AMD & Data Corruption

[Chetar Ruby]

While I welcome advice, tips or troubleshooting guiding, I'm retiring this PC.  I currently have a BioStar X370GT5 mobo with AMD Ryzen 7 1800X and 16GB DDR4-3200 DRAM.  Some background:  I'm a computer repair technician, been in the field for approximately 30 years.  My first IBM PC compatible machine was an 10mhz 8088 running MSDOS.  I have since owned 286, 386, 486, Pentium, Pentium2, Pentium3, Core2, i7, and Xeon machines.  I believe I experimented some with AMD's K6 line back in the day.

 

I've had the machine for about a year.  It seemed to work pretty decently, no tangible complaints until now.  Over this weekend, I decided it's time to refresh my long term storage drives and started copying files with this computer.  Typically USB<->SATA drive copying.  But the strange thing was/is, every time I completed a copy and it's a large dataset, about 2TB, then I'd run a compare to make sure it copied correctly, it would fail compare.  At a fairly random rate.  Recopying the 'failed to copy' data results in more corruption to other files it seems, as subsequent compares reveal more corruption to the data, but different files.  Most of the data is in archives (7z, rar, tar, zip, etc), so I can run archivers in test mode to verify the copy.  Many random failures, with no rhyme or reason, just sprinkles of data corruption all over.

 

I've basically determined this computer cannot reliably copy anything, between any two devices, be it NVME, SATA or USB.  I don't know if it's memory issue or what.  I even downspeeded the memory to the CPU's rating of 2667mhz.  No overclocking of any kind being done here.  I do want to note, running memtest86 from the Linux Mint 20.1 boot USB crashes the machine every time, with or without SMP enabled.

 

Why am I posting?  As I said, not looking for tips.  I no longer trust this computer, or AMD, with my data.  I'm posting mostly as a warning:  AMD still makes unreliable equipment.  I experienced their shoddy offerings back in the 486 clone days and swore off AMD then.  But I foolishly decided to check out their Ryzen series and got myself (what I thought was) a reasonably nice Ryzen box.  And my reward is unknown amounts of data corruption.  Feel free to comment things to try, or suspected causes, but I'm still retiring the machine, bringing my Xeon E5-2667 v2 back out of the closet.

 

So you've been warned.

[Fasauceome]

2 minutes ago, Chetar Ruby said:

Why am I posting?  As I said, not looking for tips.  I no longer trust this computer, or AMD, with my data.  I'm posting mostly as a warning:  AMD still makes unreliable equipment.  I experienced their shoddy offerings back in the 486 clone days and swore off AMD then.  But I foolishly decided to check out their Ryzen series and got myself (what I thought was) a reasonably nice Ryzen box.  And my reward is unknown amounts of data corruption.  Feel free to comment things to try, or suspected causes, but I'm still retiring the machine, bringing my Xeon E5-2667 v2 back out of the closet.

 

so, you've built a single computer and determined that every product that AMD makes has a risk of this happening?

 

this is a warrantless "warning." You have a computer with an extremely obscure issue, and you're gonna claim that the entire brand is to blame? What about the hundreds of thousands of users with AMD PCs where this isn't an issue?

[Chetar Ruby]

1 minute ago, Fasauceome said:

so, you've built a single computer and determined that every product that AMD makes has a risk of this happening?

 

this is a warrantless "warning." You have a computer with an extremely obscure issue, and you're gonna claim that the entire brand is to blame? What about the hundreds of thousands of users with AMD PCs where this isn't an issue?

Yes.  I only need one massive data loss to write off a company as 'bad.'  Call me strange?

[Fasauceome]

4 minutes ago, Chetar Ruby said:

I only need one massive data loss to write off a company as 'bad.'  Call me strange?

extremely strange. how do you know it's not the motherboard? the ram? damage you caused yourself by accident? Why isn't Biostar to blame here?

 

it's not your wary nature that's the issue, it's the self assurance that it's AMD's fault

[AlwaysFSX]

Of course it's not your drives or controller at all that could have any issues, no. I've copied many, many terabytes on my system without any issues of data corruption. It isn't AMD.

[WhitetailAni]

5 minutes ago, Chetar Ruby said:

Yes.  I only need one massive data loss to write off a company as 'bad.'  Call me strange?

Let me get this straight.

Because you - probably one of a few people - to have this problem, you believe that everyone else has a high chance to have this problem.

I think it might be your motherboard. It's using a white socket - those haven't been used on mainstream boards since AM3, it has PCI slots, and no listing on Newegg.

It doesn't look too great for an AMD X board.

 

And as for me?
I've copied maybe 15TB of data on my machine. Was any corrupted? No.

I've done many, many, Windows reinstalls, sure.

But that was my doing.

[ProjectBox153]

8 minutes ago, Chetar Ruby said:

Yes.  I only need one massive data loss to write off a company as 'bad.'  Call me strange?

And how have you determined that AMD is to blame here? Yes, it is extremely strange. 

 

12 minutes ago, Chetar Ruby said:

I do want to note, running memtest86 from the Linux Mint 20.1 boot USB crashes the machine every time, with or without SMP enabled.

And that didn't concern you or point you toward the issue?! This baffles me. If the RAM seems to be having issues then why blame AMD? The CPU doesn't randomly decide to corrupt your data. 

There are simply too many factors here to point blame at one component. Period. If I had to guess it would be the memory, but I'm sure that's not going to change your mind in any way. It simply must be the CPU, and by extension AMD as a whole. 

[Sakkura]

15 minutes ago, Chetar Ruby said:

I've basically determined this computer cannot reliably copy anything, between any two devices, be it NVME, SATA or USB.  I don't know if it's memory issue or what.  I even downspeeded the memory to the CPU's rating of 2667mhz.  No overclocking of any kind being done here.  I do want to note, running memtest86 from the Linux Mint 20.1 boot USB crashes the machine every time, with or without SMP enabled.

It's most likely a memory issue then. AMD doesn't really make memory. They do make the memory controller in the CPU, but not the BIOS code that controls things. So basically three possible culprits, only one of which is AMD.

[Chetar Ruby]

Bottom line is, I've been using Intel for 30 years.  This dataset is about 10 years old, it's survived being copied with Intel machines over and over again, running Linux or Windows.

 

And low and behold, I get an AMD machine and it trashes all my data.  I'm not just drawing on this experience, I'm adding in my experience with the K6's being garbage.

 

Look, AMD machine did this.  It could be the RAM, or could be mobo, or it could be the CPU, or any combination of the 3.  The point is.. I've never seen an Intel machine do this, ever.  They either work, or they don't.  Where AMD stuff tends to mostly work, but there's edge cases, every time.  That was the K6 and apparently, it's Ryzen too.

 

And there's is no way you're blaming the drives, I got about a dozen different drives here, doesn't matter which ones I use, it keeps trashing my data, randomly, unpredictably.  That's a HUGE HUGE no-no in computing.

 

Take the warning for what it's worth.  Watch your data.  Or don't.  I won't use that stuff anymore.  Use my experience, learn from it, or don't.

[ProjectBox153]

Just now, Chetar Ruby said:

Bottom line is, I've been using Intel for 20 years.  This dataset is about 10 years old, it's survived being copied with Intel machines over and over again, running Linux or Windows.

 

And low and behold, I get an AMD machine and it trashes all my data.  I'm not just drawing on this experience, I'm adding in my experience with the K6's being garbage.

 

Look, AMD machine did this.  It could be the RAM, or could be mobo, or it could be the CPU, or any combination of the 3.  The point is.. I've never seen an Intel machine do this, ever.  They either work, or they don't.  Where AMD stuff tends to mostly work, but there's edge cases, every time.  That was the K6 and apparently, it's Ryzen too.

 

And there's is no way you're blaming the drives, I got about a dozen different drives here, doesn't matter which ones I use, it keeps trashing my data, randomly, unpredictably.  That's a HUGE HUGE no-no in computing.

 

Take the warning for what it's worth.  Watch your data.  Or don't.  I won't use that stuff anymore.  Use my experience, learn from it, or don't.

You aren't listening to what people are saying though. You yourself said in your first post that a memory test always failed. That's literally pointing you straight to the most likely cause of all of this, but no, blame AMD. I don't understand that sort of thinking at all. You have an AMD Ryzen machine that seems to have bad memory. Why does bad memory lead you to blame the CPU?

[Chetar Ruby]

2 minutes ago, ProjectBox153 said:

You aren't listening to what people are saying though. You yourself said in your first post that a memory test always failed. That's literally pointing you straight to the most likely cause of all of this, but no, blame AMD. I don't understand that sort of thinking at all. You have an AMD Ryzen machine that seems to have bad memory. Why does bad memory lead you to blame the CPU?

Enh.  Memtest86 isn't the best judge of memory chips, when it just crashes out of the gate.  I've seen that many many times.  It's not all that unusual.  It's just an old piece of software and doesn't like some machines.  I've seen many intel machines lock up when ever you run memtest86 in SMP, but run fine in single core.  I don't really put a lot of stock in memtest86 when it fails to run at all.

 

I did run the Windows built-in memory test and it passed, so I dunno.  I'm still blaming AMD, it's their CPU, so the rest of the machine is built to accommodate it.

[Mark Kaine]

I'm sorry but you should have stopped copying stuff after you noticed something wasn't right, sounds like you just kept going? 

 

 

[Chetar Ruby]

Just now, Mark Kaine said:

I'm sorry but you should have stopped copying stuff after you noticed something wasn't riggt, sounds like you just kept going? 

 

 

You're right.  I was changing my tools, suspecting software issues, or anything other than the hardware.. I didn't want the blame the hardware!  But after many failed attempts, with different drives and different programs, I have to blame the hardware now.

[ProjectBox153]

2 minutes ago, Chetar Ruby said:

You're right.  I was changing my tools, suspecting software issues, or anything other than the hardware.. I didn't want the blame the hardware!  But after many failed attempts, with different drives and different programs, I have to blame the hardware now.

And how did you arrive to the conclusion that the AMD CPU must be to blame? There are many pieces to a computer that would be involved in the process of copying data, not just the CPU. 

[Fasauceome]

15 minutes ago, Chetar Ruby said:

  The point is.. I've never seen an Intel machine do this, ever.  

 

I've never seen any system do this, ever. Your experience in the field is worthless if you take a single occurrence as validation to declare an entire brand suspect or worthless. Look at how AMD's presence has skyrocketed in the data center, all thanks to their Epyc line of processors. Still think Ryzen isn't safe? Well the industry just isn't with you on that.

 

Why not pop some different memory in this system before writing off an entire company as "dangerous" or "unsafe" for a user's data? do any amount of testing that can validate your claim. Clearly you're upset that you lost your data so you're irrationally taking out your anger on the nearest target, which is the company that produced the CPU. But I bet if the exact same thing happened on an Intel system you wouldn't think twice.

 

I'm not a fanboy of AMD, but a user trying to write off a perfectly valid option for a home computer just because of the rarest tech issue I've ever heard of is utterly senseless.

[AlwaysFSX]

3 minutes ago, Fasauceome said:

Why not pop some different memory in this system before writing off an entire company as "dangerous" or "unsafe" for a user's data? do any amount of testing that can validate your claim.

Imagine stability testing something before using it for work?

[Chetar Ruby]

1 minute ago, Fasauceome said:

I've never seen any system do this, ever. Your experience in the field is worthless if you take a single occurrence as validation to declare an entire brand suspect or worthless. Look at how AMD's presence has skyrocketed in the data center, all thanks to their Epyc line of processors. Still think Ryzen isn't safe? Well the industry just isn't with you on that.

 

Why not pop some different memory in this system before writing off an entire company as "dangerous" or "unsafe" for a user's data? do any amount of testing that can validate your claim. Clearly you're upset that you lost your data so you're irrationally taking out your anger on the nearest target, which is the company that produced the CPU. But I bet if the exact same thing happened on an Intel system you wouldn't think twice.

Because I'm not interested in that.  It's failed.  There is something wrong with it.  I don't know what, but the only way I can be sure is to write off the entire thing.  My data is priceless.  I'm exceedingly upset about the damage and time wasted.  I'm not going to give it another chance.  It's an AMD based system, that's all that matters to me.

 

Personally?  I don't think there's anything wrong with it.  I think this is just what it does.  Maybe it has a wonky SATA controller, or USB controller, I have no idea and I'm done fiddling with it.  I've already wasted 3 days shuffling data around only to have it get corrupted by THIS machine.  THIS AMD MACHINE.

 

I'm not willing to spend any more time with that manufacturer's equipment.  I've wasted enough and I've seen enough to know it's not for me.  Works for you?  Fantastic.  I'm happy for you and the millions of others who have good luck with this stuff.  For my use-case, it's not working.  It's wasting my time, and wrecking my data.  If it was an intel machine, I'd kick the whole thing to curb just like I'm doing with this AMD machine.

 

I'm just here to share my experience.  Not looking for a solution.  I have one:  Xeon E5-2667 v2.  Use my experience however you see fit.

[WhitetailAni]

1 minute ago, Chetar Ruby said:

Because I'm not interested in that.  It's failed.  There is something wrong with it.  I don't know what, but the only way I can be sure is to write off the entire thing.  My data is priceless.  I'm exceedingly upset about the damage and time wasted.  I'm not going to give it another chance.  It's an AMD based system, that's all that matters to me.

 

Personally?  I don't think there's anything wrong with it.  I think this is just what it does.  Maybe it has a wonky SATA controller, or USB controller, I have no idea and I'm done fiddling with it.  I've already wasted 3 days shuffling data around only to have it get corrupted by THIS machine.  THIS AMD MACHINE.

 

I'm not willing to spend any more time with that manufacturer's equipment.  I've wasted enough and I've seen enough to know it's not for me.  Works for you?  Fantastic.  I'm happy for you and the millions of others who have good luck with this stuff.  For my use-case, it's not working.  It's wasting my time, and wrecking my data.  If it was an intel machine, I'd kick the whole thing to curb just like I'm doing with this AMD machine.

 

I'm just here to share my experience.  Not looking for a solution.  I have one:  Xeon E5-2667 v2.  Use my experience however you see fit.

I am personally of the opinion that it's your motherboard.

I wouldn't trust it if it was given to me for free.

[Mark Kaine]

16 minutes ago, Chetar Ruby said:

You're right.  I was changing my tools, suspecting software issues, or anything other than the hardware.. I didn't want the blame the hardware!  But after many failed attempts, with different drives and different programs, I have to blame the hardware now.

Well idk how to fix the corrupted data, although copying implies you still have the originals... 

 

 

However I can try to help you understand the situation. 

 

Yes. This is very very likely a hardware issue. Indeed. 

 

And you know it. It's your RAM, and you made a mistake thinking memtest86 isn't reliable or something - - we're doing a lot of troubleshooting here and the very clear action for faulty RAM, or rather memtest errors is to replace the RAM asap... and indeed it's often an issue (tho not that often) 

 

So if anything you should blame the RAM manufacturer, or indeed yourself because you didn't do the right thing and replace the RAM asap... this isn't and Intel or AMD thing at all, it happens regardless, the RAM is even compatible between the too! 

 

I mean it's like you think we care lol, but I would advise you to just buy the cpu that's better bang for the buck (which is currently that 10c/20t Intel CPU, but has been pretty much anything RYZEN the last couple of years, and Intel only got going because of the tremendous success of AMD Ryzen CPU's, otherwise they would probably still make 4c / no threads chips and sell them for 1k a piece lol) 

 

 

[Fasauceome]

1 minute ago, Chetar Ruby said:

I'm not willing to spend any more time with that manufacturer's equipment. 

But you are willing to waste time arguing with us forum users about how you're definitely right about this and AMD is definitely not to be trusted

 

If you're simply tired of the issue, don't come to the forum and continue to bring it up. Get on with your life. But the reason we keep posting? It's because this is a discussion forum.

 

30 years in the field huh? I'm sure you still groan every time someone says "an AMD graphics card works faster with an AMD CPU right?" It's bullshit but they don't know that, and they think it's true cause "some guy once said it." And that's you, you're that "one guy" that said AMD is less secure for your data, and even though you're the ONLY ONE who says this, some hapless, tech illiterate person will come along and just believe it. You're literally just spreading misinformation based on the most simple of occurrences.

[ProjectBox153]

6 minutes ago, Chetar Ruby said:

If it was an intel machine, I'd kick the whole thing to curb just like I'm doing with this AMD machine.

You say that, but something tells me that you wouldn't be willing to ditch Intel too and immediately blame them. You'd actually figure out what the actual issue is. For some reason you're just not willing to do that for the AMD system. 

[PlayStation 2]

I don't really get what you want from us. You already made up your mind on a fringe case that could have many variables to it just to blame it on the CPU choice, rather than experimenting with what and why this issue is happening. 

Threads like this, where you want validation for a bad conclusion rather than figuring out what's wrong always piss me off. They never give solid questions, they never bring solid answers. All they bring are stupid opinions written by someone with more time than knowledge.

[SansVarnic]

1 hour ago, Chetar Ruby said:

Yes.  I only need one massive data loss to write off a company as 'bad.'  Call me strange?

This is a massively flawed argument. FYI.

 

1 hour ago, Chetar Ruby said:

I'm posting mostly as a warning:

OP if you're not wanting to Troubleshoot your PC then this topic is unnecessary and in the wrong place.

If your wanting advice on building a new PC, please create a new topic here https://linustechtips.com/forum/18-new-builds-and-planning/.

 

And as for this topic?

 

-= Topic Locked =-

If you wish it unlocked please PM me and explain why.