes. I have been pretty lucky with ebay parts lately. My 5950X was from there, as well as 8 Samsung ECC RDIMMs, and 2 DIMMs of Patriot Viper that I added to another PC. My two Xeon X5660s and my Xeon E5-1660V3 as well. The Vega 64 Nitro I had and later traded for a 5700XT was from ebay, as well as two of my Sapphire 290 Tri X cards. All these worked fine for me.

 

[DAPUNISHER]

Yikes. I have been pretty lucky with ebay parts lately. My 5950X was from there, as well as 8 Samsung ECC RDIMMs, and 2 DIMMs of Patriot Viper that I added to another PC. My two Xeon X5660s and my Xeon E5-1660V3 as well. The Vega 64 Nitro I had and later traded for a 5700XT was from ebay, as well as two of my Sapphire 290 Tri X cards. All these worked fine for me.

Until the last 6 months or so, I'd had one or two bad used items in 23yrs of using Ebay. I have gotten a DOA PSU, DOA RAM twice, and a defective RTX 3060. No more used parts on Ebay for me for a good long while. Doesn't matter how good the return process is, I don't want to go through it anymore.

On the other hand, I've bought used Ryzens from other members here, and all have been great. Still using the 3700X and FX8350 I bought from Larry. Come to think of it, I think I bought the 5700G from him too. Stop selling me stuff Larry!

 

[aigomorla]

I usually tell people to avoid launch period hardware because even if it passes the smell test it may not work out. It's a recommendation spawned from days old where if you didn't say a few prayers your hardware would catch on fire if you looked at ir wrong.

But i made sure to pray to all the machines gods required and demanded by the order of Adeptus Mechanicus.