Want to upgrade From 3200g or 3400g to 5600g or 5700g? READ THIS. BIG BIOS PROBLEMS POSSIBLE.

[Uttamattamakin]

This is not about Motherboards or CPU's and is not news yet. Some science/tech blogger may have enough to respectably write a news article about this soon. 

 

This is a warning to any who read this board and who may want to build a computer with the Ryzen 5700g / 5600g and or upgrade from 3200g or 3400g. The Bios update you may have to do will in the case of at least one motherboard apparently corrupt the bios if you do it with a 3200g or 3400g.  

 

In my case I have a Asrock Phantom Gaming X570 ITX TB3.   Originally built with a Ryzen 3200g a Picasso APU.  I detailed this in a couple of threads....
 

Here's the twist.  

 

I got the Ryzen 5700g at MSRP day 1.   Backed up data, and followed the procedure to upgrade the bios refered to here. 

Following the special procedure for a Picasso processor I updated the bios. 

It said it completed successfully. 

 

Then when I put in the Ryzen 5700g NOTHING.  It would just power cycle, and power cycle, and power cycle.   It is possible that I got a bad chip but this is what Asrock said.

 

See that.  Following their own instructions will corrupt the bios.    WTH? 

Then They sent me this. 

So these are the three options people who want to upgrade from 3200g or 3400g to one of the new APU's have

• Get a new motherboard. 
• Get some kid that your motherboard manufacturer may act like are well known  but AMD knows nothing about. 
• Send the motherboard to the manufacturer. 
• Happen to just have a Ryzen 3000X series processor handy to use to flash the bios or take it to a repair shop that does.   Which would be really great if they said that instead of the opposite of that  on their site.

I have it in the repair shop at Microcenter and they are going to try to diagnose the problem.  Seeing if any processor will work in that motherboard or if that processor and components will work with a different motherboard.   To see if the Mo bo really is  the issue.  I've paid already so they might as well.  

 

My move might be to just go ahead and build a new PC with the parts.  I had been planning a dual system build before deciding I could just virtualize a second windows system ..... but now that I am facing this reality I see no reason not to.  If the process of getting this motherboard fixed will take more than a week I will just do that.    Then when I get it back build a second mini ITX  system in one case with an atx system.

SO TL DR be very careful with this.  Maybe even just give your computer to a repair shop to flash the bios from the start. 

 

 

Edited August 7, 2021 by Uttamattamakin
Putting more emphasis on apprently corrupt the BIOS. At this point I don't give Asrock much credibility.

[Skiiwee29]

<-- Moved to CPU's Motherboards, and Memory -->

 

Since this is going to involve some aspect of the Motherboard, CPU, or both, this would be the appropriate forum in the mean time. 

[Levent]

8 minutes ago, Uttamattamakin said:

So these are the three options people who want to upgrade from 3200g or 3400g to one of the new APU's have

 

Or flash with a CH340G (or a Arduino that has one) like manufacturers does? Its a $10 part.

[Fasauceome]

12 minutes ago, Uttamattamakin said:

Get some kid that your motherboard manufacturer may act like are well known  but AMD knows nothing about. 

Are they referring to the AMD boot kit?

https://www.amd.com/en/support/kb/faq/pa-105

Either way it's not a perfect solution because of the fact that it's a CORRUPTED BIOS. 

 

Is this only ASRock so far? I was looking at a 5600G or 4650G for my girlfriend's rig, if her Prime X570-P is vulnerable, the 4650G becomes a lot more compelling.

Edited August 7, 2021 by Fasauceome

[Arrogath]

So if you have a bios flash tool found on higher end boards that don’t require a cpu you should be fine. Sounds like an odd issue though, definitely good to know about.

47 minutes ago, Levent said:

Or flash with a CH340G (or a Arduino that has one) like manufacturers does? Its a $10 part.

Probably a bit beyond some of the people on this forum, but it’s always good to have more options 

[Uttamattamakin]

58 minutes ago, Fasauceome said:

Are they referring to the AMD boot kit?

https://www.amd.com/en/support/kb/faq/pa-105

Either way it's not a perfect solution because of the fact that it's a CORRUPTED BIOS. 

 

Is this only ASRock so far? I was looking at a 5600G or 4650G for my girlfriend's rig, if her Prime X570-P is vulnerable, the 4650G becomes a lot more compelling.

ASROCK is the motherboard I had and so far as I know it's just them and this board.  However, I can't be certain that other similar computers won't have the same issue.  

Yeah I'm not sure how a boot kit would help.  Microcenter MAY be able to fix it if. 

1 hour ago, Levent said:

Or flash with a CH340G (or a Arduino that has one) like manufacturers does? Its a $10 part.

this is a thing it'd be something a place like that may have...

 

16 minutes ago, Arrogath said:

So if you have a bios flash tool found on higher end boards that don’t require a cpu you should be fine.

It appears that it does not have that.  It does have a CMOS clear/reset button but I don't think it does that "Bios flashback" thing. 

Maybe I was crazy. I mean why would anyone with a 3200g or 3400g want to upgrade to an 8C 16T "Threadripper" of APU's?   Maybe I was crazy to do that? 

 

[Fasauceome]

2 minutes ago, Uttamattamakin said:

Maybe I was crazy. I mean why would anyone with a 3200g or 3400g want to upgrade to an 8C 16T "Threadripper" of APU's?   Maybe I was crazy to do that? 

They make it seem like it's a generational thing, so you'd probably face a similar problem if it was a 5300G.

 

If ASRock is the only vendor with this issue that's one thing, but if AMD pushed a product that has a fundamental compatibility issue that's pretty newsworthy.

[Uttamattamakin]

2 minutes ago, Fasauceome said:

They make it seem like it's a generational thing, so you'd probably face a similar problem if it was a 5300G.

 

If ASRock is the only vendor with this issue that's one thing, but if AMD pushed a product that has a fundamental compatibility issue that's pretty newsworthy.

Yeah but ... I want to see how they respond by say Monday or Tuesday.   That's only fair.

How they respond to this will dictate what kind of story it is. 

I raised the possibility of them cross shipping me a working motherboard or of the same still current model with a fully upgraded bios.  That said if they are willing to expedite this and get it back to me by Wednesday.  That's one story. 

If their corporate attitude is screw you non social media influencer.  You followed the procedure given on our website and it didn't work ... you should've known better than to think we know what we are talking about...  That's a different story. 

 

I am not a youtuber with 10 million subs but my humble blog, on a science blogging site, is treated as news by Google.  Regular plain old fashioned media can still have an impact. 

[Uttamattamakin]

UPDATE

 

Asrock is RMA'ing this board but there seems to be a bigger issue at play.  It seems that most x570 boards require a bios flash to be compatible with Ryzen 5000.  For those buying now most stores will for a small fee do that for you.  BUT if you already have a computer built you basically cannot do it with your 3400g or 3200g IF what Asrock has told me is right. 

 

I am lucky this is happening just as I have the money to be able to either buy a prebuilt OR build my own.  The stimulus money build was on the small side, just one PCI E slot but it had TB3.  So I just need to look for a computer with both AMD and Thunderbolt that supports Ryzen 5000 out of the box. That said, looking at the cost it is like ... so expensive.  But with the threadripper of APU's in it it might make sense. 

[IkeaGnome]

14 minutes ago, Uttamattamakin said:

BUT if you already have a computer built you basically cannot do it with your 3400g or 3200g IF what Asrock has told me is right. 

I wonder if this is Picasso to Cezanne specific, or if it's an issue on the side of Cezanne and would happen if you tried a flash with Renoir.

[Uttamattamakin]

On 8/9/2021 at 5:36 PM, IkeaGnome said:

I wonder if this is Picasso to Cezanne specific, or if it's an issue on the side of Cezanne and would happen if you tried a flash with Renoir.

Asrock wrote as if it was Picasso specific.  Supposedly Renoir would work.  Which leads to the question why does their website have specific instructions for Picasso?  

Anyway.  After working with MicroCenter to get the CPU swapped out it seems that the motherboard is the issue.  The chip worked in another board and this board did not work with a known working chip.  So  back to Asrock it goes.  

I will be building a new computer though.  A much more of a workstation computer.  https://pcpartpicker.com/list/zQghbh  Basically short of going full HEDT Maybe if there is a Threadripper 7000 or something like it in two years or so. 

[crnkoj]

Hi, i have to chime in here as i had a similar situation.

As i had a fully passive chassis from my old build, which mobo, ram and cpu i sold to a friend (4770s, z97 asrock itx mobo, ddr3 ram) and wasnt entirely happy (the fan spinning up.. the rest is actually awesome on the nuc) with the nuc8 (slightly faster cpu, much faster gpu - iris) i used as a replacement, i decided to reuse the old chassi, which sadly only has mounting holes for 1155/1150/1200 sockets and not am4.

Luckily i got the asrock x570 itx PG open box at a decent discount, so while waiting for the 5700g, i also got a tray 3200g masively reduced.

Now that i bought a 5700g (sadly at msrp), i updated to the official, nonbeta, bios (p3.10) with the 3200g some time ago - logically. The 3200g worked flawlessly in it apart for the rather bad ram speeds (3200 cp15 bdie), which didnt manage to do more than 2933 cl  16... I was even able to undervolt it for better performance/watt.

Anyhow poped the 5700g in and didnt work. Popped the 3200g in again, worked, flashed the 3.14 beta, swapped cpus, again no dice. Though i got a defective 5700g...Luckily i found this thread, reread the warning on asrock support page and was like WTF is wrong with them the 3200g is marketed as 3rd series ryzen (i know its not rly), so took the 3950x out of my gaming rig, booted it up with a dedicated card and the board works with 3.14. Reflashed to 3.14 with the 3950x, popped in the 5700g and it worked. Even the ram works nice (though i put in some E-die crucial ballstix 3000mhz cl15 - can even run it at 3200 with slighlty looser timings).

In any case, if someone is interested. I instantly ran the CPU at 45W, my passive case manages it (the PPT is actually at 45W and the all core freq is 3600mhz) rather well, the CB R23 score is around 11500. I tried 65w afterwards, but the than the PPT is set to 81W and the freq at start as temps are lower at 4200mhz for all core and my case doesnt manage it for more than one CB R23 run (14k score), so i went back for the 45w option in bios. This btw works great. The 3200g has a CB R23 score of 3400 ^ ^. Awesome  upgrade for a fully passive pc. And the best thing the Thunderbolt on this mobo, so i can run the external gpu that i have for my laptop, if i ever wanted to.

 

TLDR

apparently you need to update or at least boot the board once with a matisse CPU in order for the 5xxx Ryzens to work. Not sure how it is with the 4650g/4750g, but i would assume its the same situation as with 5xxx. The board does not seem to brick itself if you update the bios with Picasso cpu and put in a 5xxx CPU, at least it was that way for me.Even better the Southbridge is with the beta 3.145 bios and the 5700g not as hot as with the 3200G and the official 3.10 bios.

[Uttamattamakin]

On 8/13/2021 at 6:30 AM, crnkoj said:

TLDR

apparently you need to update or at least boot the board once with a matisse CPU in order for the 5xxx Ryzens to work. Not sure how it is with the 4650g/4750g, but i would assume its the same situation as with 5xxx. The board does not seem to brick itself if you update the bios with Picasso cpu and put in a 5xxx CPU, at least it was that way for me.Even better the Southbridge is with the beta 3.145 bios and the 5700g not as hot as with the 3200G and the official 3.10 bios.

This may be a thing that varies from board to board.  The repair shop and the local Micro Center CLAIMED that they did that test.  When I looked at the computer as I disassembled it, to get my 5700g out the thermal paste well... it looked exactly like what I did.  Either I did it like a pro or they did not do the test OR they did the test and this did not work for my board. 

It is good to see I am not insane and having one more report shows this is a real issue.  

Is there ANYONE ELSE OUT THERE WHO HAS HAD THIS ISSUE?   If there are 2 there have to be thousands. 

I mean think of what this means.  Oh' yeah I'll just pop in a Ryzen Renoir series CPU I happen to have sitting around.  This is not a reasonable thing.  

 

Personally I am lucky and you are lucky.  You had a second computer with the parts and I could afford to build a new beastly computer (using parts from the old one).  Lots of people save up and cane BARELY afford what they get.  Lots of people are likely to have just been totally screwed by this. 

[infigo]

Sounds like a similar issue I had with the 5600x launch on a asrock b550m steel legend. 

Got it, found it was using a older bios, fine the store helped me...but it didn't post. Have it back to them (together with CPU) and they couldn't figure it out. They could update to latest bios, but 5000 CPUs still would not post, they tried 3 different boards with same result. 

They said to me to change board instead which worked (I choose Asus b550m tuf). 

 

After I've seen similar issues on Reddit, so something of how they update their bios which can't update some region to support 5000 even though the rest was up to date, (p1.20 at the time). Models with bios with p1.10 apparently didn't have this issue. 

 

What cpu my store used is unknown but they used the same one and updated the Asus board for me to make sure it worked well and there it was no issue. 

 

Hopefully asrock don't have similar issues again but sticking with Asus now, (only had Asus before too). Have always been in a Asus Vs asrock choice on every one of my personal machines, before Asus fitted better in my chassi constraints but for once asrock was the better fit in the chassis bit got burnt unfortunately.

[Uttamattamakin]

Quick update.  
Recieved a new motherboard with their BIOS 2.80  I will eventually, this week or next try making a computer out of it with the old chip the R3 3200g on their website it seems to indicate that this should work.  If not then to heck with it.  At worst it may not post. 

All in all Asrock has proven willing to RMA boards in this situation just beware of updating the bios if you have a 3000 series APU.