Zen 3 CPUs - 500 Series or Newer Chipsets Only

Nimisys said:
When my X370 board died, bought an X470 so I could keep using my 1700x until I was ready to update the CPU, so an X570 was out. Gonna get burned on that one, as I was hoping to wait for the 8 core 4000 series to drop as the point to upgrade. :/
Click to expand...

Possibly... AMD has learned to not over promise things in slides. I wouldn't write off x470 or eve x450 support until things launch. AMD is not going to announce blanket x400 support cause some boards may have ROMs that are just to small. I am sure though at launch the MOBO MFGs will announce a handful of 400 series that have support.

AMD not announcing every 450 board will work is expected. I am sure some boards will not be compatible, better to not announce then have confusion. Also I am sure the MOBO MFGs appreciate AMD not committing them to support work... and perhaps helping them sell some new 500 boards.
 
Hagrid said:
But you can make the bios smaller, which is what they are doing?
Click to expand...
mnewxcv said:
right, but you can add something if you remove something.
Click to expand...
I said that same thing twice in the thread.

The issue is like i said earlier if you update your bios to get better stability for your 1 or 2000 cpu and get the wrong bios you lose support for your 1 to 2000 cpu.

There was even a quote from MSI i put in that said that.

It's up to motherboard manufacturers to do that not AMD.
 
Master_shake_ said:
I said that same thing twice in the thread.

The issue is like i said earlier if you update your bios to get better stability for your 1 or 2000 cpu and get the wrong bios you lose support for your 1 to 2000 cpu.

There was even a quote from MSI i put in that said that.

It's up to motherboard manufacturers to do that not AMD.
Click to expand...
I don't think it would be that hard to do a microcode check when flashing, like a model/version check that already occurs, to make sure you don't end up with a non functional computer. We will see soon enough though.
 
mnewxcv said:
I don't think it would be that hard to do a microcode check when flashing, like a model/version check that already occurs, to make sure you don't end up with a non functional computer. We will see soon enough though.
Click to expand...

I could be wrong, but doesn't BIOS flashing require an active CPU?
Hence that whole thing about flashing in UEFI/DOS to prevent CPU preemption in OS from disrupting the flash process.

EDIT: Oh I think I see what you mean - ie. check the current flashing CPU and enable that in the BIOS along with 4xxx.
 
I think my biggest gripe is really the b550 being so late to the show. For the past year nobody buying a 3600 wanted to spend as much or more on their motherboard than their CPU. So, they had to get a b450 or x470. Now they are at a dead end because they had no low/mid options (still!!). If the b550 was released and was promised forward compatible, a feel a lot of folks running 3600/3600x would have got that instead of the 450/470. Had the b550 been out and people had the option it wouldn't make this a big deal. Not being up front (like not squashing rumors of backwards compatibility when it pops up) is almost as bad as lying about it. They sold all the b450's with 3xxx series knowing people weren't going to spend the $$ on the x570 by allowing them to believe they still had a path forward. Frustrating for sure, I was looking forward to handing down or repurposing my CPU and upgrading.
 
schoolslave said:
I could be wrong, but doesn't BIOS flashing require an active CPU?
Hence that whole thing about flashing in UEFI/DOS to prevent CPU preemption in OS from disrupting the flash process.

EDIT: Oh I think I see what you mean - ie. check the current flashing CPU and enable that in the BIOS along with 4xxx.
Click to expand...

There are some motherboards out there that allow the flashing of the bios without a cpu installed.
 
schoolslave said:
I could be wrong, but doesn't BIOS flashing require an active CPU?
Hence that whole thing about flashing in UEFI/DOS to prevent CPU preemption in OS from disrupting the flash process.

EDIT: Oh I think I see what you mean - ie. check the current flashing CPU and enable that in the BIOS along with 4xxx.
Click to expand...
yeah, I meant so that it checks to make sure the microcode for the current CPU is within the BIOS being flashed.
 
mnewxcv said:
yeah, I meant so that it checks to make sure the microcode for the current CPU is within the BIOS being flashed.
Click to expand...

The problem is that the existing BIOS that people have is not doing such checks. As shipped, there was not the thought that support would be bifurcated and thus require CPU validation / confirmation.
Hopefully it isn't an issue, as the folks willing to do an in-the-field CPU upgrade would also be savvy to these issues. But - it's a potential of a "my system is bricked" so is a little scary.
 
Face it: AMD is at the pinnacle of performance, and their engineers are going to have a boring time trying to integrate ancient chipset limitations into their brand new CPUs. They've already given their older motherboards a nice treat with their "Conquest of Intel" CPU series, so why hold them back at this point?

Release the wolves.

I'd love to see this next era of CPUs actually perform an absolute cut of all backwards compatibility, and start completely fresh, removing all aging, obsolete, and unoptimized microcode and structures, such that their newest socket actually provides a significant step forward, taking us into a new glorious era of advanced technology.
 
Master_shake_ said:
Size limit is a real limitation.

You can't just add space to a BIOS chip.
Click to expand...
There are 400 series MBOs with chips the same size as 500 series. And you can just remove support for Zen, Zen+ and Zen2 chips for a Zen3 version and fit into a smaller BIOS chip.

schoolslave said:
I could be wrong, but doesn't BIOS flashing require an active CPU?
Click to expand...
Most better motherboards don't.
 
Aegir said:
Face it: AMD is at the pinnacle of performance, and their engineers are going to have a boring time trying to integrate ancient chipset limitations into their brand new CPUs. They've already given their older motherboards a nice treat with their "Conquest of Intel" CPU series, so why hold them back at this point?

Release the wolves.

I'd love to see this next era of CPUs actually perform an absolute cut of all backwards compatibility, and start completely fresh, removing all aging, obsolete, and unoptimized microcode and structures, such that their newest socket actually provides a significant step forward, taking us into a new glorious era of advanced technology.
Click to expand...
I also sometimes fantasize about non-existing problems.
 
Aegir said:
I'd love to see this next era of CPUs actually perform an absolute cut of all backwards compatibility, and start completely fresh, removing all aging, obsolete, and unoptimized microcode and structures, such that their newest socket actually provides a significant step forward, taking us into a new glorious era of advanced technology.
Click to expand...
Go buy an Itanium, I hear they're awesome! No baggage to worry about there. (insert: Vietnam helicopter flashback meme).

The concerns about the impact of legacy are mostly overblown. It's not nothing, but it's also not really what many think it is. You do spend a percentage in the decode phase to translate to native micro-ops, but it isn't really a big deal. And the funny thing - it's becoming less of a big deal all the time, as that relatively fixed cost becomes a smaller percentage of the overall effort / die / cost.
 
Aegir said:
Face it: AMD is at the pinnacle of performance, and their engineers are going to have a boring time trying to integrate ancient chipset limitations into their brand new CPUs. They've already given their older motherboards a nice treat with their "Conquest of Intel" CPU series, so why hold them back at this point?

Release the wolves.

I'd love to see this next era of CPUs actually perform an absolute cut of all backwards compatibility, and start completely fresh, removing all aging, obsolete, and unoptimized microcode and structures, such that their newest socket actually provides a significant step forward, taking us into a new glorious era of advanced technology.
Click to expand...
Except if the microcode is the same and it clocks the same... It doesn't really matter how many other CPUs are supported. You could remove all but a single CPU from the bios and it wouldn't make it run any better. The only thing the new CPU brings different is pcie 4.0, but since pcie spec says it has to be backward compatible then it doesn't make any difference.
 
  • Like
Reactions: Meeho
like this
Ready4Dis said:
Except if the microcode is the same and it clocks the same... It doesn't really matter how many other CPUs are supported. You could remove all but a single CPU from the bios and it wouldn't make it run any better. The only thing the new CPU brings different is pcie 4.0, but since pcie spec says it has to be backward compatible then it doesn't make any difference.
Click to expand...

I'm referring to more general backwards compatibility sacrifices, with also a large spoonful of hyperbole.
 
noko said:
Any board with flashback bios, where you can flash the bios even without a CPU would never have a bios size problem if the bios is selected for a given Ryzen version(s). There were bios's that lacked Ryzen 1 support just for this reason. The reasoning for all boards not being able to support is really BS.

I used PCIe 4 for 3-4 months, x370, made zero difference on the 5700XT performance, issues etc. Many motherboards had very good tracing for PCIe that would support it. Should have been left to the manufacturers. I can see motherboard makers maybe pushing the issue so they can sell enough motherboards. Also the motherboard manufacturers would also had to update the bios, support the users in an endless cycle for aged motherboards which may have other issues not related to the bios changes.

Personally I think AMD could sell more CPUs if most of the mid to upper end AM4 boards would support all AM4 CPU's but the issues covering 4 chipsets, X370, X470, X570, X670 would probably make that painful. When you have 10s of thousands complaining how crappy the Ryzen 4 CPUs are bacause it will not post on a B350 board or one of the VRMs caught fire, AMD/Board Manufacturers has to make a call.
Click to expand...
AMD doesn’t need to sell more CPU’s they sold more than 90% of their production runs that doesn’t leave a lot of room to move up and their production runs are limited. If they sell more than that cuts into any reserve for RMA’s, lost shipments, etc. AMD has a pretty good grasp on what they are doing.
 
gigatexal said:
they can't just keep microcode for the 2xxx to 4xxx series and omit the 1xxx chips?
Click to expand...
I think they might be doing something like that since the bios update says to not use older cpu's.
Asrock ab350 pro4 bios 6.00, 6.20, 6.30 state this:
Update AMD AGESA Combo-AM4 1.0.0.4 Patch B
*ASRock do NOT recommend updating this BIOS if Pinnacle, Raven, Summit or Bristol Ridge CPU is being used on your system.
*Before updating this BIOS, please also read the description in previous BIOS version.
 
Aegir said:
I'm referring to more general backwards compatibility sacrifices, with also a large spoonful of hyperbole.
Click to expand...
I'm not sure which backwards compatibility you're talking about then... DDR4 is current, it's not like they have dual memory controllers or anything. PCIe has to be backwards compatible, so not much they can do there. All other peripherals are attached to the PCIe bus, so that is all irrelevant. If you mean they should have ditched AM4 earlier, well they did a lot of work getting the signalling right so if it works why mess it up unless there is good reason (like ddr5 comes out or something). They use 1331 pins while Intel uses around 1150... It's not as if they are lacking pins for power or signals. Rather than break backward compatibility they decided to overbuild the socket the first time so it would last as long as ddr4... They originally probably thought ddr5 was not going to get delayed and intended zen3 to be on a new platform. This fell through and AM4 is still viable.
 
The BIOS size limitation claim is 100% BS. It is true that there is not enough room in some motherboards BIOS to have support for all socket AM4 CPUs. However, as some have pointed out there is the option of removing support for older CPUs like Zen 1 to add support for newer ones. Bottom line, this is nothing but a convenient excuse by AMD to get out of adding support for Zen 3 to older 300 and 400 series boards.

That said, I understand why they are doing this. As someone who owns a 300 series board and experienced first hand the issues of upgrading to Zen 2, I know that things did not go smoothly. There were bricked motherboards, motherboards that lacked a flasbhack feature or the ability to flash without a CPU installed, boards that would not boot consistently with a Zen 2 chip, and CPUs that were not running at their rated speeds. It took a couple months of troubleshooting to resolve some of these issues. AMD had to send out loaner CPUs to customers who needed one to flash their boards in some situations. Motherboard makers had to deal with irate customers and a bunch of RMAs. It was a complete mess. There was a lot of bad press. Then there was the whole PCIE 4.0 fiasco for the cherry on top.

I believe this is THE REASON that AMD will not allow 300 or 400 series boards to run Zen 3. They are going to keep things much more straightforward at launch. Want Zen 3? Buy a 500 series board. Simple and easy.

I'm hoping they relent or MSI and others can release BIOS updates that support Zen 3 on older boards, but I'm not holding my breath. At this point I'm thinking my Ryzen 3600 is my new 2500k and I'll be using it for the long haul. That's probably for the best anyway. The upgrade from a 1600x to a 3600 was significant and just one part to replace at a pretty low cost. It remains to be seen how much there is to be gained by going to a 4600 or similar. Also, I'm feeling way more GPU bound in general these days.
 
I don't have an issue with them blocking PCIe 4.0 on older boards because some manufacturers were trying to enable it on it on boards that weren't really up to it spec wise and some boards that did have the specs had some occasional issues. AMD could have created a certification program and the occasional issues were likely due to teething issues that could have been fixed with some BIOS updates but I can understand why AMD wouldn't want to deal with the headaches. More importantly I don't recall them ever advertising or even hinting at PCIe 4.0 when the 400 series boards came out.

On the other hand if the report that they're blocking it on all boards before x570 because of BIOS chip size(as if they're all the same) then that's some straight up bullshit. They have heavily marketed the idea that their CPUs and chipsets are backwards and forwards compatible across the socket as long as a board can handle it electrically and gets a BIOS update. Changing that policy so late in the socket cycle without good reason or advance warning is a shitty move at the very least but I could also see it opening them up to a potential class action suit.
 
Im sad. Guess Ill just wait for Zen4 or the Intel Equivelent on my next complete Overhaul.
 
Meeho said:
AMD proved with the PCIE 4.0 B450/X470 removal that they are not above bullshit, and countless times before.
Click to expand...
That was due to incompatibility and stability issues due to trace length/board implementation primarily. Otherwise, then people would just bitch that AMD SUXX DRIVERS BAD etc otherwise... I think restricting it did less damage than the other option as much as I'd like to have PCIe4.0 on a pre-500 board.
MSI also supported win7 on zen+ and I used it thanks to their drivers, AMD didn't officially support that. So I think in this case we may also see some limited 'beta' bioses supporting 3000 series on those with enough eprom to support it.

Lmao at juantel showing up again though, the loserbenchmark of CPU forum shilling in this thread of all threads.


Way I see this is wording. They said through 2020. If Zen3 is delayed to 2021 as it is supposedly, then...? They kept their word. Of course, I'd love to have the option to run a beta bios or similar if needed, same with PCIe4, but you have to make do sometimes in this world. They did a pretty good run already for many years of socket support so if this is the worst part then damn, it's still much better than the competition.
But they deserve the flack and hopefully will change their decision for the good of consumers. People are still going to buy new boards, thing is there are almost no AM4 boards left in USA supposedly (likely spreading elsewhere too), so it doesn't really matter as much at this point.. you'll have to wait anyway.
 
N4CR said:
That was due to incompatibility and stability issues due to trace length/board implementation primarily. Otherwise, then people would just bitch that AMD SUXX DRIVERS BAD etc otherwise... I think restricting it did less damage than the other option as much as I'd like to have PCIe4.0 on a pre-500 board.
Click to expand...
That was just AMD's excuse. MBO vendors both had top quality SKUs and verification process to see if it works or not. And if AMD feared that any problems would come back to them, they could just state it's not officially supported and have the BIOS be designated as beta. But it's a made up potential problem, as any 500 series board could be made with bad PCIE traces and the blame would certainly be on the MB company, not AMD.

MSI also supported win7 on zen+ and I used it thanks to their drivers, AMD didn't officially support that. So I think in this case we may also see some limited 'beta' bioses supporting 3000 series on those with enough eprom to support it.
Click to expand...
And would you have blamed AMD if it didn't work?

In any case, it can't be justified this time with AMD blaming BIOS size when there are both 400 series boards with the sams size and the option to segment the BIOS support per generation like had been done before.
 
have had the itch to swap my z390/8700k out and try AMD again for the first time in 10+ years, but being this is the last AM4 socket chip they will make. Might just wait a little longer.
 
GotNoRice said:
Aren't there x370 and B350 boards that support 3000 series? Because according to that picture, that should not work either.

I think that board makers do have a bit of latitude in this. There could easily be "optional" bios versions that only support certain processors in order to free up room. You probably won't see that on most boards, but maybe higher-end boards.
Click to expand...
I think my Asrock ITX has a limitation when going to 3000, ifI updated the bios I would lose support for some earlier CPUs. I could be wrong but they advised against it.

https://www.asrock.com/mb/AMD/Fatal1ty AB350 Gaming-ITXac/index.asp


Update AMD AGESA Combo-AM4 1.0.0.4 Patch B

*ASRock do NOT recommend updating this BIOS if Pinnacle, Raven, Summit or Bristol Ridge CPU is being used on your system.
*Before updating this BIOS, please also read the description in previous BIOS version.
 
Aegir said:
I'd love to see this next era of CPUs actually perform an absolute cut of all backwards compatibility, and start completely fresh, removing all aging, obsolete, and unoptimized microcode and structures, such that their newest socket actually provides a significant step forward, taking us into a new glorious era of advanced technology.
Click to expand...

ARM64 then. :)
 
we can keep bitching about AMD pulling this stunt or we can move on -- who are we to change the mind of AMD?

so who wants to sell me their x570 ?
 
juanrga said:
ARM64 then. :)
Click to expand...

Just to avoid any misinformation. AFAIK there is only one ARM socket supporting a single generation of product. The ThunderX2. Which they list as an "Optionally with LGA 4077 socket"

Lets not go there.
 
oMek said:
have had the itch to swap my z390/8700k out and try AMD again for the first time in 10+ years, but being this is the last AM4 socket chip they will make. Might just wait a little longer.
Click to expand...
To AMDs benefit, this is Intel's last chip until a new socket also... ;).
 
kirbyrj said:
ASMedia is making the B550. They made the X570.
Click to expand...

I thought that the X570 was the only chipset that ASMedia _didn't_ make:

https://www.techpowerup.com/251964/...-supplier-to-amd-but-x570-an-in-house-chipset

That's what makes all of this absolute horsepucky. All of these chipsets are basically the same damned silicon with only slight revisions. All B550 brings to the table over the older chipsets is AMD giving their blessing to allow the motherboard manufacturers to support PCIe 4.0 from the CPU on 3000+ chips.
 
snclawson said:
I thought that the X570 was the only chipset that ASMedia _didn't_ make:

https://www.techpowerup.com/251964/...-supplier-to-amd-but-x570-an-in-house-chipset

That's what makes all of this absolute horsepucky. All of these chipsets are basically the same damned silicon with only slight revisions. All B550 brings to the table over the older chipsets is AMD giving their blessing to allow the motherboard manufacturers to support PCIe 4.0 from the CPU on 3000+ chips.
Click to expand...

I was responding to a previous post. In context, Asmedia makes B550 while AMD made x570.
 
Chimpee said:
Couldn't motherboard manufacturer in theory just give up supporting Ryzen 1000 and 2000 CPUs to increase ROM space for Ryzen 3000 and 4000?
Click to expand...
That has effectively been done. When I upgraded BIOS on my CHVI to support the 3900X, my 2700 couldn't run Windows anymore.
 
Hardware Unboxed makes a case for why AMD is wrong about this decision and is screwing some of their customers (400 series specifically)



Updates: AMD has refrained from commenting and therefore addressing any of our questions at this point, but we have had some interesting conversations with a few of their partners. Again to be perfectly clear on this point, board partners cannot support Ryzen 4000 series processors on 400-series motherboards without AMD’s help, it’s simply not possible. So don’t expect an AIB to crack the code and open up support, again without AMD’s support it’s not going to happen. It does seem as though this was a recent decision by AMD and their partners found out the same time we did, so that’s truly bizarre, but then given the last few product releases from AMD it’s getting harder and harder to be suprised by this stuff. I’ve also had industry contacts confirm that the AMD BIOS excuse is rubbish and that simple workarounds are possible, just like the one I discussed. In one example there would be a single large BIOS file that you download, then upon flashing you select the CPU series you want to support and it flashes the appropriate code. So at this point it’s now up to the rest of the community to pressure AMD into changing this decision and to open up support for 400-series boards. You guys had better believe that if you give them an inch, they’ll take a mile and we’ll be back to where we were just a few years ago.
Click to expand...
 
Last edited:
I think part of the reason AMD is doing this is to get away from the CPU loaner program. Even if they were to follow Hardware Unboxed plan to have multiple BIOS options available, for new motherboards you would still need at least one generation of CPU's available in both BIOS's which would mean the CPU loaner program chip would need to fall into that category, which it currently does not. The loaner CPU is a GEN1 Ryzen based chip. They would have to maintain compatibility with that chip in the newer BIOS as well so the flash would finish. Then consider if they ever need to change what CPU, or the situation where a user may be trying to flash to a BIOS that doesn't support their current CPU, etc.
 
I will worry about this when I see what ASUS plans when these chips are close to launch.
 
You must log in or register to reply here.
Back
Top