Ryzen issues with Linux

Yeah, this has me stumped...if there were any kernel logs, that may help, but I'm not expecting anything obvious.

Thinking RMA is the way to go. Could be motherboard, CPU, or PSU, I'm leaning toward motherboard (and I'm not sure a replacement would fix the problem), but if they offer free shipping and you have the spare time it's worth a shot I guess (cpu or mb).
 
I really doubt a PSU has anything to do with this, I've used four different PSUs, two of them new out of box.

I did get a new error during the compile when mucking around with kernel flags:

Kernel: show_signal_msg: 13 callbacks suppressed
Kernel: bash[19407]: segfault at ffffffffffffffee ip 000055f3b6410382 sp 00007fff210be470 error 7 in bash[55f3b63a1000+104000]
 
Your Ryzen is affected by the segfault bug, so independent of whether the mobo is defective I would advise you to RMA the CPU. You don't have to pay for shipping, but might be a couple weeks (depending on where you live) without CPU as AMD offers no cross shipping.
 
Finally got the replacement CPU from AMD. The RMA process less than stellar, but I just got back a replacement CPU.

I just finished testing it and it has the exact same fucking problem. kill-ryzen.sh crashes and the system has random hard locks.

Time to trash this garbage and go back to Intel, this is what I get for giving AMD a chance after all these years.
 
Last edited:
I've tried 3 different BIOS revisions (F22, F23 and F23d) which all have the same issue. Other than that, I can't verify if the board is the issue because I don't have another AM4 board, and Gigabyte is refusing to RMA this one because Linux is an unsupported OS in their eyes. The latest BIOS available for this board only has AGESA version 1.0.0.4, when the latest is like 1.0.0.6b or something.

I don't have an easy way to get another AM4 board.
 
Try Fedora stable instead of ubuntu. I hear that ts better on those boards Linux and those boards are very iffy. For Linux I would stay away from Giga for now,

my current stable fedora kernel .17.19-200.fc28.x86_64

the 200 kernel is pretty good.

Linux laptop 4.17.19-200.fc28.x86_64 #1 SMP Fri Aug 24 15:47:41 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
 
I installed Ubuntu at the direction of other people to test the kill-ryzen script, I WAS running Fedora 28 before and it had the same problem.

The kill-ryzen script would not run at all on Fedora because of missing dependencies.
 
what was your mem speed with kill ryzen? should be run at 2133mhz only. also gcc 5.4 or 6.3 .

other than the killryzen errors is there anything else that points to issues with the new cpu?
 
i didnt mention init3 to fix dependencies. it is the way to test without loading x and other shit to better isolate the kernel. sorry if i didnt catch everything in your thread. im trying to help but if your start getting aggressive - you will end up eating your build. i would rely on seg faults to determine errors and not only the script.
 
actually, not init3 - use single user mode. can you figure that out? i can be aggressive too. see how that works?
 
GiGaBiTe said:
Please see yourself out, your input is no longer needed here.
Click to expand...

with pleasure.. if you cant even figure out Linux dependencies -and debugging kernel issues at runlevel 5, in an xterm - you need more help than I can bother to provide. Funny how people come beg for help and then when you try to help they act like they know it all. Well if you know it all, you wouldnt be here asking for help, would you? Plus on on top of that, they second guess your suggested help on invalid presumptions on your line of thought.

So, what is the incentive to help you? I have NONE. You seem to forget that, Too bad, because that makes me less inclined to help other people on this board that dont have a shit attitude and are truly deserving of help.
 
Last edited:
Mega6 said:
Funny how people come beg for help and then when you try to help they act like they know it all.
Click to expand...

What's really funny is how much of a pompous ass you are. You come in here asking questions which have already been answered by me, in two cases multiple times; Then you go on a tirade of spinning hilarious straw man arguments attacking my technical knowledge because you're too self-important to read the thread and must be spoon fed information.

Your ego is too big to be of any help to anyone really, I don't need to give you some incentive to give helpful information.

By the way, your idea of running in single user mode or runlevel 3 didn't do anything, the script still failed and still getting random segfaults and hard locks.
 
Oh well, we tried. Sometimes shit just don't work out. :/
 
GiGaBiTe said:
What's really funny is how much of a pompous ass you are. You come in here asking questions which have already been answered by me, in two cases multiple times; Then you go on a tirade of spinning hilarious straw man arguments attacking my technical knowledge because you're too self-important to read the thread and must be spoon fed information.

Your ego is too big to be of any help to anyone really, I don't need to give you some incentive to give helpful information.

By the way, your idea of running in single user mode or runlevel 3 didn't do anything, the script still failed and still getting random segfaults and hard locks.
Click to expand...
You did ask for help and I see he actually spent quite a bit of time trying to help you out. He certainly didn't deserve a response like that as thanks for the effort he gave you - even if you feel it was less than you wanted.
 
Nobu said:
Oh well, we tried. Sometimes shit just don't work out. :/
Click to expand...

Doing more research, it seems like it's a Gigabyte specific BIOS issue. After smashing together a bunch of key words on Google, I eventually opened the flood gates on bunches of people complaining about buggy BIOS revisions in regards to AGESA versions and other features not working properly. There's actually a guy on another forum that has made BIOS mods for several GB boards with 3x0 chipsets to fix a lot of the issues, unfortunately mine isn't one of them.

The board I have (AB350M-DS3H) ends at AGESA version 1.0.0.4 with BIOS F23, when the latest is something like 1.0.0.6b and is supposed to fix a bunch of the problems. GB doesn't regard Linux being broken as a problem and to them as long as Windows 10 works, they don't give a flying toss.

Be this a warning to anyone wanting to run Linux on a 3x0 chipset Gigabyte board, you're going to have a bad time. You need to find a board with a confirmed BIOS revision that has AGESA 1.0.0.6b or above, which I think I have found and will try buying one later this week.

cyberguyz said:
You did ask for help and I see he actually spent quite a bit of time trying to help you out. He certainly didn't deserve a response like that as thanks for the effort he gave you - even if you feel it was less than you wanted.
Click to expand...

Not really sure we're reading the same thing? Nobu and several other people in the thread were of great assistance and I appreciate that. Mega6 only showed up after I got the RMA'd CPU and spent most of his time being a pompous ass and spouting straw man attacks. I wouldn't exactly call that effort.
 
GiGaBiTe said:
Doing more research, it seems like it's a Gigabyte specific BIOS issue. After smashing together a bunch of key words on Google, I eventually opened the flood gates on bunches of people complaining about buggy BIOS revisions in regards to AGESA versions and other features not working properly. There's actually a guy on another forum that has made BIOS mods for several GB boards with 3x0 chipsets to fix a lot of the issues, unfortunately mine isn't one of them.

The board I have (AB350M-DS3H) ends at AGESA version 1.0.0.4 with BIOS F23, when the latest is something like 1.0.0.6b and is supposed to fix a bunch of the problems. GB doesn't regard Linux being broken as a problem and to them as long as Windows 10 works, they don't give a flying toss.

Be this a warning to anyone wanting to run Linux on a 3x0 chipset Gigabyte board, you're going to have a bad time. You need to find a board with a confirmed BIOS revision that has AGESA 1.0.0.6b or above, which I think I have found and will try buying one later this week.
Click to expand...
Ah, that explains why I haven't seen this. My board (x370 k5) is on 1.0.0.6 iirc, either that or 1.0.0.1 (they reset the version numbers with the release of the ryzen apus, for whatever reason).
 
Yeah the version nonsense threw me off. One of the older BIOS revisions had 1.0.7.2 IIRC which made no sense until I went digging in the AGESA version history.

While I despise ASUS, it looks like I'll have to end up getting one of their boards because it's the only one I can verify the AGESA version on, and there are a few reports of Linux working properly.
 
GiGaBiTe said:
I just finished testing it and it has the exact same fucking problem. kill-ryzen.sh crashes and the system has random hard locks.
Click to expand...

Have you tried testing with super boring jedec spec Crucial DDR4-2133 CL15 ram? I have a set of that ram, because the first fancy ram I bought just wouldn't fscking work reliably. I also bought some DDR4-3600 CL15 that I'll be throwing in the trash since it's past return date, and doesn't seem to work no matter what settings or how slow it runs. Eventually the system reboots or gets corruption. Working ram is the crazy expensive 2x16GB Trident Z DDR4-3200 CL14 stuff.
 
xorbe said:
Have you tried testing with super boring jedec spec Crucial DDR4-2133 CL15 ram? I have a set of that ram, because the first fancy ram I bought just wouldn't fscking work reliably. I also bought some DDR4-3600 CL15 that I'll be throwing in the trash since it's past return date, and doesn't seem to work no matter what settings or how slow it runs. Eventually the system reboots or gets corruption. Working ram is the crazy expensive 2x16GB Trident Z DDR4-3200 CL14 stuff.
Click to expand...
Don't toss it! Most ram is lifetime warranted, you could sell it as is, or RMA it and sell what you get back.
 
Yeah, not an AMD problem per se - except, it is somewhat - in a way - it's restricted or limited, though, to the first generation of Ryzen - but mostly BIOS /chipset related, I think.

Posted on August 24, here: "Also the new 2000 series Ryzen cpus seem fine from what I've heard."

The X470 don't have too many issues although there are a few - I found - from a google search about it on Linux.

But, the biggest culprit is Gigabyte - researching that, there's a lot of complaints and critiques about it. I wouldn't go with Gigabyte on a Ryzen platform. The issue is that Gigabyte owners have concerns and complaints but the response is too slow or unsatisfactory.
 
I am finally ready to go into production with my X470 ECC system. Replacing the WDC Black G1 5XX GB NVMe SSD with a Samsung 960 1TB seems to have fixed my problems.
 
xorbe said:
Have you tried testing with super boring jedec spec Crucial DDR4-2133 CL15 ram? I have a set of that ram, because the first fancy ram I bought just wouldn't fscking work reliably. I also bought some DDR4-3600 CL15 that I'll be throwing in the trash since it's past return date, and doesn't seem to work no matter what settings or how slow it runs. Eventually the system reboots or gets corruption. Working ram is the crazy expensive 2x16GB Trident Z DDR4-3200 CL14 stuff.
Click to expand...

Yeah, I have two generic sticks of DDR4-2133, one from Altex and another is even more generic. I get same result with both. Take a look at the hilarious video in the OP of the first pair of Patriot sticks I got running Memtest.

The Patriot stuff is 2666, but I've tried bog standard 2133 timings as well.

Nobu said:
Don't toss it! Most ram is lifetime warranted, you could sell it as is, or RMA it and sell what you get back.
Click to expand...

I might end up doing that. I just got back from Frys with an Asus Prime B450M-A, so I'm about to tear the machine apart to swap boards. If this doesn't work, I'm going to likely get rid of everything but the case, PSU and drives.

pavel said:
But, the biggest culprit is Gigabyte - researching that, there's a lot of complaints and critiques about it. I wouldn't go with Gigabyte on a Ryzen platform. The issue is that Gigabyte owners have concerns and complaints but the response is too slow or unsatisfactory.
Click to expand...

My pains are a result of blindly going with Gigabyte. I've had literally dozens of their motherboards in the past and only ever had one issue with an EP45-DS3L which I literally blame on cosmic rays. The board had dual BIOS and developed a problem where it would randomly hard reboot and go into an infinite boot loop, which ended up killing one of my hard drives. After two RMAs (which GB said there was no issue) I decided as a last ditch effort to reflash the latest BIOS which was already on it, which wiped out both BIOS chips and put two new copies on. The board never had a problem after that and works to this day. I figured since Ryzen had already been out a year that most of the bugs had been worked out, apparently not.
 
  • Like
Reactions: pavel
like this
drescherjm said:
I am finally ready to go into production with my X470 ECC system.
Click to expand...

Might I ask what memory you ended up using? and mobo? I've been running non-ECC, and it seems OK, but after 9 years of ECC this business of running computers non-ECC is a bit like walking down the street with no trousers on.
 
Definitely looking forward to hearing if the board replacement resolves the issue for you...
 
GiGaBiTe said:
Yjust got back from Frys with an Asus Prime B450M-A
Click to expand...

I have that motherboard + rma'd 1500X that works fine (oops wait, I have the B350M-A). Also have Asus Prime X470-Pro + 2700X working well.
 
GiGaBiTe said:
Yeah, I have two generic sticks of DDR4-2133, one from Altex and another is even more generic. I get same result with both. Take a look at the hilarious video in the OP of the first pair of Patriot sticks I got running Memtest.

The Patriot stuff is 2666, but I've tried bog standard 2133 timings as well.


I might end up doing that. I just got back from Frys with an Asus Prime B450M-A, so I'm about to tear the machine apart to swap boards. If this doesn't work, I'm going to likely get rid of everything but the case, PSU and drives.

My pains are a result of blindly going with Gigabyte. I've had literally dozens of their motherboards in the past and only ever had one issue with an EP45-DS3L which I literally blame on cosmic rays. The board had dual BIOS and developed a problem where it would randomly hard reboot and go into an infinite boot loop, which ended up killing one of my hard drives. After two RMAs (which GB said there was no issue) I decided as a last ditch effort to reflash the latest BIOS which was already on it, which wiped out both BIOS chips and put two new copies on. The board never had a problem after that and works to this day. I figured since Ryzen had already been out a year that most of the bugs had been worked out, apparently not.
Click to expand...
I had a P35 Gigabyte board - I think the EP35-DS3R - does that sound right? There was nothing wrong with it but I bought an Asus P45 P5Q (used?) and sold the Gigabyte mobo on ebay. Nothing was wrong with it but the Asus mobo was a good price. I just swapped the cpu in and I don't know if I gained much.

But, a recent google search and further reading, I discovered a lot of complaints with Gigabyte and AMD Ryzen hardware. I don't think you should feel bad at all because who would know? I was considering Gigabyte until I read all that. Found it on multiple sites and sources so it's not just some random, one-time or rare thing. They need to get their act together.

I suspect you should be okay - I think the B450 and X470 mobos - especially if other brands - are usually better bets with less issues and problems. Although, I think the 1st generation of Ryzen hardware still holds some issues regardless of brand - but, more likely to encounter it with Gigabyte. I suspect, a lot of it is BIOS-related, too.
 
pavel said:
I had a P35 Gigabyte board - I think the EP35-DS3R - does that sound right?
Click to expand...

Yeah, that's a model of their board. I had an EP35-DS3L with an E6750 (still have it actually) and never had any issues with it either.

pavel said:
They need to get their act together.
Click to expand...

Isn't that the truth. Their BIOSes are so bad now that there are random guys on the internet making custom BIOS revisions for some of their boards to fix things they screwed up.

pavel said:
I suspect you should be okay - I think the B450 and X470 mobos - especially if other brands - are usually better bets with less issues and problems.
Click to expand...

I finished swapping boards late last night and reinstalled Fedora 28. So far, I've have had zero issues. The easiest way I could get it to crash before was by running Quake timedemos, and I let it run all night and day and it is still working with no segfaults or hard locks. I'll have to let it run a few more days before I'll be satisfied that it's stable to put in service as my network server.

pavel said:
Although, I think the 1st generation of Ryzen hardware still holds some issues regardless of brand - but, more likely to encounter it with Gigabyte. I suspect, a lot of it is BIOS-related, too
Click to expand...

I suspect it's down to GB not implementing the latest AGESA firmware in their BIOSes, but according to the guy modding GB BIOSes, there are a whole lot of other problems as well. I don't dabble in the dark wizardry of BIOS edits so I'll have to take his word for it.
 
GiGaBiTe said:
Yeah, that's a model of their board. I had an EP35-DS3L with an E6750 (still have it actually) and never had any issues with it either.



Isn't that the truth. Their BIOSes are so bad now that there are random guys on the internet making custom BIOS revisions for some of their boards to fix things they screwed up.



I finished swapping boards late last night and reinstalled Fedora 28. So far, I've have had zero issues. The easiest way I could get it to crash before was by running Quake timedemos, and I let it run all night and day and it is still working with no segfaults or hard locks. I'll have to let it run a few more days before I'll be satisfied that it's stable to put in service as my network server.



I suspect it's down to GB not implementing the latest AGESA firmware in their BIOSes, but according to the guy modding GB BIOSes, there are a whole lot of other problems as well. I don't dabble in the dark wizardry of BIOS edits so I'll have to take his word for it.
Click to expand...
Yeah, gigabyte AM4 boards (at least 300 series chipset) have/had pretty flaky firmware. I was fairly happy with my fm2+ a88x d3h (except for the common fm2 platform issues and completely unhelpful bios menu descriptions), but their AM4 execution was pretty abysmal (at least for their first run of boards). Mine is working for now, but I'll probably trade it for an MSI board eventually.
 
Hopefully this shitty experience hasn't turned you off on Ryzen. I've enjoyed the heck out of mine since initial launch last year. Also... I feel a little gratified that it turned out to be the mainboard. :)
 
mvmiller12 said:
Hopefully this shitty experience hasn't turned you off on Ryzen. I've enjoyed the heck out of mine since initial launch last year. Also... I feel a little gratified that it turned out to be the mainboard. :)
Click to expand...

I can say I was a little frustrated with the system however once I diagnosed that the NVMe drive appeared to be the cause and I replaced the WDC Black G1 NVMe drive I am happy that it appears to be working fine for 2 to 3 weeks with the Samsung 960 1TB. With that said the system it replaces is over 10 years old and has run 24/7 with > 90% uptime for most of that time so I am looking for a little more reliability than a typical desktop..
 
Last edited:
Hahahahahah. Gigabyte is a shit company now.

My question to their support:

"I just got a RMA replacement CPU from AMD and they said that I need to update the AGESA to 1.0.0.6b, but the BIOS page for this motherboard only goes up to 1.0.0.4?

Is there any BIOS revision that has AGESA 1.0.0.6b?"

Their response:

"Dear xxxxxx,
That code is not available on this platform currently, any particular reason on it?
The current code can support your 1500X CPU"

Not only are they clueless to why it's needed, they do everything possible to be lazy and avoid fixing the problem. Earlier in the conversation, they stated only Windows 10 is supported as an OS on their motherboards "due to 3rd party drivers for other operating systems."

It's been over two days since I swapped it with an ASUS motherboard and I've had no issues since, so I'm going to put it down to a trash board from a now garbage company. Not that ASUS is any better, at least their board works.
 
I wasn't particularly thrilled when my first TR board - a GB Aorus x399 Gaming 7 - died a sudden death. I won't be using any more of their products.
 
GiGaBiTe said:
Doing more research, it seems like it's a Gigabyte specific BIOS issue. After smashing together a bunch of key words on Google, I eventually opened the flood gates on bunches of people complaining about buggy BIOS revisions in regards to AGESA versions and other features not working properly. There's actually a guy on another forum that has made BIOS mods for several GB boards with 3x0 chipsets to fix a lot of the issues, unfortunately mine isn't one of them.

The board I have (AB350M-DS3H) ends at AGESA version 1.0.0.4 with BIOS F23, when the latest is something like 1.0.0.6b and is supposed to fix a bunch of the problems. GB doesn't regard Linux being broken as a problem and to them as long as Windows 10 works, they don't give a flying toss.

Be this a warning to anyone wanting to run Linux on a 3x0 chipset Gigabyte board, you're going to have a bad time. You need to find a board with a confirmed BIOS revision that has AGESA 1.0.0.6b or above, which I think I have found and will try buying one later this week.



Not really sure we're reading the same thing? Nobu and several other people in the thread were of great assistance and I appreciate that. Mega6 only showed up after I got the RMA'd CPU and spent most of his time being a pompous ass and spouting straw man attacks. I wouldn't exactly call that effort.
Click to expand...

Wendell form Level1Techs mentioned some things about AGESA updates in some of his videos AMD / linux videos. have skimmed it sounds like an issues with Gigabyte. I
m more of an Asus guy, never had an problems save for electrical fires due to lightening and faulty PSU and the asus boards work fine melted or not. too bad they are charging an arm and a leg. I also won't touch much anything form them except desktop parts. Fordecades we have run Asus boars with Kingston ram.

Kingston honors their lifetime warranties if you do have problem. One Asus stopped booting after a faulty PSU shot flames out of the express gate drive, Asus flashed my bios for free and sent it back at no cost other than shipping (though i did desolder the express gate myself as it was WELL out of warranty) One Asus board has it's PCI modem struck by lightening back in the day and ran fine with no issues other than no connection until the tech arrive and was greeting with plumes of smoke and flames when the side cover was opened. It worked fine you just would have had hard time installing PCI cards, ended up being a display for years and still worked. I'm under the impression Asus is not the same company. especially wouldn't own any of thier mobile or non desktop / server parts though.
 
GiGaBiTe said:
Hahahahahah. Gigabyte is a shit company now.

My question to their support:

"I just got a RMA replacement CPU from AMD and they said that I need to update the AGESA to 1.0.0.6b, but the BIOS page for this motherboard only goes up to 1.0.0.4?

Is there any BIOS revision that has AGESA 1.0.0.6b?"

Their response:

"Dear xxxxxx,
That code is not available on this platform currently, any particular reason on it?
The current code can support your 1500X CPU"

Not only are they clueless to why it's needed, they do everything possible to be lazy and avoid fixing the problem. Earlier in the conversation, they stated only Windows 10 is supported as an OS on their motherboards "due to 3rd party drivers for other operating systems."

It's been over two days since I swapped it with an ASUS motherboard and I've had no issues since, so I'm going to put it down to a trash board from a now garbage company. Not that ASUS is any better, at least their board works.
Click to expand...

Really this is AMD's fault. Some time ago they reset the AGESA numbers and started over.
1.0.0.4 is newer than the old 1.0.0.6b from last year.

Here is an example. Look through the bios updates: https://www.gigabyte.com/us/Motherboard/GA-AB350-Gaming-3-rev-1x#support-dl
 
You must log in or register to reply here.
Back
Top