Processors

[ipegoraro]

Hello,

I recently purchase a Ryzen 5 3600 along with a Asus Prime B450 Plus and a G.SKILL Aegis 16 GB (2x8) f4-3000c16d and the power supply is a Crosair CX450M. I have installed both Windows 10 and Fedora 30, and in both OSes the system is unstable, meaning after a phew minutes of usage (browsing or even staying idle) the system would crash with a Blue Screen of Death or Kernel Panic.
BIOS is up to date to the latest version for the motherboard (1820 released 2019/09/24), already tried with another motherboard (Asus Prime X370 Pro) and same problem.

I've also tried removing one ram DIMM and system is still unstable.

No overclock has been done everything is set to the default values.

Any advice? or should i return the processor?

Thanks in advance

[fiska]

I am using bios with Agesa 1.0.0.3 ab, without BSOD, but when I turn off the computer it takes about 5 minutes to shut down completely, anyone with the same problem?

[hardcoregames_]

fiska wrote:

I am using bios with Agesa 1.0.0.3 ab, without BSOD, but when I turn off the computer it takes about 5 minutes to shut down completely, anyone with the same problem?

The latest BIOS shuts down immediately (almost)

[fiska]

I know that but there is BSOD

[hardcoregames_]

try backing off on the RAM timing in the BIOS a bit and see if that works better

[fiska]

I'm using 18 22 22 42 (native XMP) - 3600mhz, I tested without XMP, I also tested with ryzen dram 16 16 16 32, and yet BSOD happens, which timming would you recommend?

[fiska]

Before updating the chipset was worse, it happened several times, now with the latest BIOS and chipset rarely happens 2-3 per day, the problem is that I work using chrome and this is where the BSOD happens most.

[nejcella]

Sorry, but the CPU should work (and it does if it's not from a faulty batch) without changing RAM timings and other settings in the BIOS. 

Mine even freezes on ASUS splash screen if I update to latest version. The only so-so stable version still produces BSOD (watchdog_timeout), but not 10 times a day. When you change every component (RAM, PSU, Motherboard,...) and it's still the same problem, then you now something is wrong with the CPU. When I put in Ryzen 7 2700x or Ryzen 5 3400G it works with every BIOS version.

[trevayne]

nejcella wrote:

Sorry, but the CPU should work (and it does if it's not from a faulty batch) without changing RAM timings and other settings in the BIOS. 

Mine even freezes on ASUS splash screen if I update to latest version. The only so-so stable version still produces BSOD (watchdog_timeout), but not 10 times a day. When you change every component (RAM, PSU, Motherboard,...) and it's still the same problem, then you now something is wrong with the CPU. When I put in Ryzen 7 2700x or Ryzen 5 3400G it works with every BIOS version.

Yeah, I don't care what I did with the timings of my RAM -- there was simply no cooperation with the current AGESA.  So I'm back on 1.0.0.3abb.  RMA folks just got to me and asked for more information, so I guess I'm in for the long haul -- or at least until my patience runs dry.

[hardcoregames_]

Until the H.60 BIOS for My MSI X570-A PRO surfaced, My G.Skill memory did not work at default properly

F4-2400C17-8GNT

F4-2400C15-4GNT

waiting on some new faster memory but UPS has been dragging their feet and its 3 days late

[hardcoregames_]

My RAM arrived finally

G.Skill F4-3200C16D-16GFX

So far the BIOS was able to set it to DDR4-2933 which the best my R5-2400G can muster.

[fotoamg]

Hi!

I shared my problem here in a previous post last year and than went to older abba bios.

With abba bios it is stable but there are about 5% performance decrease in benchmarks.

I am testing it since a while and so far no BSOD or at least not while using it, sometimes I came home it was turned off but it could be electric outage or whatever.

As for people RMAd their CPU and using the same config the replacement one works without issues it is for sure CPU problem.

Now after the Holidays I plan to RMA my CPU, I am sure the CPU is faulty, and sure this thread for AMD developers is not where they reading about product issues

For those who run an RMA I would like to ask is it working only in specific countries?

I am from Europe, Hungary and bought the CPU at local shop where it has warranty, but I am afraid it would be hard to convince local shop that the CPU is faulty and needs RMA.

So can I RMA it from here myself?

Where to start the process?

What kind of pictures should I prepare?

(I don't want to remove CPU many times of course before sending it.)

Also where are these info:

Product Part Number

Product Serial Number

Thanks!

[trevayne]

I was told to voice my problems to MSI.  They will not RMA my CPU because 1.0.0.3abb has the system stable, so obviously in their eyes, it's not my CPU.  And I've tried to fiddle with the RAM timings, with no success on 1.0.0.4b.  Guess I just have to see if MSI fixes whatever is wrong in the BIOS for the X570 Gaming Plus?

[hardcoregames_]

there are still lots of BIOS problems to be fixed but no timetable is available for a fix to be posted

[trevayne]

You actually had a point about the timings, Hardcore.  I looked at the timings/voltage while on 1.0.0.3abb and noted them.  I flashed to 1.0.0.4b, and saw that the timings/voltage were... well, let's say "off," especially the voltage (1.22V on B vs. 1.37 on ABB).  So I applied the timings and voltage to 1.0.0.4b, and here we are stable for the last two hours, which is... a definite improvement.

[fotoamg]

Can it be that the processor itself has instructions (metadata) inside for voltages and load calculation and the BIOS is just using it  and the faulty CPUs which make BSOD with newer bios has wrong data saved in their microcode (firmware) only?

AMD could add a firmware patch if this is the issue.

[trevayne]

I have no idea.  I am so very beyond knowledgeable in this department.  All I did was compare the timings and voltages between the two AGESA, and 1.0.0.4b was way undervolted, plus the timings were off (especially the tRC value).  Mind you, this was on my board, and it could be an entirely different can of worms for other motherboards; but if you feel adventurous, you can look at your timings on a stable BIOS version and compare to 1.0.0.4b.

If I had to hazard a guess, though, it'd be that it can be a faulty CPU and/or it could be the wrong coding in the AGESA.  Like Hardcore mentioned, 1.0.0.4b does look like it needs to be ironed out more, so I'm going to assume that it's probably safer right now to be on a stable BIOS.  You know, rather than fiddling with timings and voltages to HOPE to be stable on the latest, 'cause my culprit could possibly not be your culprit.

[hardcoregames_]

trevayne wrote:

You actually had a point about the timings, Hardcore.  I looked at the timings/voltage while on 1.0.0.3abb and noted them.  I flashed to 1.0.0.4b, and saw that the timings/voltage were... well, let's say "off," especially the voltage (1.22V on B vs. 1.37 on ABB).  So I applied the timings and voltage to 1.0.0.4b, and here we are stable for the last two hours, which is... a definite improvement.

I also had to overvolt memory to stabilize the machine.

I have lots of experience with overclocking which is what I used to stabilize my rig with the memory I was using.

I have new memory now so new issues to contend with

[al_sami96]

I sold my previous 3600 and bought a new 3600.Guess what !! It's working flawlessly..From this perspective I can assure,a particular batch of 3600 is having bsod constantly after ab bios update.

[trevayne]

al_sami96 wrote:

I sold my previous 3600 and bought a new 3600.Guess what !! It's working flawlessly..From this perspective I can assure,a particular batch of 3600 is having bsod constantly after ab bios update.

I wish I were in the position to get another 3600, but if adjusting the RAM timings/voltage for me was my solution, then I don't think I had a faulty CPU like a lot of folks in this thread -- just poor coding in 1.0.0.4b for my board.

Thanks again to Hardcore for steering me in the right direction.  I honestly wouldn't have looked into comparing the timings/voltage between the BIOS versions, if you didn't strike up my curiosity.

[al_sami96]

Specific batch of 3600 wasn't configured in a optimized way..Thus voltage and timings were conflicting with latest Bios.And latest 3600 I meant the later batch which is now available everywhere is working in every bios without any issues and thus no changes need to be made manually to run it properly in latest Bios.

[leukosaima]

trevayne‌ let me know if you BSOD... because i've noted my timings and voltages for ram in 1.0.0.3abb and applied the same ones to 1.0.0.4b and it has not helped, so i'd be interested to know if somehow it works for you.

[trevayne]

leukosaima‌ Yeah, I spoke too soon, I'm afraid.  Weird, though, 'cause it was fine all this time, and then I reloaded a game, and BOOM, there it went. :\

I'm at a loss here.  And AMD isn't going to replace my CPU, so... yeah.  All I can say is that it has to be the AGESA.

[leukosaima]

maybe next step is to make sure the cpu voltages are the same across AGESAs

[hardcoregames_]

New processors are coming soon so new BIOS updates will surface to support them

[fotoamg]

I started the RMA process, I was asked for pictures,

I've put my pictures also to drive, I hope it is enough for them for RMA...

AMD Ryzen 5 3600 RMA - Google Drive 

will share here how the process goes...

[Anonymous]

Update, I took my CPU, RAM back to Micro Center and did a swap, no restarting at all. I wish i knew what one was bad ( ( i really think it was the CPU). So everything seems good now.

My issue before i took back the parts was the computer would restart during low idle power times. I first noticed that my computer was resatrting during windows install. Underclocking the CPU (3600x) and RAM 3200 G skill 16gb x2 let me install.

Before returning my CPU and RAM, i changed motherboards 3 times. Tested every bios, Asrock b450, asus strix b450 changed HDDs, m.2, made sure case wasn't casuing the issue, changed GPU.

There really seems to be some bad batch CPUs out there, because newegg has a bunch of 1 stars with people talking about this same issue of restarts during low power. I also want to bet there are more people out there with this issue, that just don't know it. Changing the power setting in windows let me stay in windows without that many restarts.

So now i get to use my computer, no more issues. Also while trying to find my issue i learned some bad support companies. AMD, you'll are crazy bad. I sent in my first email, and they came back and acted like they didn't know what i wanted help WTF.

Anyone know that message you get from AMD, don't reply in 10 days and this will bla bla bull.

I messaged evey company part i owned. Seasonic was the fastest. They messaged back in 2 damn mins, like wtf plus they were ready to do an RMA on a 7 year old XP2 660 just like that.

AMD needs to learn from them, shoot all the companies do. G skill took 2 days, like get the hell out of here. Asus was okay, with online chat. Micro center was crazy good help to be honest though. Now i guess if you don't have their warranty, umm they might try and use their in house techs that charge you $34 just to look at it lol

So now i do understand why no one wants to RMA with AMD, they're slow as hell haha. Hope you all get it fixed.

[fotoamg]

I sent them my pictures for RMA more then 1 day ago, 2 days now, not reply yet and I keep my CPU in its box, not able to use my new system as waiting for the next steps  If I put CPU back I need to wast an other bunch of my expensive silver thermal compound and not easy to remove my Arctic Freezer Extreme cooler...

So waiting still for their response...

p.s.: I am sure the issue is CPU related and metadata in their microcode which can be fixed by them with firmware update in the CPU which cannot be done at home, but it is just my technical related idea what's going on here.

Of course this metadata can be overridden by bios so BIOS can solve the issue skipping wrong 'fine tune' data from CPU power cycles but if it can be completely eliminated by a replacement CPU (with fixed microcode) without BIOS hacks so it is AMD's fault.

[trevayne]

Lucky for you that they're even considering an RMA.  I guess I shouldn't have told them that things were stable on 1.0.0.3abb, 'cause they immediately pointed at MSI and said, "Their problem, not ours."  So I think it's a combination of AMD and the mobo manufacturers, Foto.  There's a beta for my X570 Gaming Plus currently, but I don't think that's going to solve anything on my end, when it gets OUT of beta.

Sad state of affairs, honestly.

[fotoamg]

Oh after I sent the pictures they answered now whith that:
"Thank you for your email and providing with requested pictures.
Could you please provide your complete system configuration and let me know have you installed the latest chipset drivers and provide the make & model of your graphics card.
Also, if feasible, could you please check your processor on different motherboard that has compatible BIOS and let me know the issue status. "

So I need to buy some other mobo and CPU to try it? What....

[trevayne]

No, just tell them you don't have another motherboard to test with (that's silly), and give them whatever else info they want.  They're probably going to tell you to test other GPU drivers and to make sure you're up-to-date there.  I think it's ridiculous what hoops you have to go through with them.

[kkimic]

Strange, I did not send any photos and they immediately asked me to RMA my 3900X through the supplier. I was unlucky with some many bios flashes that it restarted whilst flashing in the end and I ended up bricking the mobo in the process, really frustrated.

Went back to the supplier and had lots of issues them not accepting these until they could test for sure that they were faulty. Mobo was easy as it was bricked, so there it went through the RMA process.

Processor they installed in another mobo and it worked (of course), but I insisted they installed windows and test it longer and at idle. BSOD during windows installation at the shop so straight into RMA.

3 weeks they told me I have to wait, went back to my 6700K and could not be any happier whilst I wait.

[trevayne]

B450 Tomahawk BIOS bsod issues  <--- Just want to point out this forum post.  This more or less addresses stability problems after ABB, with a focus on the interactions with a Ryzen 3600 and AGESA ABBA and up.  Yeah, this is MSI, but I think it can be applied to other manufacturers.  Someone mentioned a workaround, but that involves downclocking the CPU, which... I would think nobody really wants to do.

[hardcoregames_]

trevayne wrote:

B450 Tomahawk BIOS bsod issues  <--- Just want to point out this forum post.  This more or less addresses stability problems after ABB, with a focus on the interactions with a Ryzen 3600 and AGESA ABBA and up.  Yeah, this is MSI, but I think it can be applied to other manufacturers.  Someone mentioned a workaround, but that involves downclocking the CPU, which... I would think nobody really wants to do.

It seems that BIOS issues with Ryzen 3000 apply widely across several chipsets. This suggests there may be some fault with some of the processors as I have seen anecdotal posts were a CPU was returned for a replacement that worked.

[trevayne]

Unfortunately for me, I already tried barking up that tree.  Because things are stable on 1.0.0.3abb, the AMD RMA people more or less told me to take a hike and to take issue up with MSI.  I'm... in the middle of figuring things out on their end.

[hardcoregames_]

trevayne wrote:

Unfortunately for me, I already tried barking up that tree.  Because things are stable on 1.0.0.3abb, the AMD RMA people more or less told me to take a hike and to take issue up with MSI.  I'm... in the middle of figuring things out on their end.

I have also got MSI motherboards and I have griped a lot.

[formaldisaster]

Update

Received replacement cpu today from Amazon

Batch number shows a wk 41 manufacture date and unlike the first one (wk18 made in china), the new one is made in malaysia. - Just thought worth mentioning.

So was on ASUS 2704 bios 1.0.0.3 Patch ABB

I installed cpu flashed bios to latest 3003 1.0.0.4 patch B

and set Memory to DCOP profile

Booted...held breath....into windows

Stress tested CPU with OCCT for couple of hours, and been using it on and off through the day, left on at idle too (which was one of the bsod points)

According to ryzen master the max core speed was 3977mhz, across all 6 cores, now (unless something has changed - and I dont know where to even look for this) what is noticeable is that while OCCT is running the 6 cores are showing the same speed at the same time (set to current in the histogram), whereas on the old chip the speeds were all over the place (by core) up and down ranging from 700 - 3980

So far so good.

[pepz]

Had the same problem with Ryzen 5 3600. Produced in week 19, 2019. I send it back and fit one from week 45,2019. Now no BSOD in idle even with latest BIOS. So ist was the CPU.

[nodaone]

Same here! Had my 3600 replaced and all problems where gone.. 

[fotoamg]

Hi!

Here is how my story going....
I started RMA process at AMD, but it would be about 4-5kHUF to send it to the Netherland by post so I went back to the shop where I bought it and wanted to replace under warranty.
They said it will be about 31days they send to the supplier and the supplier's supplier to check etc.
I was waiting 25 days no response from the shop, I sent email, they replied that it s faulty cannot be repaired I will get back the price or can shop using that balance at their shop to buy a new one.
Ok, but they sold it to me as 'opened box' version for 63kHUF (so cheaper but they did not tel me it is faulty just opened for demo etc.) and the new one is 70kHUF now.
So I had to pay 7kHUF extra to get a new one, with AMD RMA it would cost only about 4-5kHUF postage cost :
So I' lost abut 10% of the price again and has to wait for the shop to et me a new one for an extra 7kHUF now.. still waiting for them...

I hope the next CPU won't be faulty because I've lost 2-3 months from my life debugging and arguing with them etc.... and lost 10% of the price, as I could already buy a full price version 3 months ago

[hatsantos]

I created an account to report the same problem as you guys.

I built a new AMD system on the first weeks of the Ryzen 3000 series. Since the first day my system is unstable...
My initial PC had the following hardware:

Ryzen 3700x
Gigabyte 450 Pro
Corsair RGB Pro 3200 CL16
GPU Gigabyte 5700XT
PSU Corsair RMx 750W

With this setup I got a lot of BSoD when in idle, doing simple tasks like browsing or coding.
When I play games or other heavy tasks the PC works fine with no shutdowns. Never got a BSoD during long game sessions or benchs.

Trying to fix the issue I have already RMA the motherboard and the RAM.
I replace the board with a Gigabyte x580 Elite (spent more 100 bucks hopping for a stable machine) and better bios but no luck on that. 
Recently I also replaced the RAM with a G.Skill Trident Neo, since it is recommend and designed to be compatible with Ryzen systems, but the BSoD keep comming...

I'm trowing money away in the hope to fix the issue without luck... Never tougth that could be the CPU, but hopelessly, I found this thread.
I will downgrade the bios and see it that fixes my issue if not I will try to RMA my CPU.

This came in a bad time... Since this corona virus situation I have to work from home, and it's been almost impossible... I can't RMA the CPU at this moment... and if that's the only thing that fixes the problem I'm screwed...