The bar ends at the beginning of 2020. Look carefully. The whole chart begins at 2017 when Ryzen was launched, and the last section starts at 2019 and then ends at 2020.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD Zen 3 (5000 Series), rumored 17% IPC gain.
- Thread starter AStaley
- Start date
- Status
The bar ends at the beginning of 2020. Look carefully. The whole chart begins at 2017 when Ryzen was launched, and the last section starts at 2019 and then ends at 2020.
Caporegime
People are frustrated because they were savey enough to have bought half decent B450 / X470 boards thinking they would be future proofed for Zen 3 and now AMD have dropped this bombshell on them which means they have to buy another board for Zen 3.
That's completely understandable, but AMD never said Zen 3 would be supported on older boards.
AMD experimented with mixed support "Until 2020" for business reasons, offering mixed support is a good thing to offer if you want to bring people on side and i'm sure had it worked they would have loved to continue it, but it didn't, its hurting their business, its hurting their partners and probably retailers.
Last edited:
Soldato
I fully get the frustration because I share it, but that doesn't change the fact people seem to be missing the point and focussing on ROM sizes that would work in their specific instance, rather than the bigger picture of business, partners, retailers, Joe Public, perceptions, etc. etc. etc.
Caporegime
Which would mean something if AM4 was dead, but it's going through 2020 at least.
The graph is for the socket
Caporegime
Also the fact that plenty of the X570 motherboards have 16MB BIOS chips. Its been show Bristol Ridge to Zen2 support takes up 8MB~12MB of memory.
As HUB said,they have talked to people in the industry too:
The fact is they are correct,and they wouldn't be saying what they say,as AMD could easily just cut them out for reviews,which they are capable of doing.
Also a lot of the MSI motherboards support BIOS flashback,meaning you don't have to use a big BIOS,ie,no CPU is required an only a Zen3 BIOS is required. The fact is they are withholding the microcode from motherboard OEMs,who should be able to make the decision on what motherboards can be supported.
It was the same with PCI-E 4.0 support,OEMs made a list of what motherboards were supported,and which ones could support the PCI-E slot and which could only do the M2 slot.
People don't realise,if anything is a problem,its on the OEMs,and they wouldn't have made support if they didn't feel confident. After all they are designing the motherboards themselves. They should have a better clue than AMD themselves on their own motherboards.
Caporegime
What if AMD renamed the socket for 500 series boards?
Caporegime
This assumes the Zen 3 BIOS alone is small enough to fit on on these older boards with the small ROM.
This is the problem, people who have no clue what they are talking about assuming their limited and probably incorrect knowledge as fact so the only possible explanation is: AMD are being cynical. You're smarter than that CAP.
Caporegime
Both motherboard OEMs and system integrators were told by AMD that B450 would support Zen3. They literally told a system integrator who buys 1000s,maybe 10000s of its motherboards each year,it would work The fact AMD communicated it to it's own industrial partners,who appeared to only find out the same day we did. Forget the end users,they are confusing their own partners.
This is 100% the fault of AMD,not system integrators fault,not the end users fault and neither is even the motherboard OEM's fault. This is AMD just again doing what it does best,just pull out random stuff at the last minute. They did it with the RX5600XT launch,when their own board partners had to contend with a last minute BIOS update,AMD kept quiet about. This is why the RX5600XT was a mess,you couldn't tell whether you were getting a slower card or a faster card.
People deconstructed BIOSes for the last year. The BIOSes are between 8MB~12MB for Bristol Ridge to Zen2 support,ie,4 different generations of CPUs,and 3 distinctly different designs.AMD has admitted by allowing X570 motherboards to have 16MB BIOSes,you don't actually need huge BIOSes to run Zen3,or are we going to find out X570 motherboards next year will have the same problem??
There is zero reason for them to stop OEMs from implementing their own support,but they are holding back on CPU microcode.
That is what people are missing,no microcode is being supplied,so OEMs themselves can't make their own decisions. All they are doing is passing the buck to their own partners.
Some of you really need to listen to Martini. AMD will pull such stunts - they did it for socket 754,socket AM2,socket AM3,socket FM1 and socket FM2. I have seen how much people got annoyed when AMD did the same things,and they didn't give a damn back then either. Maybe some of you are too young to remember these things.
Caporegime
And some of them do, you're assuming from what was said that "All of them do".
And you're still ignoring the mess of expecting users to understand that they need to #### about with first having access to older CPU's to then flash the BIOS. You haven't addressed this point at all.
Don
I must remember most of those socket changes (and a few others) quite differently to you then.
Caporegime
Put yourself in AMD's shoes.
You're selling a product, it requires two components to work, you only have control of one of those components, you allow the other component in older EOL form to support your product but its up to the product manufacture to enable that support and if they do its up to you the consumer to modify that product, possibly bricking it and rendering it out of warranty in order to get it to work with the component you make.
That's unacceptable...
You're selling a product, it requires two components to work, you only have control of one of those components, you allow the other component in older EOL form to support your product but its up to the product manufacture to enable that support and if they do its up to you the consumer to modify that product, possibly bricking it and rendering it out of warranty in order to get it to work with the component you make.
That's unacceptable...
Caporegime
OEMs will decide what motherboards will get supported as they are the ones who have to support the motherboards,not AMD.
They are doing exactly what Intel did with CFL as Asus went on record,saying Intel actually stopped them from implementing support on earlier motherboards. Then we found out certain laptop OEMs implemented support,so people on this forum have CFL CPUs in Z170 motherboards working fine.
They even mislead a system integrator.
They didn't consult motherboard OEMs,they didn't consult system integrators - they pulled another stunt. So in the end some of you have not addressed the point,AMD mislead it's own system integrators and motherboard partners,now opening them up to liabilities. Its bad enough doing it to end customers,they are now doing it commercial partners too.
In the end I am not going to agree with you on some of you about this. I have seen so many here have a go at Intel doing the same stuff,but AMD does not get a free pass then. Either criticise both,or don't criticise any of them for doing things like this. You can't have it both ways!
Maybe they were more fondly remembered by some,but I think Martini is hinting back to some of them when he said what he said.
AMD introduced socket 754 and socket 939 concurrently,and then dropped 754 like a lead balloon leaving people high and dry. They had some nice CPUs no doubt,but their were arguments about what socket to use,like the whole B450 vs X570 discussion now.
Socket 754 was a mainstream socket,so many enthusiasts probably were on socket 939. So like here the whole should you get the cheaper socket/motherboard argument lost!
But either way the two socket thing for the Athlon 64 was a bit silly IMHO.
Then you had the quad FX platform - that was hilarious and short lived too. Introduced in 2006,and cancelled in 2007.That was a dead end
Then with AM2,closer to the Phenom II launch there was hints the Phenom II CPUs would work on them,and then AMD launched AM2+ but some motherboards got hacked support for the Phenom II CPUs.
Then there were hints the newer AM3 motherboards would have better socket compatibility with future AMD CPUs. Then AMD launched AM3+ around 5~6 months previous to the Bulldozer launch. Look at people being dissapointed by AM3 motherboards not working OK with Bulldozer:
https://forums.overclockers.co.uk/threads/support-bulldozer-gigabite-880gb-ud2h.18288197
Not the first or last thread on the whole AM3/AM3+ thing. The same why my "insert name AM3" motherboard won't work. Again more hacked support eventually.
They probably dodged a bullet there TBF!
This is the same period a senior AMD guy said Bulldozer would have higher IPC than a Phenom II on xtremesystems forums. That was another own goal there.
The FM sockets you would only care about,if you ever used them - probably not most here.
Socket FM1 - one generation socket and I have a Llano system myself.
Socket FM2 changed,and then people thought it might be longer lived,but then AMD announced if you want to move to Steamroller based APUs,you need socket FM2+ and the worse thing is the only difference between the motherboards was a few pins on the socket. One generation again. I had both side by side to have a good look at.
The FM2 was probably the worst of all of them,especially as the later CPUs required less power than the Piledriver based APUs. This is probably why some of the FM2+ versions(mini-ITX) had no VRM heatsinks,and the FM2 versions did.
Last edited:
Soldato
It also doesn't make sense for AMD to restrict Zen 3 only to 500 series boards artificially because they would sell a lot more chips if they where also compatible with previous generation boards.
And as some one pointed out a while back, when considering options and thinking of going Zen 3 plus board that also puts Intel's offering on the table at the same time.
And as some one pointed out a while back, when considering options and thinking of going Zen 3 plus board that also puts Intel's offering on the table at the same time.
Caporegime
OEMs will decide what motherboards will get supported as they are the ones who have to support the motherboards,not AMD. tHEY
They are doing exactly what Intel did with CFL as Asus went on record,saying Intel actually stopped them from implementing support on earlier motherboards. Then we found out certain laptop OEMs implemented support,so people on this forum have CFL CPUs in Z170 motherboards working fine.
They even mislead a system integrator.
They didn't consult motherboard OEMs,they didn't consult system integrators - they pulled another stunt. So in the end some of you have not addressed the point,AMD mislead it's own system integrators and motherboard partners,now opening them up to liabilities. Its bad enough doing it to end customers,they are now doing it commercial partners too.
In the end I am not going to agree with you on some of you about this. I have seen so many here have a go at Intel doing the same stuff,but AMD does not get a free pass then. Either criticise both,or don't criticise any of them for doing things like this. You can't have it both ways!
Not really. AMD introduced socket 754 and socket 939 concurrenly,and then dropped 754 quiet suddenly leaving people high and dry.
Then with AM2,closer to the Phenom II launch there was hints the Phenom II CPUs would work on them,and then AMD launched AM2+ but some motherboards got hacked support for the Phenom II CPUs.
Then there were hints the newer AM3 motherboards would have better socket compatability with future AMD CPUs. Then AMD launched AM3+ around 5~6 months previous to the Bulldozer launch. This is the same period a senior AMD guy said Bulldozer would have higher IPC than a Phenom II on xstremesystems forums. That was another own goal there.
Socket FM1 - one generation socket. Socket FM2 changed,and then people thought it might be longer lived,but then AMD announced if you want to move to Steamroller based APUs,you need socket FM2+ and the worse thing is the only difference between the motherboards was a few pins on the socket. I had both side by side to have a good look at.
The FM2 was probably the worst of all of them,especially as the later CPUs required less power than the Piledriver based APUs.
Intel rename the same CPU and stick it on a different socket, that's not what AMD are doing, Zen 2 is not Zen and Zen 3 is not Zen 2.
You keep circling back to this fact that older boards can support Zen 3, which is factually correct, the socket pins are the same, The BIOS is not the same, we can be pretty sure of that because the BIOS for Zen is not the same for Zen+ or Zen 2, its why you need to flash a new BIOS on these older boards, you're still ignoring the problems with making customers do that.
Caporegime
The fact is if they are telling their own motherboard OEMs that it works,their own system integrators it will work,then it must work. That was AMD's words to those companies. These companies then found out via the press release we all heard that it does not work. So these companies now are opened up to liabilities for advertising such support,and also in the terms of system integrators,they also have bought tons of motherboards which probably now have less of a use to them,then before the announcement. Also if so many X570 motherboards use 16MB BIOSes,it indicates to me that it's sufficient to cover Zen3 compatiblity and again its upto the OEMs.
AMD is holding back from motherboard OEMs,making that decision for themselves. This is not motherboard OEMs holding back updates after AMD giving them the microcode,its AMD doing this. This means people like MSI who use 32MB BIOS chips on their MAX series motherboards and BIOS flashback could do this. This is why they advertised support for all AM4 socket CPU,as they literally made a range of improved motherboards for this. We can't even have a go at OEMs,if AMD doesn't give them a chance.
The problem is if they are dicking around with system integrators and their own motherboard OEMs,then how is this going to play out in the future?? It does not seem the wisest thing to do,and if these motherboard OEMs are lying,then AMD needs to clarify this.
This reminds so much of the 5600XT launch,and all the BIOS stuff for that,and landing it on OEMs laps,at the last moment. This is really not the best thing to do IMHO,as even if they follow AMD reference specifications it means diddly squat. OEMs will just put more effort into other products as they know where they stand.
Caporegime
Right. AMD would be lying to those OEM's if they had said there is no way Zen 3 would work on older boards, that's simply not true, they will work in those MAX boards at least, doesn't mean they will work in all older boards, i don't think ASRock could get Zen 3 to work in my board even if they wanted to, they barely managed to get Zen 2 to work in it and they only did that with reduced BIOS features and the removal of all other Ryzen generation support, they squeezed it in.
AMD are simply telling them the truth of the matter, they also explained why they decided not to add Zen 3 support despite this, its a good faith pragmatic reason. You're ignoring that.
Just to point out the other side of this IIRC X570 boards dont support original Ryzen 1xxx series CPUs.
Seem to recall it was hit and miss with some X/B3xx boards supporting the 2xxx chips.
Caporegime
No they told system integrators and motherboard OEMs it would work. Schenker is a reasonably well known OEM in Germany. They literally said AMD told them officially it WOULD work,and then found out the same way we all did. You seem to be ignoring this.
AMD told Schenker "though official channels". So in the end with HUB,saying motherboard OEMs have also found out at the same time. If AMD has pragmatic reasons,so does Intel despite Asus also telling us otherwise. No different.
The fact is they pulled the same stunt with the RX5600XT after OEMs had validated the cards,etc. They plonked a new BIOS just before launch,which most OEMs couldn't actually test for it. The fact is if this is what you do for your own partners,then we are nothing in the scheme of things,as we don't have direct contact with AMD.
OTH,there are those which support later chips too. You can make selective BIOSes to target individial users.
So if a Zen3 compatible BIOS can support Zen2/Zen3,nothing stopping Zen+ to Zen3,etc and that is what some here don't realise. The package you download can selectively support specific CPUs.
But the problem is again,the reference default BIOS chip size,is mandated by AMD,so as Martini said that is really also AMD not making a more robust base specification.
Like I said both system integrators and apparently motherboard OEMs thought it was fine,so basically AMD really needs to work on its corporate communication then,because that does not bode well IMHO. Doing these sorts of last minute decisions just like their 5600XT launch,ie,all the last minute BIOSes,does not really help their cause.We always wonder why laptops and prebuilt PCs,seem to have more Intel CPUs,and Nvidia graphics cards,if this is the way they communicate things,its not entirely surprising to me.
Last edited:
I certainly agree and understand what you are **** off about, I do see why AMD has done this too. The confusion that arised on a lot of forums with regards having to do bios updates prior to install the next gen CPUs was high. Even worse for those that bought a 350/450 board only to find it required a update for the CPU they have. They got some stick for it to IIRC.
Don't make it any better for people like yourself on previous gen boards though.
Associate
AMD know this isn't a morally defensible move when they are trying to justify socket compatibility lifetime starting with Excavator.
This is a typical corporate muppet 'solution' to some problem they have likely created themselves and can't think a way around it. Too busy managing upwards probably. It is extremely telling that whoever communicated this thought their flimsy excuse was watertight. Says a lot about their competence and understanding of the technology they represent. (It seems a common theme these days that most of the people who work for technology corporates don't know that much about the technology that pays their wages except for soundbites and/or marketing.)
If AMD had a technologically-sound reason for this we'd have heard it seep out before now from partners and it would be the ONLY thing AMD was communicating.
If PCIe4 was such a big deal they should have changed the socket to AM4+ and telegraphed it. Of course that would have made people hold off for the next latest/greatest boards and that would have stopped revenue growth dead in it's tracks. Notice they announced this decision not long after Q1 numbers had been communicated. The numbers would have cratered if B450 was known to not have been compatible. AMD caused this by not getting B550 out in a timely manner and the consumer has to pay for their failure to execute.Not for the first time and probably not the last.
I can't see that many people getting that much real-world benefit out of PCIe4 yet anyway. PCIe4 is currently just AMD ePeen the consumers are paying for....in more ways than just financially it seems now.
AMD should already have been splitting out AGESA for the different CPUs. TheStilt has advocated not upgrading bios past a certain revision on each CPU for a long time now. We currently have to define hardware specifics to obtain the correct drivers on AMD's website so doing the same for the correct BIOS/CPU combination is not some new concept that users couldn't understand. There are motherboard manufacturers out there who state on their bios download pages ALREADY not to upgrade to specific bios if using an older version of AM4 CPU.
There are probably loads of pragmatic, practical solutions here that people would accept if informed correctly, some of them may even incur a cost. That is, if what we're being told is true.
AMD have just done a great job of offloading some of the goodwill and mindshare they have built up since Ryzen was introduced. They've been getting away with it for a bit now (they need shot for the APU generational naming, got to feel a bit sorry for TR early-adopters) but it looks like getting away with it just made them bolder and they didn't know when they would go too far. I've wondered for a while if someone from the GPU side has been promoted into the CPU side as the very recent decisions and launches of CPUs have felt very much as amateurish as the GPU launches have been for what feels like ever. AMD isn't cheap anymore (nominal, not relative) so they can't afford to take their customers for granted if they want to maintain their upward trajectory. What they have just done is take their customers for granted.
With the world the way it is where a huge amount of people's financial security is either already gone or under threat AMD couldn't have picked a worse time to make people think twice about what to spend their money on.....especially for a major launch item. This isn't blowing both feet off. This is removing everything just below the chest.
As a long-term AMD/Radeon-only buyer my planned Zen3 household upgrade (Desktop - CPU/GPU, Server - APU, HTPC - APU) is now indefinitely on hold and i'm considering options. I don't only buy/build for myself so all other requests that come to me are on hold too.
Surprised there isn't an obvious poll on the forum about this subject.
I hope you're reading this AMD.
This is a typical corporate muppet 'solution' to some problem they have likely created themselves and can't think a way around it. Too busy managing upwards probably. It is extremely telling that whoever communicated this thought their flimsy excuse was watertight. Says a lot about their competence and understanding of the technology they represent. (It seems a common theme these days that most of the people who work for technology corporates don't know that much about the technology that pays their wages except for soundbites and/or marketing.)
If AMD had a technologically-sound reason for this we'd have heard it seep out before now from partners and it would be the ONLY thing AMD was communicating.
If PCIe4 was such a big deal they should have changed the socket to AM4+ and telegraphed it. Of course that would have made people hold off for the next latest/greatest boards and that would have stopped revenue growth dead in it's tracks. Notice they announced this decision not long after Q1 numbers had been communicated. The numbers would have cratered if B450 was known to not have been compatible. AMD caused this by not getting B550 out in a timely manner and the consumer has to pay for their failure to execute.Not for the first time and probably not the last.
I can't see that many people getting that much real-world benefit out of PCIe4 yet anyway. PCIe4 is currently just AMD ePeen the consumers are paying for....in more ways than just financially it seems now.
AMD should already have been splitting out AGESA for the different CPUs. TheStilt has advocated not upgrading bios past a certain revision on each CPU for a long time now. We currently have to define hardware specifics to obtain the correct drivers on AMD's website so doing the same for the correct BIOS/CPU combination is not some new concept that users couldn't understand. There are motherboard manufacturers out there who state on their bios download pages ALREADY not to upgrade to specific bios if using an older version of AM4 CPU.
There are probably loads of pragmatic, practical solutions here that people would accept if informed correctly, some of them may even incur a cost. That is, if what we're being told is true.
AMD have just done a great job of offloading some of the goodwill and mindshare they have built up since Ryzen was introduced. They've been getting away with it for a bit now (they need shot for the APU generational naming, got to feel a bit sorry for TR early-adopters) but it looks like getting away with it just made them bolder and they didn't know when they would go too far. I've wondered for a while if someone from the GPU side has been promoted into the CPU side as the very recent decisions and launches of CPUs have felt very much as amateurish as the GPU launches have been for what feels like ever. AMD isn't cheap anymore (nominal, not relative) so they can't afford to take their customers for granted if they want to maintain their upward trajectory. What they have just done is take their customers for granted.
With the world the way it is where a huge amount of people's financial security is either already gone or under threat AMD couldn't have picked a worse time to make people think twice about what to spend their money on.....especially for a major launch item. This isn't blowing both feet off. This is removing everything just below the chest.
As a long-term AMD/Radeon-only buyer my planned Zen3 household upgrade (Desktop - CPU/GPU, Server - APU, HTPC - APU) is now indefinitely on hold and i'm considering options. I don't only buy/build for myself so all other requests that come to me are on hold too.
Surprised there isn't an obvious poll on the forum about this subject.
I hope you're reading this AMD.
- Status