Question B550 chipset, so AMD joins the dark side after all.

Page 18 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Shivansps

Diamond Member
Sep 11, 2013
3,855
1,518
136
I just read the article...

Ryzen%203_B550_Press%20Deck_NDA%20Until%20May%207th-page-008_575px.jpg


Ryzen%203_B550_Press%20Deck_NDA%20Until%20May%207th-page-005_575px.jpg


Ryzen%203_B550_Press%20Deck_NDA%20Until%20May%207th-page-004_575px.jpg


So let me get this straight, this chipset is coming out like a year later, they did not even bother to add CPU PCI-E 4.0 uplink support or to increase the number of sata ports that is ALREADY a problem on every 6 sata B450 motherboard (NVME x4 disables the 2 SOC Sata, thus 6 sata B450 mbs losses 2 sata if NVME is used), and they even dare to futher reduce backguard compatibility?

I was not expecting for the PCI-E lanes FROM the chipset to be 4.0, but only USB 3.2 G2, no more satas, CPU link still 3.0 and the PCI-E lanes 3.0 is beyond disappointing.
 
Last edited:
  • Like
Reactions: PingSpike

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
28,496
20,605
146
I am hoping that the consumer advocacy the tech press is engaging in over this, will bear fruit. My speculation is that this will go the way the xbox one drama went. MS made numerous announcements that were unfriendly to the consumer. The tech press did a full court press, consumers went full ham, resulting in MS reversing course/doing the best they could at damage control.

I am fairly confident that the current ham level will result in officially supported 4 series boards.

.
Called it


 

Ajay

Lifer
Jan 8, 2001
15,454
7,862
136
As i said, people only had to make some noise and they will listen. I hope everyone learned a lesson here today, this is not thanks to the ones that were searching hard for justifications at every corner.

We all win from this, AMD included.
Eh, AMD has a black eye over this, IMHO. But black eyes heal and are forgotten about in time.
One thing I think we will see with AM5 is far less promises from AMD on future compatibility.
 

SPBHM

Diamond Member
Sep 12, 2012
5,056
409
126
good change by AMD!

still a little worrying in regards to how they worded it with limited beta bios and not being widely available requiring the person to show that they have Zen3 and the board already, weird,
still better than giving no options like before,
they had to react to all the negativity (rightly so) from their previous decision.
 

dlerious

Golden Member
Mar 4, 2004
1,787
724
136
I think it was MSI that was responsible. They promised support, opening them up to possible lawsuits. Floating the possibility of leaving the AMD ecosystem would open a few eyes.
 

AtenRa

Lifer
Feb 2, 2009
14,001
3,357
136
I think it was MSI that was responsible. They promised support, opening them up to possible lawsuits. Floating the possibility of leaving the AMD ecosystem would open a few eyes.

With Intel DIY desktop market share in the abyss, I dont believe any board maker would like to leave the AMD platform.
 

DrMrLordX

Lifer
Apr 27, 2000
21,634
10,849
136
There is always going to be someone whining about something.

Some whine more than others. It is somewhat-reasonable for people to want Summit Ridge, Pinnacle Ridge, Matisse, and Vermeer to all work in any AM4 board (outside of A320), but the technical feasibility of that feat is questionable. AMD and the OEMs are now taking on the herculean task of trying to figure out which CPUs the market wants working in which boards and making it happen seamlessly.

I suspect that only x570 and b550 will "go well". Everything else will be at least as messy as Matisse on 300 and 400-series boards last year. Probably worse.


Not surprising. If you have microcode available to get Vermeer working in 400-series boards, it's probably possible to adapt that to 300-series boards. They aren't THAT different.

We all win from this, AMD included.

Yes and no. AMD just opened a can of worms. Matisse was already a hot mess on some old boards. This could get really bad, and if it delays launch, it's going to suck for people who were actually willing to pony up for a current board to run Vermeer.
 

CHADBOGA

Platinum Member
Mar 31, 2009
2,135
832
136
Excellent stuff from AMD. :D

I still haven't made up my mind about what AMD cpu I will be getting later this year, but now I have more options, as I might go for a 1600AF or 3300x this year with a B450 MSI Mortar MAX and upgrade the CPU to a Zen 3 in 2022.
 

AtenRa

Lifer
Feb 2, 2009
14,001
3,357
136
Excellent stuff from AMD. :D

I still haven't made up my mind about what AMD cpu I will be getting later this year, but now I have more options, as I might go for a 1600AF or 3300x this year with a B450 MSI Mortar MAX and upgrade the CPU to a Zen 3 in 2022.

I would get the 3300X with B550 and then upgrade to 8C16T ZEN 3 when you feel you need more performance.
 

maddie

Diamond Member
Jul 18, 2010
4,740
4,674
136
No you not getting it. New BIOS will lock those board with only 4000 Series CPU's. People who buys 1000/2000/3000 CPU/APU now have to look hard to buy new B450 that supports their CPU/APU limiting their motherboard choice. Remember B550 dont support 1000/2000 series CPU/APU and 3000 Series APU.
I don't think any stock B450 board should support the Ryzen 4xxx. Leave it up to the interested to do that for themselves.
 

misuspita

Senior member
Jul 15, 2006
401
452
136
it's probably possible to adapt that to 300-series boards. They aren't THAT different.
I wonder if the new 4xxx APUs will work on AsRock A300 thingie. That would be the first time the AMD one will be equal/better than Intel one, since you could technically put a 9900k in there (AsRock 310). Now, with a small box of 15x15x8 cm you could have a enormously potent mobile deskop computer at your fingertips, because 3400G wasn't really that great.
 

Shivansps

Diamond Member
Sep 11, 2013
3,855
1,518
136
Yes and no. AMD just opened a can of worms. Matisse was already a hot mess on some old boards. This could get really bad, and if it delays launch, it's going to suck for people who were actually willing to pony up for a current board to run Vermeer.

Whats exactly the issue? At home im running a 3600 on a non premium B350 and it is running whiout issues. At work i have a "server" running 24/7 with a A320+3600 for months now with no issues... And we sold around 300 Matisse (3600,3600X,3700X) with 300 boards, mostly AB350M-DS3H V2,. It may had been some voltage issues with early bios but no now.
 
  • Like
Reactions: DAPUNISHER

zir_blazer

Golden Member
Jun 6, 2013
1,165
408
136
Look at this crazy thing:


They must be taking slices off the GPU 16x block for the extra 2 m.2 slots. But then again, there are 3 GPU slots so I'm interested how the breakdown works out. It looks like its going to an odder duck than the Asus B450 Strix board.
The only surprising thing is that the main 16 PCIe Port can bifurcate to 8x/4x/4x. At least according to official AMD slides of previous Chipset generations you need a X series Chipset to allow bifurcation to 8x/8x, and there was no mention of 8x/4x/4x at all (Note that Intel consumer Processors supports both 8x/8x and 8x/4x/4x modes, but only allows doing so on Z series Chipsets only). Keep in mind that Zen itself can do 4x/4x/4x/4x, but I'm not sure if this capability was ever exposed in desktop, as it falls onto the category of "Chipset controlled Processor features". Threadripper can do it. Not sure if someone showcased 8x/4x/4x + x4 on AM4 before.
Basically, think that an AM4 Processor itself with its 24 PCIe Lanes loses 4 as it is the downlink to the Chipset, then it can use the other 20 as 8x/4x/4x + 4x. Sounds like you could have a full sized ATX Motherboard with 4 PCIe 4.0 Slots if you wanted to.


So its the CPUs that can't address more than 16MB. Well, that is going to be a big mess to support then, but it was already going to be a big mess. I think people will appreciate the option nonetheless.

Socketed bios chips would kind of solve this. The manufacturers could even charge for them to mitigate the cost of the extra work. I'm not sure how common those are these days.
Socketed Flash EEPROMs would NOT solve this, because they introduced a multitude of issues with unique Motherboard unit data. UUID, MAC Addresses, UEFI NVRAM, and so on, would be lost if you just swapped chips. There goes your Windows activation, too. Too complicated.

To address 32 MiB and bigger SPI Flash EEPROMs you were supposed to support 4 Byte addressing, which I thought that Zen supported since there were Motherboards with 32 MiB SPI chips in the wild. I suppose that the SPI chips themselves support some form of bank switching so that you can tell them to show either the first 16 MiB or the last 16 MiB of the ROM contents on the 16 MiB addressing window, and the problem seems to be that you may require to work with more than 16 MiB at once. Otherwise you're back to the good old days of DOS and the Expanded Memory/EMS.
 
  • Like
Reactions: lightmanek

DrMrLordX

Lifer
Apr 27, 2000
21,634
10,849
136
Whats exactly the issue?

It took over three months to get the majority of 300-series and 400-series boards running Matisse properly without issue. It took maybe another 1-2 months to get them all updated to 1.0.0.3ABBA or later. And 1.0.0.3ABBA was the first truly "good" AGESA version for Matisse. Not to speak of the numerous boards that had to hide features or cut basic support for things like a mouse in the UEFI. And the weird boost/voltage behavior when running Matisse on some older boards.
 

PingSpike

Lifer
Feb 25, 2004
21,732
561
126
The only surprising thing is that the main 16 PCIe Port can bifurcate to 8x/4x/4x. At least according to official AMD slides of previous Chipset generations you need a X series Chipset to allow bifurcation to 8x/8x, and there was no mention of 8x/4x/4x at all (Note that Intel consumer Processors supports both 8x/8x and 8x/4x/4x modes, but only allows doing so on Z series Chipsets only). Keep in mind that Zen itself can do 4x/4x/4x/4x, but I'm not sure if this capability was ever exposed in desktop, as it falls onto the category of "Chipset controlled Processor features". Threadripper can do it. Not sure if someone showcased 8x/4x/4x + x4 on AM4 before.
Basically, think that an AM4 Processor itself with its 24 PCIe Lanes loses 4 as it is the downlink to the Chipset, then it can use the other 20 as 8x/4x/4x + 4x. Sounds like you could have a full sized ATX Motherboard with 4 PCIe 4.0 Slots if you wanted to.

The Asus B450 Strix boards, or at least one of them does an 8x/4x/4x m.2 split to get two pci-e 3.0 m.2 slots. Its weird because as you said, it doesn't seem like this should be possible on that chipset. I suspect the secret on this new one is that the the secondary GPU slot can't be in use with all 3 pci-e 4.0 m.2 slots. This is a cool board but we'll have to wait and see its price I guess.

I'm not sure about the bifurcation (quadfurcation?) 4x/4x/4x/4x support being locked behind a chipset. From the b450 tomahawk max manual:

PCIe SlotX Lanes Configuration
PCIe lanes configuration is for MSI M.2 Xpander / MSI M.2 Xpander-Z / Other M.2 PCIe storage card. The options in this item will vary with the installed processor

I have that same option in my x470 bios. I don't have any expander cards though.

I still think if they sold B450 boards with a second Zen3 flashed bios chip in a bag that was for that board it would still work wouldn't it? I doubt they'll go that route though.
 

PingSpike

Lifer
Feb 25, 2004
21,732
561
126
Something still doesn't make sense to me. x570 boards often have 16mb bios and they already supported Zen+ through Zen2 and were definitely going to get Zen3. And people have at least gotten them to boot with even Zen1.

If that is the case...then it implies 16mb is enough space for all the cpu generations. So why is a bios that supports all impossible on 400 series?
 
  • Like
Reactions: Tlh97 and maddie

Makaveli

Diamond Member
Feb 8, 2002
4,718
1,054
136
Not just take away, but take away permanently. Meaning once you flash for Zen3 compatibility, you can't flash back to use an older CPU again that was dropped. There's going to be other unintended (by people wanting this) consequences pop up as well, I'm sure. It's going to be a mess to deal with on a global scale. I hope customer support teams for all the Mobo makers get a nice Christmas bonus this year, lol.

Just imagine the call to customer support for someone who bought a new b450 board and a Zen3 CPU. The board won't boot with their CPU because they have to update the BIOS first so customer support has to tell them to get an older CPU that the BIOS supports to update to Zen3 at which point the older CPU is likely useless in said board. Or they have to pick the right model that is supported by both BIOS versions and make sure not to pick from the list that is no longer supported in the later BIOS. I don't envy them.

I c you point.

But do normies really call for support on updating a bios over the phone?

I would figure most would just take it back to where they got the machine and ask the shop to do it. local computer store, best buy etc.

As much as I would like that AMDs backtracking is due to us enthusiasts, it is not. Well, not in a measure we would like.
It is due to motherboard manufacturers and ODMs.
As to all other things regarding AMDs problems,well, they are AMDs problems.

But I do get what people are saying.

Now AMD has push all the responsibility to the Motherboard Manufactures. So they can deal with Bricked motherboards, people flashing incorrectly. People not knowing they will need to flash a Zen 3 compatible bios to their board first before purchasing the cpu.

Now all the blame they can deal with so kinda of smart move if you ask me.
 
Last edited:

Hitman928

Diamond Member
Apr 15, 2012
5,290
7,920
136
I c you point.

But do normies really call for support on updating a bios over the phone?

I would figure most would just take it back to where they got the machine and ask the shop to do it. local computer store, best buy etc.

I was talking about the DiY market but even in the case of prebuilts, Best Buy and many local shops aren't going to help you upgrade the computer for free, so there's that. But what about all the computers purchased from online only stores or places like Walmart that don't have repair teams?
 

Shivansps

Diamond Member
Sep 11, 2013
3,855
1,518
136
It took over three months to get the majority of 300-series and 400-series boards running Matisse properly without issue. It took maybe another 1-2 months to get them all updated to 1.0.0.3ABBA or later. And 1.0.0.3ABBA was the first truly "good" AGESA version for Matisse. Not to speak of the numerous boards that had to hide features or cut basic support for things like a mouse in the UEFI. And the weird boost/voltage behavior when running Matisse on some older boards.

Yeah, there was issues with early bios, but there was bios issues with every Ryzen launch to date. The most important thing it was eventually fixed.