News AMD Nixes Support for PCIe 4.0 on Older Socket AM4 Motherboards, Here's Why

[rajag89]

I'm not sure why your post is directed at me...

Oops, just noticed I copy pasted your name by mistake. Anyways I will let the thread rest here. So peace

 

[littleleo]

I don't see it being a big deal right now since there isn't a bunch a PCI-E 4.0 VGA cards out there. And I don't see how this has anything to do on CPU backward compatibility,

 

[Redneck5439]

I know some people will be disappointed by this news, but it really isn't a huge deal. I know that AMD said they wouldn't lock the feature out, but looking at the design of the new X570 motherboards it would appear that PCIe 4.0, at least in its current form needs a lot of cooling. Current X470 boards, even high end boards don't have the huge heat sinks and built in fans that the X570s do. More than likely even if the board would be technically capable of running a PCIe 4.0 port it would probably shorten the lifespan of the motherboard, possibly by a lot.

Furthermore I don't really see this as being a huge issue for a lot of people. I know that PCIe 4.0 will give much more bandwith to M.2 SSD devices, however most people were really hoping for a boost in GPU performance. That's not going to happen till at least 2020. The most powerful GPUs on the market don't support PCIe 4.0 because Nvidia chose to give the RTX chips a PCI-Express 3.0 x16 bus interface, even though the PCI-Express gen 4.0 specification was expected. AMD's new GPUs will take advantage of PCIe 4.0 however they are expected to only compete against the RTX 2060, RTX 2070, possibly the RTX 2080. That means that the very best gaming GPUs will still be the RTX 2080Ti and Titan, neither of which have native support for PCIe 4.0. By in large the most popular GPUs in use are the GTX 1000 series and they will of course have no native support either.

By the time that AMD has a truly worthy high end GPU on PCIe 4.0 and Nvidia releases their next gen GPUs with PCIe 4.0 native support we will all be awaiting the release of X670 motherboards and Zen 2+. With regards to the new M.2 SSDs that will support PCIe 4.0- they will be new, they will be expensive, and they will have "teething" issues in their very first generation. I know that I will be content with my current SSD till at least Zen 2+.

PCIe 4.0 in this generation is a "nice to have" feature but far from practical. As long as the 3000 series processors perform just as good on X470 as on X570 I see no reason to upgrade to X570 when the only real difference will be PCIe 4.0 support.

 

[TJ Hooker]

looking at the design of the new X570 motherboards it would appear that PCIe 4.0, at least in its current form needs a lot of cooling. Current X470 boards, even high end boards don't have the huge heat sinks and built in fans that the X570s do. More than likely even if the board would be technically capable of running a PCIe 4.0 port it would probably shorten the lifespan of the motherboard, possibly by a lot.

The built in fans on X570 are for cooling the chipset. Nobody ever though the chipset PCIe lanes were on older boards were going to get upgraded. The PCIe controller for the lanes for the primary slot (and the M.2 slot) is on the CPU itself.

 

[Redneck5439]

Where did you see that there won't be 16 core cpus? Until now, only the 12 core has been announced, but 16 was never denied.
And are you disappointed that old motherboards can't run above what they were designed to do? "Oh no, AMD said their 486 could be compatible with PCI Express 7.0, but now they say it can't, so that's a let down" (exaggerated to make the issue stand out)

AMD definitely has a 16 core Ryzen processor that they are working on, its been spotted several times "in the wild". The 16 core part more than likely won't be launching in July, but who knows what AMD might announce at E3. I think it much more likely that AMD will hold off on releasing a 16 core processor until they see if Intel has any real answer to its Ryzen 3000 line, which right now they simply don't. AMD in fact may hold off on releasing a 16 core till their next generation Zen 2+ which will allow them to work out any latency issues the monster processor may have. The point is Intel doesn't have anything that will rival the 12 core 24 thread processor already announced so rapidly releasing a 16 core 32 thread processor isn't necessary, they can continue to refine and optimize it.

 

[setx]

Looks like motherboard vendors screwed the potential PCIE 4.0 support to the point that AMD decided that it's safer to lock this feature out than face bad publicity from users later due to non-working devices and errors.

What that really shows is: all the motherboards nowadays are designed as cheap as possible regardless of their final price. Otherwise we would see support on top boards that could handle it thanks to better build quality than required by old standard. But no, there is no "better quality" anywhere nowadays.

 

[Redneck5439]

Looks like motherboard vendors screwed the potential PCIE 4.0 support to the point that AMD decided that it's safer to lock this feature out than face bad publicity from users later due to non-working devices and errors.

What that really shows is: all the motherboards nowadays are designed as cheap as possible regardless of their final price. Otherwise we would see support on top boards that could handle it thanks to better build quality than required by old standard. But no, there is no "better quality" anywhere nowadays.

There is truth in what you are saying, however there were a few X470 boards that could have handeled the upgrade, like the Asus Crosshair Hero VII. If you look at the construction of the X370 motherboards, then the X470 motherboards and then the upcoming X570 motherboards you can clearly see the X570 motherboard's build quality is much greater than the previous generations. The motherboard manufactures were treating the Ryzen processors as "second best" and the build quality of a lot of the X470 motherboards shows it. There were many articles that pointed out that a lot of the X470 motherboards were nothing more than B550 boards with X470 nomenclature as their VRMs and heatsinks were subpar at best.

Now that Ryzen is set to topple Intel in the high end CPU marketplace you are seeing much better VRMs, much better heatsinks and motherboard cooling. While I don't believe its really "worth it" to upgrade to X570 if you already have a quality X470 board (and there are only a handful of them) its undeniable that the new boards are made much better than the previous generations. I personally plan on upgrading my motherboard with the next generation (X670), however will be upgrading my processor to either the R7 3800X or possibly the R9 3900X if it doesn't suffer a latency penalty.

https://docs.google.com/spreadsheet...MvYeCLI5ZbIpnq5fyiWD4NCkkU/edit#gid=229691480

- a decent tier list of AM4 motherboards based on VRM. The "top tier" and "high end" would probably have been able to have some PCIe 4.0 support.

 

[Redneck5439]

The built in fans on X570 are for cooling the chipset. Nobody ever though the chipset CPIe lanes were on older boards were going to get upgraded. The PCIe controller for the lanes for the primary slot (and the M.2 slot) is on the CPU itself.

https://www.pcgamesn.com/amd-x570-chipset-fans-user-configurable

From the article "But take one look at the Godlike and it’s evident that MSI is expecting some serious heat from these boards, which, according to AMD, mostly originates from the PCIe 4.0 connectivity for the chipset."

PCIe 4.0 is new technology, I know they added a lot of extra cooling to the new X570 motherboards and until we have some deep reviews of the new boards and technology I'm thinking the extra cooling was necessary.

 

[TJ Hooker]

@Redneck5439 Ok, your quote seems to be agreeing with me. Putting a PCIe 4.0 controller in the chipset (AKA PCH) increased the power consumption of the chipset considerably, to the point where it needed active cooling. But that has nothing to do with PCIe 4.0 support on 300/400 series boards (because they have PCIe 2.0 from the chipset and that can't be changed), and therefore is not relevant to what's being discussed here. The only question was whether the PCIe lanes straight from the CPU (typically the primary x16 slot and the first M.2 slot) might be able to run at 4.0 speeds with a Ryzen 3K CPU on a 300/400 series motherboard, which we now know the answer to (no).

 

[Math Geek]

i do like the idea of a PCIE 3.5 or other such thing which let's it run at a somewhat faster rate to squeeze that last little bit of SSD bandwidth. but it seems even that was too much for the old design to handle. i'd accept that with a smile as a good compromise if it could be done.

i can only assume based on the fact they are not even trying that, how much stress it must be placing on the board. i wonder if some of the over-engineered, top end boards would be able to handle such a thing??

 

[Redneck5439]

@Redneck5439 Ok, your quote seems to be agreeing with me. Putting a PCIe 4.0 controller in the chipset (AKA PCH) increased the power consumption of the chipset considerably, to the point where it needed active cooling. But that has nothing to do with PCIe 4.0 support on 300/400 series boards (because they have PCIe 2.0 from the chipset and that can't be changed), and therefore is not relevant to what's being discussed here. The only question was whether the PCIe lanes straight from the CPU (typically the primary x16 slot and the first M.2 slot) might be able to run at 4.0 speeds with a Ryzen 3K CPU on a 300/400 series motherboard, which we now know the answer to (no).

I got ya. I was thinking that by enabling PCIe 4.0 (or close to it) that extra voltage would also be pushed through the existing 400 series chipset which in return would require additional cooling.

 

[paul horn]

Funny how AMD can make a statement about backward compatible parts and fail to keep their promise and everyone goes oh well. And Intel gets the hot coals. It is nice AMD finally provided some comp.

 

[cryoburner]

...however most people were really hoping for a boost in GPU performance. That's not going to happen till at least 2020.

Or maybe closer to 2030. : D

The reality is that even PCIe 3.0 still offers more than enough bandwidth than today's highest-end graphics cards require for optimum performance, at least as far as gaming is concerned. And only recently have these highest-end graphics cards started to have their performance limited a bit by PCIe 2.0, despite GPUs having supported 3.0 for more than eight years. Upcoming graphics cards will undoubtedly "support" PCIe 4.0, but it isn't likely to improve their gaming performance in any notable way. They'll perform more or less the same in a 3.0 slot, and mid-range cards will even continue to perform similar in a 2.0 slot. The bandwidth of the PCIe interface isn't really limiting performance of these cards. Just making more bandwidth available isn't going to help if they aren't even using what they already have.

 

[Ncogneto]

1.) AMD said it would not lock out the feature.
2.) AMD has locked out the feature.

None of which changes the fact that they are still keeping their promise of backwards compatibility. which makes this comment

"taking a bit of the shine of AMD's vaunted backward compatibility with Ryzen processors on the AM4 socket." from your article, highly suspect and reeking of click bait.
AMD has delivered on their promise. It's not as if AMD users are being asked to disable half the cores of the CPU (cough cough hyperthreading). Owners of older boards will still be able to enjoy a higher IPC, as well as support for faster memory timings. Should AMD also be expected to support PCI 5.0 and USB 16.0 on older boards in the future (and btw, AMD has little to do with how different vendors decide to build their boards for future scalability, they only require them to meet certain standards for current gen support).

Here is a little hint, this has nothing to do with the AM4 socket. Had the board vendors taken the time to make the boards a little more future proof, this would be a non-issue. The shade should be thrown out at the likes of ASUS, Gigabyte, MSI, etc. AMD most likely heard reports from vendors that their older boards would still support PCIe 4.0, and upon further review, AMD found that in fact this what not the case at all.

 

[Ncogneto]

Funny how AMD can make a statement about backward compatible parts and fail to keep their promise and everyone goes oh well. And Intel gets the hot coals. It is nice AMD finally provided some comp.

Except the CPU still is backwards compatible, so your saying what exactly? Should the USB ports also be expected to support future versions of USB, and if not, it is AMD's fault? Such nonsense.

 

[Redneck5439]

Funny how AMD can make a statement about backward compatible parts and fail to keep their promise and everyone goes oh well. And Intel gets the hot coals. It is nice AMD finally provided some comp.

I don't see how AMD failed to keep their promise. They promised to support AM4 through 2020, and true to their word you can buy a Ryzen 3000 series processor and use it in a first gen motherboard and be good to go. I upgraded from Piledriver to the X470 and R7 2700X, I will have no issues with putting a R7 3800X or R9 3900X processor in my existing X470 motherboard and have a worthy upgrade. I have a friend who has a Asus X370 Corsshair Hero VI who is planning to do the same thing. Name me one Intel socket that has had 3 generations of support.

Changing their mind (for the greater good) on PCIe 4.0 support, which can barely be used by anything at this point, is nothing when compared to the bigger promise that was kept of AM4 support though 2020. AMD has already supported the AM4 socket for 3 generations and may have a fourth before switching sockets for 5nm with Zen 3, Intel has never come close to that kind of socket support.

 

[TJ Hooker]

None of which changes the fact that they are still keeping their promise of backwards compatibility. which makes this comment

"taking a bit of the shine of AMD's vaunted backward compatibility with Ryzen processors on the AM4 socket." from your article, highly suspect and reeking of click bait.
AMD has delivered on their promise.

In a perfect world, new CPUs would be backwards compatible with older motherboards without any loss/reduction in the features of the CPU. That wouldn't normally be assumed in a situation like this, but for a time, based on AMD's own comments, it looked like this might actually be the case for Ryzen 3k. But now that we know that's not the case, the backwards compatibility does look a tiny bit less rosy than it did before this news, even though the lack of PCIe 4.0 is understandable and has minimal impact.

 

[King_V]

Funny how AMD can make a statement about backward compatible parts and fail to keep their promise and everyone goes oh well. And Intel gets the hot coals. It is nice AMD finally provided some comp.

What's not backward compatible? Are you saying that PCIe 4.0 parts won't work on PCIe 3.0 slots?

 

[Deleted member 14196]

no, he's moaning about pcie 4.0 not being supported on OLD MOTHERBOARDS that can't handle it and WERE NEVER DESIGNED FOR IT. LOL, it's hilarious!

If anybody thought that pcie 4.0 would work properly with any current motherboards, well, you have a LOT to learn, let's just leave it at that.

 

[jimmysmitty]

I don't see how AMD failed to keep their promise. They promised to support AM4 through 2020, and true to their word you can buy a Ryzen 3000 series processor and use it in a first gen motherboard and be good to go. I upgraded from Piledriver to the X470 and R7 2700X, I will have no issues with putting a R7 3800X or R9 3900X processor in my existing X470 motherboard and have a worthy upgrade. I have a friend who has a Asus X370 Corsshair Hero VI who is planning to do the same thing. Name me one Intel socket that has had 3 generations of support.

Changing their mind (for the greater good) on PCIe 4.0 support, which can barely be used by anything at this point, is nothing when compared to the bigger promise that was kept of AM4 support though 2020. AMD has already supported the AM4 socket for 3 generations and may have a fourth before switching sockets for 5nm with Zen 3, Intel has never come close to that kind of socket support.

He is not wrong though. You, I and everyone here knows if Intel did the same people would rage. I think AMD gets a break being the smaller company.

That said I think its the smart move. They did somewhat shoot themselves in the foot, maybe just a toe, by making the statement that they would not lock it out and leave it to the motherboard vendors to decide but it is a smarter move to disable it overall.

And technically AM4 has supported 2 gens, Zen and Zen 2. I wouldn't consider Zen+ a new gen as it was just a refinement on Zen, much like the old steppings that used to be better power wise (like the Q6600 G0 stepping vs the B3 stepping). And you are correct but also not correct. Intels LGA775 socket supported 4 generations of CPUs. Pentium 4, Pentium D, First generation Core 2 (65nm) and second generation Core 2 (45nm). And yes there were some 900 series chipsets that supported Core 2, it just mattered who the vendor was. As well LGA1151 has supported 4 series of Core CPUs, most of the change was in the chipset and power delivery which does get harder to maintain when you add in cores.

In my opinion its holds the CPU back trying to stay compatible with an older socket and chipset. But thats just my opinion.

 

[King_V]

no, he's moaning about pcie 4.0 not being supported on OLD MOTHERBOARDS that can't handle it and WERE NEVER DESIGNED FOR IT. LOL, it's hilarious!

If anybody thought that pcie 4.0 would work properly with any current motherboards, well, you have a LOT to learn, let's just leave it at that.

Yeah, I probably should've put a /s at the end of my post...

 

[Ncogneto]

In a perfect world, new CPUs would be backwards compatible with older motherboards without any loss/reduction in the features of the CPU. That wouldn't normally be assumed in a situation like this, but for a time, based on AMD's own comments, it looked like this might actually be the case for Ryzen 3k. But now that we know that's not the case, the backwards compatibility does look a tiny bit less rosy than it did before this news, even though the lack of PCIe 4.0 is understandable and has minimal impact.

Let me get out my crayons and I will draw you a picture.

1. It is no the fault of the CPU that the older motherboards don't support PCIe 4.0
2. AMD doesn't make motherboards

It's really not that tough........AMD did not go back on it's promise (unlike Intel, when are those 10 nm CPU's coming?)

Yet the Author, clearly points to AMD as the one in which is losing shine, when in fact it is the likes of MSI, ASUS, Gigabyte, etc that are the actual culprits here. Do you not think they knew full well the specifications for PCIe 4.0 that was on the horizon when they were making their 400 series motherboards? This has nothing to do with the CPU itself, rather the signal integrity of the lanes in the MOTHERBOARD. It would be like the Author placing the blame on Porsche for not being able to drive faster than 70 mph in the US. It's stupid.

AMD's only misstep here is taking the motherboards makers words that some of there older generations of motherboards would indeed support PCIE 4.0, and when AMD did some investigation, they found out they in fact were not up to snuff as the motherboard makes claimed, and as such they disabled support.

 

[TJ Hooker]

Let me get out my crayons and I will draw you a picture.

1. It is no the fault of the CPU that the older motherboards don't support PCIe 4.0
2. AMD doesn't make motherboards

It's really not that tough........AMD did not go back on it's promise (unlike Intel, when are those 10 nm CPU's coming?)

Yet the Author, clearly points to AMD as the one in which is losing shine, when in fact it is the likes of MSI, ASUS, Gigabyte, etc that are the actual culprits here. They knew full well what they had to do to make there older motherboards forward compatible with PCIe 4.0 and choose not too. This has nothing to do with the CPU itself, rather the signal integrity of the lanes in the MOTHERBOARD. It would be like the Author placing the blame on Porsche for not being able to drive faster than 70 mph in the US. It's stupid.

I never said anything about whose fault it was that older mobos don't support PCIe 4.0, nor did the author. It doesn't really matter at the end of the day, as AM4 is AMD's platform so it behooves them to stay on top of what their motherboard partners are doing and work with them, because those motherboards will ultimately reflect on AMD's product because they're part of AMD's ecosystem.

AMD's only misstep here is taking the motherboards makers words that some of there older generations of motherboards would indeed support PCIE 4.0, and when AMD did some investigation, they found out they in fact were not up to snuff as the motherboard makes claimed, and as such they disabled support.

You have any evidence that's how it played out? I.e. the mobo manufacturers lied to and misled AMD into making that initial statement, which AMD now has had to go back on.

 

[jimmysmitty]

Let me get out my crayons and I will draw you a picture.

1. It is no the fault of the CPU that the older motherboards don't support PCIe 4.0
2. AMD doesn't make motherboards

It's really not that tough........AMD did not go back on it's promise (unlike Intel, when are those 10 nm CPU's coming?)

Yet the Author, clearly points to AMD as the one in which is losing shine, when in fact it is the likes of MSI, ASUS, Gigabyte, etc that are the actual culprits here. Do you not think they knew full well the specifications for PCIe 4.0 that was on the horizon when they were making their 400 series motherboards? This has nothing to do with the CPU itself, rather the signal integrity of the lanes in the MOTHERBOARD. It would be like the Author placing the blame on Porsche for not being able to drive faster than 70 mph in the US. It's stupid.

AMD's only misstep here is taking the motherboards makers words that some of there older generations of motherboards would indeed support PCIE 4.0, and when AMD did some investigation, they found out they in fact were not up to snuff as the motherboard makes claimed, and as such they disabled support.

The issue is not that isnt does or does not support it. The issue is that AMD stated they would not lock the feature out and leave it up to motherboard manufactures, which I guarantee some of the better X470s at least could support it, yet now they are locking it out completely. Thats the issue.

I agree with their decision but AMD has some culpability to the claims. It is not just the manufactures who made the claims, TH even got a statement from AMD on this 5 months ago.

 

[Ncogneto]

I never said anything about whose fault it was that older mobos don't support PCIe 4.0, nor did the author.

BS.

The Author's opening statement

" taking a bit of the shine of AMD's vaunted backward compatibility with Ryzen processors on the AM4 socket. "

I laid out a perfectly sound argument why this doesn't take any shine of AMD's backwards compatibility. It's not my fault you can't follow simple logic. My main issue was with the wording in which the Author used. And while you can claim he never said it was AMD's fault, it most certainly is either implied, or just sloppy writing.

It doesn't really matter at the end of the day, as AM4 is AMD's platform so it behooves them to stay on top of what their motherboard partners are doing and work with them, because those motherboards will ultimately reflect on AMD's product because they're part of AMD's ecosystem.

Isn't that what they are doing, not allowing motherboard vendors to utilize PCIe 4.0 when their testing of these third party boards shows they cannot do and comply with the PCIe 4.0 spec? They are still backwards compatible btw.

You have any evidence that's how it played out? I.e. the mobo manufacturers lied to and misled AMD into making that initial statement, which AMD now has had to go back on.

Do you have any evidence of the contrary? Was it AMD's job to check backwards compatibility of all the various motherboards out there to see if they could conform to the PCIe 4.0 spec? Before they had release ready silicon?

This is what AMD Said specifically


"Users today may find a PCIe 4.0 option available in their pre-X570 motherboards. However, users should expect this option to be disabled when final retail BIOSes are released to implement full performance and stability for new 3rd Gen Ryzen processors.

As pre-X570 motherboards were not designed with PCIe 4.0 in mind, their designs may be incapable of running PCIe 4.0 signaling with the requisite stability and performance. To ensure a reliable and consistent experience in the field, PCIe 4.0 will not be an option ultimately available to pre-X570 motherboards. Users may continue with a beta BIOS if they desire, but performance and stability cannot be guaranteed."

Reading comprehension, it's your friend. AMD is not locking the CPU's from any older motherboards, if this were the case, a beta bios would not work either.