MSI may not support Zen2 on 300-series

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
25,229
Certainly a dick move if true. Then again, the "rumblings" about the "necessary" 300 series chipset turned out to be true also.
 

N4CR

Supreme [H]ardness
Joined
Oct 17, 2011
Messages
4,450
This is just marketing testing market to see if people will be pissed off enough or not, in order to evaluate if they need to program BIOS for 300 series or not.
They should do it or else AMD fans will take a massive Micro Shit International all over them.
 
Joined
Dec 30, 2017
Messages
55
Well, for me it was a major factor in buying my B350 Pro Carbon.
AMD promised socket compatibility so I bought what I considered to be a quality board to last me.

If this is true, it will be the last MSi board i ever buy.
 

PhaseNoise

2[H]4U
Joined
May 11, 2005
Messages
2,375
This is just marketing testing market to see if people will be pissed off enough or not, in order to evaluate if they need to program BIOS for 300 series or not.
They should do it or else AMD fans will take a massive Micro Shit International all over them.
My gut feel aligns with yours - this is a trial balloon "leak" which can be easily dismissed if there is enough outrage.
 
  • Like
Reactions: N4CR
like this

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
55,548
Like I said, I just don't think MSI wants to spend the money updating firmware and doing QVL testing of motherboards it no longer produces more than two years after their release.

This is taken from the article link above: "To recap, AMD announced on numerous occasions that it doesn't want to be a greedy clique like its competitor, by forcing motherboard upgrades and promised that socket AM4 motherboards will be backwards and forwards compatible with at least four generations of Ryzen processors, running all the way up to 2020."

Its important to note the wording here. This is something AMD promised, not motherboard manufacturers. This isn't something that MSI necessarily agreed to do and didn't do as far as I know. While Intel maintains strict control over its motherboard partners, AMD doesn't. Even though Z170 could do everything Z270 could do regarding processor compatibility, those companies were forced into limiting support via firmware because Intel said so. I've talked to people in the industry who admitted this was the case. AMD doesn't have the same level of control over its motherboard partners that Intel does. Its very likely AMD made a promise with no real way of forcing motherboard manufacturers to honor it. I know some people think this is ultra-greedy of motherboard manufacturers, but I can't think of any other industry that would carry the same kind of expectations. To put it another way, 3 years in the computing industry is a long time. Ryzen was released in 2017. 2020 would be four years after release. X470 motherboards launched about a year after the X370 and B350 parts did. Effectively, anything MSI produced using those older chipsets was discontinued and replaced by X470 and B450 variants of those chipsets. While those motherboards carry many similarities to their predecessors, they are not identical.

In other words, a motherboard manufacturer adding CPU support to X370 and B350 motherboards through 2020 would involve continued development and QVL testing of CPUs and memory modules three years or more after the product was discontinued and four years after release. Given the plethora of potential models from some manufacturers and the difficulty in doing this, I can't say I blame any manufacturer from dropping support for these older models after a time. I certainly think two years is an acceptable reasonable minimum, or supporting at least two generations of CPU. However, I can't necessarily see doing so for three or four generations as I know something about the difficulty in doing so. You also have to understand, that the VRM's of all X370 and B350 motherboards aren't equal. We may find that some higher end models may be capable of supporting CPU's through this time line and will get that support, but entry level motherboards may not if their VRM's aren't up to the task. For one thing, lower end motherboards often lack the necessary hardware for the range of adjustments the higher end motherboards do. This could be a problem if a new CPU needs to be able to drop to lower ranges than the older board can accommodate or if a new motherboard has to output more power than it was really designed to do over a certain period of time at a given temperature.

I think people often don't understand the difficulty in broad CPU compatibility, nor do they recall all the difficulties in actually trying to leverage it. People seem to be caught up in the idea of buying a platform and upgrading that CPU every year or two and revitalizing their systems on the cheap. Keep in mind that not all motherboard manufacturers in the past updated the BIOS ROMs of AM2 and AM3 motherboards throughout all of the Phenom, Phenom II and Bulldozer product life cycles. This effectively locked some people out of certain upgrades even though the socket would theoretically allow compatibility. While I can speculate as to the reasons for this, the end result was that some customers felt betrayed and cheated by the forced obsolescence. We also had cases where you had some motherboards with cut rate VRM's that wouldn't do 130 watt TDP processors, which were limited to 95 watts. So the electrical limitations with some CPUs were well known. This too pissed off many people. Intel hasn't generally had this issue as it VRM specifications have to be followed to the letter. Again, AMD doesn't have that kind of control.

When your talking about keeping things economical, I understand where people are coming from. However, from the enthusiasts perspective it doesn't make much sense. I don't think most enthusiasts tend to go the route of continually upgrading a CPU on the same motherboard. Take the 890FX for example once again. While it was virtually the same chipset as the 990FX, with only a stepping increase and a name change, the VRM implementation changed a fair amount. You could run Bulldozer CPU's on the older 890FX motherboards but they didn't support the new C-states and power savings features of 990FX and you were limited in terms of overclocking as those older VRM designs weren't optimized or designed for Bulldozer. Just as using a Ryzen 2700X on a X370 motherboard won't give you access to Precision Boost Overdrive and XFR2. Let's also not forget that X370 isn't nearly as good as X470 for RAM compatibility. While I agree its a nice sentiment for a CPU maker to guarantee support for CPU's on a given motherboard platform for four or more years, it isn't entirely realistic and attempts at doing this in the past have been met with all kinds of specific requirements and the results have been a mixed bag for end users.

It will be interesting to see how or if MSI responds to this news and whether or not its true. MSI may also back peddle on this as well and decide to support those older motherboards going forward. I'm also curious what other motherboard manufacturers are going to do here. Will they follow suit or take the opportunity to proclaim compatibility through 2020 and paint MSI in a bad light?
 

whateverer

Gawd
Joined
Nov 2, 2016
Messages
994
How is this a surprise? Precision Boost 2.0 requires a new motherboard chipset to support, and that is really the key performance improvement that added to the 400 series.

Why would you castrate a Zen 2.0 processor without Precision Boost 2.0? It fixes the broke boost system in the first gen Zen, and made the systems a lot more capable under lighter loads.

The people looking to upgrade their current Zen 1.0 systems for pure compute are pretty rare. For everyone else, they would be missing half the upgrade.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
25,229
How is this a surprise? Precision Boost 2.0 requires a new motherboard chipset to support, and that is really the key performance improvement that added to the 400 series.

Why would you castrate a Zen 2.0 processor without Precision Boost 2.0? It fixes the broke boost system in the first gen Zen, and made the systems a lot more capable under lighter loads.

The people looking to upgrade their current Zen 1.0 systems for pure compute are pretty rare. For everyone else, they would be missing half the upgrade.
No it doesn't. The only advantage is StoreMi and maybe PBO on some 4XX boards. At the very least it should support minus PCIe 4 on the 5XX...and that likely won't have a big impact other than marketing.
 
Joined
Mar 3, 2017
Messages
514
Well scratch MSI off my mobo list. Cost of BIOS updates is peanuts. I understand MSI is a small manufacturer, but geez, gotta keep up with the times if you expect to grow. It's AMD, not intel...
 

Chimpee

[H]ard|Gawd
Joined
Jul 6, 2015
Messages
1,480
Well scratch MSI off my mobo list. Cost of BIOS updates is peanuts. I understand MSI is a small manufacturer, but geez, gotta keep up with the times if you expect to grow. It's AMD, not intel...
It is also AMD making the promised not the motherboard manufacturer and I don't see why the board makers have to keep a promise that AMD made. Are we suppose to hold motherboard manufacturer responsible for something AMD promise?
 

sirmonkey1985

[H]ard|DCer of the Month - July 2010
Joined
Sep 13, 2008
Messages
21,769
How is this a surprise? Precision Boost 2.0 requires a new motherboard chipset to support, and that is really the key performance improvement that added to the 400 series.

Why would you castrate a Zen 2.0 processor without Precision Boost 2.0? It fixes the broke boost system in the first gen Zen, and made the systems a lot more capable under lighter loads.

The people looking to upgrade their current Zen 1.0 systems for pure compute are pretty rare. For everyone else, they would be missing half the upgrade.
nope it doesn't.. most of the x370 boards have been updated to support PB2 and PBO. the only thing they don't natively support is storeMI.
 
Joined
Dec 30, 2017
Messages
55
Like I said, I just don't think MSI wants to spend the money updating firmware and doing QVL testing of motherboards it no longer produces more than two years after their release.

This is taken from the article link above: "To recap, AMD announced on numerous occasions that it doesn't want to be a greedy clique like its competitor, by forcing motherboard upgrades and promised that socket AM4 motherboards will be backwards and forwards compatible with at least four generations of Ryzen processors, running all the way up to 2020."

Its important to note the wording here. This is something AMD promised, not motherboard manufacturers. This isn't something that MSI necessarily agreed to do and didn't do as far as I know. While Intel maintains strict control over its motherboard partners, AMD doesn't. Even though Z170 could do everything Z270 could do regarding processor compatibility, those companies were forced into limiting support via firmware because Intel said so. I've talked to people in the industry who admitted this was the case. AMD doesn't have the same level of control over its motherboard partners that Intel does. Its very likely AMD made a promise with no real way of forcing motherboard manufacturers to honor it. I know some people think this is ultra-greedy of motherboard manufacturers, but I can't think of any other industry that would carry the same kind of expectations. To put it another way, 3 years in the computing industry is a long time. Ryzen was released in 2017. 2020 would be four years after release. X470 motherboards launched about a year after the X370 and B350 parts did. Effectively, anything MSI produced using those older chipsets was discontinued and replaced by X470 and B450 variants of those chipsets. While those motherboards carry many similarities to their predecessors, they are not identical.

In other words, a motherboard manufacturer adding CPU support to X370 and B350 motherboards through 2020 would involve continued development and QVL testing of CPUs and memory modules three years or more after the product was discontinued and four years after release. Given the plethora of potential models from some manufacturers and the difficulty in doing this, I can't say I blame any manufacturer from dropping support for these older models after a time. I certainly think two years is an acceptable reasonable minimum, or supporting at least two generations of CPU. However, I can't necessarily see doing so for three or four generations as I know something about the difficulty in doing so. You also have to understand, that the VRM's of all X370 and B350 motherboards aren't equal. We may find that some higher end models may be capable of supporting CPU's through this time line and will get that support, but entry level motherboards may not if their VRM's aren't up to the task. For one thing, lower end motherboards often lack the necessary hardware for the range of adjustments the higher end motherboards do. This could be a problem if a new CPU needs to be able to drop to lower ranges than the older board can accommodate or if a new motherboard has to output more power than it was really designed to do over a certain period of time at a given temperature.

I think people often don't understand the difficulty in broad CPU compatibility, nor do they recall all the difficulties in actually trying to leverage it. People seem to be caught up in the idea of buying a platform and upgrading that CPU every year or two and revitalizing their systems on the cheap. Keep in mind that not all motherboard manufacturers in the past updated the BIOS ROMs of AM2 and AM3 motherboards throughout all of the Phenom, Phenom II and Bulldozer product life cycles. This effectively locked some people out of certain upgrades even though the socket would theoretically allow compatibility. While I can speculate as to the reasons for this, the end result was that some customers felt betrayed and cheated by the forced obsolescence. We also had cases where you had some motherboards with cut rate VRM's that wouldn't do 130 watt TDP processors, which were limited to 95 watts. So the electrical limitations with some CPUs were well known. This too pissed off many people. Intel hasn't generally had this issue as it VRM specifications have to be followed to the letter. Again, AMD doesn't have that kind of control.

When your talking about keeping things economical, I understand where people are coming from. However, from the enthusiasts perspective it doesn't make much sense. I don't think most enthusiasts tend to go the route of continually upgrading a CPU on the same motherboard. Take the 890FX for example once again. While it was virtually the same chipset as the 990FX, with only a stepping increase and a name change, the VRM implementation changed a fair amount. You could run Bulldozer CPU's on the older 890FX motherboards but they didn't support the new C-states and power savings features of 990FX and you were limited in terms of overclocking as those older VRM designs weren't optimized or designed for Bulldozer. Just as using a Ryzen 2700X on a X370 motherboard won't give you access to Precision Boost Overdrive and XFR2. Let's also not forget that X370 isn't nearly as good as X470 for RAM compatibility. While I agree its a nice sentiment for a CPU maker to guarantee support for CPU's on a given motherboard platform for four or more years, it isn't entirely realistic and attempts at doing this in the past have been met with all kinds of specific requirements and the results have been a mixed bag for end users.

It will be interesting to see how or if MSI responds to this news and whether or not its true. MSI may also back peddle on this as well and decide to support those older motherboards going forward. I'm also curious what other motherboard manufacturers are going to do here. Will they follow suit or take the opportunity to proclaim compatibility through 2020 and paint MSI in a bad light?
Then maybe MSi should have qualified their stance on this on Ryzen release, not kept quiet and let motherboards be sold under false pretences.
 
Joined
Mar 3, 2017
Messages
514
It is also AMD making the promised not the motherboard manufacturer and I don't see why the board makers have to keep a promise that AMD made. Are we suppose to hold motherboard manufacturer responsible for something AMD promise?
When you're a board partner, yes, we should hold them accountable. There should be no expectation that every feature works with the new chips, but it should at least run at a basic level of functionality.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
55,548
Well scratch MSI off my mobo list. Cost of BIOS updates is peanuts. I understand MSI is a small manufacturer, but geez, gotta keep up with the times if you expect to grow. It's AMD, not intel...
The cost of BIOS updates isn't peanuts. As I explained above, there is a lot more to it than just updating a few lines of code.

It is also AMD making the promised not the motherboard manufacturer and I don't see why the board makers have to keep a promise that AMD made. Are we suppose to hold motherboard manufacturer responsible for something AMD promise?
I have to wonder if this is even a promise that AMD was in a position to make. That is, did AMD have the agreements in place with motherboard manufacturers to make promises about socket support. AMD can support it on their end, but that doesn't mean that companies like MSI are in any way bound to such promises.

nope it doesn't.. most of the x370 boards have been updated to support PB2 and PBO. the only thing they don't natively support is storeMI.
StoreMI support is a matter of licensing the software. X370 and B350 simply don't include the license for it. From a storage perspective, X370 and X470 are identical.
 

pitingres

Weaksauce
Joined
Jul 25, 2018
Messages
95
BIOS rom size is an issue here, I'm sure. I gather that many boards simply don't have room to store the code and data necessary to support the entire Ryzen line. If that's true of the MSI B350's, that could mean for instance that you couldn't update the BIOS to a Zen 2 version without a Ryzen 2 processor, since you couldn't run 1st and 3rd gen code together. Or alternatively, MSI would have to have a whole selection of BIOS's with various CPU combinations, and that's a non-starter - I can hear the "I bricked my board" calls from here.

I hardly think MSI is doing this because they are lazy and enjoy pissing off customers. I also think that you're going to hear pretty much the same song from the other mobo brands; I've already read a rumbling in that direction from someone plausibly claiming to be an Asus rep. I guess we'll see how it plays out.
 

ir0nw0lf

Supreme [H]ardness
Joined
Feb 7, 2003
Messages
6,347
I was going to post this story yesterday but as far as I can tell it's rated FUD (as of last night) and I didn't want to potentially start a ginormous flame war over it. But it does bring up a good talking point: *if* this is true is there a chance others will do the same?
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
55,548
BIOS rom size is an issue here, I'm sure. I gather that many boards simply don't have room to store the code and data necessary to support the entire Ryzen line. If that's true of the MSI B350's, that could mean for instance that you couldn't update the BIOS to a Zen 2 version without a Ryzen 2 processor, since you couldn't run 1st and 3rd gen code together. Or alternatively, MSI would have to have a whole selection of BIOS's with various CPU combinations, and that's a non-starter - I can hear the "I bricked my board" calls from here.

I hardly think MSI is doing this because they are lazy and enjoy pissing off customers. I also think that you're going to hear pretty much the same song from the other mobo brands; I've already read a rumbling in that direction from someone plausibly claiming to be an Asus rep. I guess we'll see how it plays out.
There aren't that many Ryzen CPU's. AMD motheroards have had to support a much wider array of CPU's with smaller BIOS ROMs than they do today. That's probably not the issue. The issue is that these things have to be tested. You've got to test it with some actual CPU's, memory, and make sure that the flash doesn't brick the board or cause other problems. If there are a lot of motherboard models, than stock of them has to be kept around for testing. Then there is paying all the people that work on these things. People who code, test, etc. Personnel costs are no doubt one of the most expensive parts of the equation.

I was going to post this story yesterday but as far as I can tell it's rated FUD (as of last night) and I didn't want to potentially start a ginormous flame war over it. But it does bring up a good talking point: *if* this is true is there a chance others will do the same?
If this was or is true, then I would wager its because AMD has no contractual way to force motherboard manufacturers to keep socket longevity promises and you better believe other vendors would follow suit.
 

mvmiller12

Gawd
Joined
Aug 7, 2011
Messages
930
My specific recollection is that AMD stated they would stay on the same socket (AM4) through 2020, not that every socket AM4 motherboard ever made would support every AM4 chip ever made.

(But I AM glad that most boards seem to be supporting all the chips anyway)
 

N4CR

Supreme [H]ardness
Joined
Oct 17, 2011
Messages
4,450
Can understand why some boards are VRM limited but many should not be...

If MSI spends minimum time required, releases update as 'at your own risk' and makes this clear I don't see the issue. Just don't be cheap fuckers.
 

Geforcepat

Gawd
Joined
Jun 2, 2012
Messages
925
AMD Better get on the phone here. this is a critical time for them. don't need to lose momentum now.
I see what dan is saying but it doesn't matter much because at the end of the day a gurantee is a gurantee.
 

Chimpee

[H]ard|Gawd
Joined
Jul 6, 2015
Messages
1,480
When you're a board partner, yes, we should hold them accountable. There should be no expectation that every feature works with the new chips, but it should at least run at a basic level of functionality.
You actually think it is reasonable for the board partners to test every single of their 300 series chipset motherboard that they no longer sale to make sure Ryzen 3000 series can run is insane. Not saying MSI couldn't do it, just it doesn't make a lick of sense to do it as there is no guarantee all 300 series motherboard will work with Ryzen 3000, makes much more sense to do it for 400 series chipset motherboard as that is recent and with better vrm.
 

Chimpee

[H]ard|Gawd
Joined
Jul 6, 2015
Messages
1,480
AMD Better get on the phone here. this is a critical time for them. don't need to lose momentum now.
I see what dan is saying but it doesn't matter much because at the end of the day a gurantee is a gurantee.
I can guarantee this will not affect AMD all that much. This is conjecture on my part, but I suspect the number of people who do CPU upgrades are pretty small.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
55,548
I can guarantee this will not affect AMD all that much. This is conjecture on my part, but I suspect the number of people who do CPU upgrades are pretty small.
I would agree with this. I've done more upgrades over the years than I can ever count for various people. I've worked on thousands of systems and I can tell you most of the CPU upgrades I have done were primarily adding CPU's to multi-socket systems. I rarely ever upgraded consumer machines. Memory, storage, video cards and power supplies were all common. CPU upgrades were as rare as hens teeth. In two decades I could probably count all the CPU upgrades I've seen done or have done myself on one hand. People often ask about it but very few actually do it. They typically find that the cost is too high to get a meaningful upgrade.

I think the bulk of CPU upgrades done are by enthusiasts as they'll be the ones more likely to do an incremental upgrade for better overclocking, more cores or whatever. However, that's the crowd that's most likely to replace the motherboard at the same time.
 

Geforcepat

Gawd
Joined
Jun 2, 2012
Messages
925
Dan_D i understand what you're saying but if i remember a few years ago i uprgraded an anthlon x2 to a phenom x3. an then years later upgraded sandybrige setup i5 2400 to i7 2600k about 3 years ago. still running it btw waiting on this newfangled 7nm ryzen 3000.
at the end of the day it's the principle of the thing.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
55,548
Dan_D i understand what you're saying but if i remember a few years ago i uprgraded an anthlon x2 to a phenom x3. an then years later upgraded sandybrige setup i5 2400 to i7 2600k about 3 years ago. still running it btw waiting on this newfangled 7nm ryzen 3000.
at the end of the day it's the principle of the thing.
There are certainly motherboard manufacturers that did update their BIOS ROMs to support the newer CPUs from the beginning of the chain to the end of it. And that's great. However, not every motherboard manufacturer did despite AMD's statements about compatibility. It is up to the motherboard manufacturers to do those updates and AMD has no control over whether or not those companies do so.
 

Chimpee

[H]ard|Gawd
Joined
Jul 6, 2015
Messages
1,480
Dan_D i understand what you're saying but if i remember a few years ago i uprgraded an anthlon x2 to a phenom x3. an then years later upgraded sandybrige setup i5 2400 to i7 2600k about 3 years ago. still running it btw waiting on this newfangled 7nm ryzen 3000.
at the end of the day it's the principle of the thing.
To be honest, I am not sure if AMD will continue support of AM4 socket beyond 2020. 4 years for AM4 is a pretty long for a socket.
 

mvmiller12

Gawd
Joined
Aug 7, 2011
Messages
930
To be honest, I am not sure if AMD will continue support of AM4 socket beyond 2020. 4 years for AM4 is a pretty long for a socket.
It's really not that big a deal when you consider how little the chipset in a Ryzen system actually does. Almost everything of note is built directly into the CPU, and the only real constraint with the socket itself is the number of PCIe lanes that route to it, and the number of power and grounds necessary to make sure the CPU gets good, solid power.

The only real reason to change the socket is to support newer RAM types such as DDR5, and even so, AMD has precedent where there has been an interim series of chips that had 2 different memory controllers and would fit in both the older and newer sockets. I believe there was a 1-pin difference between sockets AM2+ and AM3, with AM2+ supporting DDR2 and AM3 supporting DDR3. The initial run of Phenom II's would work in either socket.
 

BassTek

Supreme [H]ardness
Joined
Jul 13, 2002
Messages
5,449
Looks like people got worked up a little early, glad to see my board will get support. I will probably buy a new board anyway but was planning to use my current one in a secondary system possibly with Zen2.
 

PhaseNoise

2[H]4U
Joined
May 11, 2005
Messages
2,375
Looks like people got worked up a little early, glad to see my board will get support. I will probably buy a new board anyway but was planning to use my current one in a secondary system possibly with Zen2.
Arguably, nobody got worked up early, they got worked up at precisely the right time. Either the theoretical support engineer was just wrong and thus those incorrect statements still needed clarification from the company, or this was actually a test to see if the market would bear such a change.

In both cases, the response given so far from enthusiasts is roughly: "We think that's a poor choice, we expected forwards compatibility. Not providing that could cost you future business". Yes, in some cases it was phrased more ... coarsely.

I think that's a reasonable and fair response from people who felt that they would be purchasing something with forward compatibility. Yes, it was AMD's promise technically, but the end user would view any deviation from motherboard-makers as weasel-words, and someone doing that is likely to lose good faith in the enthusiast community. While not huge in market share, they are (I hate to use this term, damn millenials...) "influencers". Every person here has a sphere of influence far larger than the average person.

As Dan indicates, BIOS updates are a significant investment. A single byte change means re-validation, which is a very time consuming task (top tier vendors have a pretty amazing validation process - I worked with them when I was designing CPUs, and as full-disclosure of opinions - the ASUS folks are the top of the game, from my perspective. <Not speaking to warranty issues, just engineering>). And yes, having 10-20 models exacerbates the problem, but all vendors are quite afraid of not having a model at any given very-granular price point. It's hyper-competitive.

I posted this knowing it was probably either fake or a market probe, but I think this is an interesting discussion to have nonetheless. When do we give vendors leeway to force an upgrade? It's an interesting question. I know people had strong opinions on the Intel side, so wanted to explore it when the shoe could possibly be on the other foot.

Dan points out some very real issues. I suspect some boards from some vendors will actually have insurmountable issues. Hopefully no enthusiast-class boards, but we'll see.

(I love having actual discussions on tech)
 
Top