ASUS Begins Enabling Limited PCIe Gen 4.0 on AMD 400-series Chipset Motherboards

sknight

Limp Gawd
Joined
Oct 7, 2009
Messages
131
https://www.techpowerup.com/257380/...en-4-0-on-amd-400-series-chipset-motherboards

4oMEGGd0uTm1vkxB.jpg
 
SOmeone should tell AMD this is a time for them to contrast versus Intel and gain PR points by leaving well enough alone. Trying to block this functionality, even when it works, will only be seen as attempting to push people onto 500 series chipsets which are more costly, "hiding" PCI-E 4.0 features capable on 400 series even unofficially If they leave this alone it will affect relatively few people and even fewer negatively. If mobo manufacturers enable this via BIOS as sort of an unsupported update, then if it doesn't work for whatever reason people will go back down to to 3.0 and be none the worse for wear.

The last period of AMD serious competition before Ryzen, during the Phenom 1 and 2 era, AMD won a lot of converts from allowing motherboards to try to "unlock" disabled cores on their dual or tri-core parts, rather than physically burning them out. Sometimes the unlocked core performed around or at as well as a higher-tier model, sometimes it didnt, but AMD was appreciated for giving people the chance and motherboard manufacturers made it a feature, especially on higher end boards, to support attempting to unlock and having good VRMs etc.

AMD would be making a big mistake to block this, in the wake of all the good PR they've gotten for Ryzen and its chipsets.

Note: Nice to see Asus doing a good job with support and these features. I wonder how the quality would stack up instead of using mid-tier B450 TUF boards versus something like X470 ROG instead, which have better VRMs and more layers to the PCB etc.
 
Ugh the back and forth will or wont it work and will or wont amd allow it is confusing as hell and only going to huet AMD.. Just say its not officially supported and move on AMD-- let the mobo manufacturers support it if they want and get some goodwill from the enthusiasts that have and keep you going.

Leave it up to amd marketing to burn through the ryzen and navi goodwill theyve earned by pulling this crap. Typical amd pr -.-
 
Ugh the back and forth will or wont it work and will or wont amd allow it is confusing as hell and only going to huet AMD.. Just say its not officially supported and move on AMD-- let the mobo manufacturers support it if they want and get some goodwill from the enthusiasts that have and keep you going.

Leave it up to amd marketing to burn through the ryzen and navi goodwill theyve earned by pulling this crap. Typical amd pr -.-


I think the folks that even have an inkling as to what PCE-E is, will be able to do the necessary research.
 
All fun and games until those 4 layer PCB versions start catching fire or become unstable. AMD suggested to not enable PCI-E 4.0, not force it. Like what was said before. Its unofficial support, just like windows 7.

If AMD really wanted to lock out support, they would disable CPU support for anything above 8 cores.
 
Either the motherboards are capable of running PCI 4.0 or they are not. There is no middle "almost pregnant" here. The signal integrity is there or it is not.
The MB manufacturers need to show AMD that, through thorough testing and validation, PCI 4.0 is viable on (some of) their motherboards. Not all of the boards will pass validation and not all of the slots will pass validation, but some will. And some is better than none, which seems to be what AMD wants.

Yes, some users will be disappointed that their chosen MB model can't do PCIe 4.0. And other users will be mad about it. So what. It was never a promised feature for older boards.

Let the engineers and engineering tests decide which boards can handle PCIe 4.0, and which slots (GPU and/or M.2) work reliably for each one.
I think AMD should allow PCI 4.0 on older boards that can demonstrate reliable performance via stable signal integrity and quality PCB design methodologies and manufacturing. As stated above, AMD can say PCIe 4.0 is not officially supported on pre-X570 boards. I'm sure the lawyers are involved in these decisions as well. But there is always the fallback to PCI 3.0 just a BIOS setting away if a user runs into trouble with PCI 4.0 enabled. Spending the extra bucks on a PCI 4.0 SSD would suck though if you later found out you couldn't run it wide open. Caveat emptor.
 
Either the motherboards are capable of running PCI 4.0 or they are not. There is no middle "almost pregnant" here. The signal integrity is there or it is not.
The MB manufacturers need to show AMD that, through thorough testing and validation, PCI 4.0 is viable on (some of) their motherboards. Not all of the boards will pass validation and not all of the slots will pass validation, but some will. And some is better than none, which seems to be what AMD wants.

Yes, some users will be disappointed that their chosen MB model can't do PCIe 4.0. And other users will be mad about it. So what. It was never a promised feature for older boards.

Let the engineers and engineering tests decide which boards can handle PCIe 4.0, and which slots (GPU and/or M.2) work reliably for each one.
I think AMD should allow PCI 4.0 on older boards that can demonstrate reliable performance via stable signal integrity and quality PCB design methodologies and manufacturing. As stated above, AMD can say PCIe 4.0 is not officially supported on pre-X570 boards. I'm sure the lawyers are involved in these decisions as well. But there is always the fallback to PCI 3.0 just a BIOS setting away if a user runs into trouble with PCI 4.0 enabled. Spending the extra bucks on a PCI 4.0 SSD would suck though if you later found out you couldn't run it wide open. Caveat emptor.

You missed the part where AMD has, so far, claimed that the next big AEGSA update that they rollout to the board makers will outright remove this functionality, board quality or not.
 
You missed the part where AMD has, so far, claimed that the next big AEGSA update that they rollout to the board makers will outright remove this functionality, board quality or not.
I did not miss that at all. I am giving my opinion on how AMD could handle this, rather than by an iron fist (locked AGESA). I do wonder why the MB manufacturers are even bothering with the work they have done so far to enable PCIe 4.0 if AGESA will be locking it out anyway. Nobody would want to use a beta BIOS forever just to gain PCIe 4.0 support.
 
All fun and games until those 4 layer PCB versions start catching fire or become unstable. AMD suggested to not enable PCI-E 4.0, not force it. Like what was said before. Its unofficial support, just like windows 7.

If AMD really wanted to lock out support, they would disable CPU support for anything above 8 cores.

Something tells me if AMD was able to pull more power through a PCIE socket than it was spec'd for, then this will also be a non-issue. It just sounds more like AMD wanting to force the use of X570 more than anything.
 
Something tells me if AMD was able to pull more power through a PCIE socket than it was spec'd for, then this will also be a non-issue. It just sounds more like AMD wanting to force the use of X570 more than anything.


I would too because there are too many B350/B450/X370/X470 boards with questionable VRMs especially for Ryzen 9
 
If they didn't do the work and can't guarantee it, then they should remove it. Its not like x570 mobos don't come in all kinds of price ranges.
Yeah let's castigate AMD for ' removing' never offered options.. yeah that's the company to castigate... /S ... Not four core no ht , need new mobo for a sneeze Intel.
 
If they didn't do the work and can't guarantee it, then they should remove it. Its not like x570 mobos don't come in all kinds of price ranges.
Yeah let's castigate AMD for ' removing' never offered options.. yeah that's the company to castigate... /S ... Not four core no ht , need new mobo for a sneeze Intel.
No no no, because bad bad Intel do wrong, AMD get free pass. "/s"

Long term result: yet another company turns heel, all because consumers made poor excuses and never pushed back.

EDIT: "/s," of course.
 
I would too because there are too many B350/B450/X370/X470 boards with questionable VRMs especially for Ryzen 9

It's not like Ryzen 2 was something AMD had no idea would be like. They knew full well where their roadmap was going and what the differences would be like. AMD should have worked with board partners to make sure that their own promise of full comparability was fulfilled.

But they didn't, it really is that simple.

If any other company would have done this they would have been lambasted 7 ways from Sunday, and not a single person can deny that.

And let's not even get into the fact that tons of their CPUs aren't even hitting their advertised boost clocks and there hasn't even been any criticism of any notable worth on this regard. This is yet another aspect that other companies would have been burned at the stake for.
 
I don't even see this as an issue. If AMD blocks the use of PCI 4 on older 470 and 450s so what. I agree with them that it's probably not a good idea to run powered things far above spec... and the spec on the 400x stuff is pci 3, and 4 is known to draw much more power and create a lot more heat. I say its a non issue as right now anyone willing to drop the extra $ for a PCI 4 device isn't upset about going 570 and probably already have. Anyone that has a high end 470 has likely already populated at least on of their M.2 slots with a high end PCIE nvme... is anyone really considering pulling their Samsung 970s or WD blacks out. I know down the line more stuff will use 4 and be capable of exploiting it. Still I don't see how AMD choosing to lock a feature out that really does have the potential to melt less then boards is a bad thing. That is all AMD needs at this point is stories in mainstream publications about users upgrading to AMDs 12 core chips and setting their houses on fire. (no one is going to blame the nvme running at 120 degrees)

I mean we all know the basics of PCI 4... PCI 3 was an efficiency bump. PCI 4 gets its performance increase through raw increased clockspeed. It draws somewhere around 30% more power. Now sure it's possible some over engineered x470 boards could handle that... but I'm not so sure the 470 chipset is up to snuff. AMD claims its due to stability issues they are pulling unofficial support and I think we should likely believe them. AMD has been pretty generous in the past and present allowing core unlocking on old bulldozers and not locking multipliers ect. I tend to believe what they are saying. There is a very good reason that the 570 chip has active cooling. I don't think I would trust a 470 running pci4 on the nvme and / or GPU... sounds to me like it will be nothing but unstable with a high potential of melting my components.
 
There is a very good reason that the 570 chip has active cooling.



I agree wholeheartedly. Active cooling on chipsets is a real PITA, and was abandoned long ago (until just recently) for very good reasons.
 
I don't even see this as an issue. If AMD blocks the use of PCI 4 on older 470 and 450s so what. I agree with them that it's probably not a good idea to run powered things far above spec... and the spec on the 400x stuff is pci 3, and 4 is known to draw much more power and create a lot more heat. I say its a non issue as right now anyone willing to drop the extra $ for a PCI 4 device isn't upset about going 570 and probably already have. Anyone that has a high end 470 has likely already populated at least on of their M.2 slots with a high end PCIE nvme... is anyone really considering pulling their Samsung 970s or WD blacks out. I know down the line more stuff will use 4 and be capable of exploiting it. Still I don't see how AMD choosing to lock a feature out that really does have the potential to melt less then boards is a bad thing. That is all AMD needs at this point is stories in mainstream publications about users upgrading to AMDs 12 core chips and setting their houses on fire. (no one is going to blame the nvme running at 120 degrees)

I mean we all know the basics of PCI 4... PCI 3 was an efficiency bump. PCI 4 gets its performance increase through raw increased clockspeed. It draws somewhere around 30% more power. Now sure it's possible some over engineered x470 boards could handle that... but I'm not so sure the 470 chipset is up to snuff. AMD claims its due to stability issues they are pulling unofficial support and I think we should likely believe them. AMD has been pretty generous in the past and present allowing core unlocking on old bulldozers and not locking multipliers ect. I tend to believe what they are saying. There is a very good reason that the 570 chip has active cooling. I don't think I would trust a 470 running pci4 on the nvme and / or GPU... sounds to me like it will be nothing but unstable with a high potential of melting my components.

the chipset handling it doesn't matter, the only thing the aib's are supporting are the pcie 4.0 lanes coming off the CPU e.g. first x16 slot and/or first m.2 slot if the board supports it.. the chipset is completely unaffected since the interconnect is pcie 3.0. the limiting factor is signal integrity which i doubt there are to many x470/b450 boards with the pcb and trace quality needed to support pcie 4.0.
 
the chipset handling it doesn't matter, the only thing the aib's are supporting are the pcie 4.0 lanes coming off the CPU e.g. first x16 slot and/or first m.2 slot if the board supports it.. the chipset is completely unaffected since the interconnect is pcie 3.0. the limiting factor is signal integrity which i doubt there are to many x470/b450 boards with the pcb and trace quality needed to support pcie 4.0.

Fair enough I stand corrected there. I believe your correct. Bottom line is as you say those traces are going to be asked to handle 30% higher power usage. So I agree for this feature to be stable on 470 boards they would have had to be over engineered to a pretty high degree. It sucks that 570 boards are in general a lot more expensive... still it seems likely there is a good reason beyond OEMS greed. :)

I think we all may have to get used to MOBOS costing a bit more in the future. Once Intel ups their game and moves to PCI 4 or 5 down the road... it looks like boards are going up. At least AMD is giving users the choice of using an older PCI 3 loving board with their shiney new CPUs. I expect Intels new parts with PCI 4 or 5 are going to be on a new socket. lol
 
No no no, because bad bad Intel do wrong, AMD get free pass. "/s"

Long term result: yet another company turns heel, all because consumers made poor excuses and never pushed back.

EDIT: "/s," of course.
Thats the thing, its not a free pass. We don't know what's involved in validating this shit.
Intel just cuts shit, OLD shit out of making different levels of products, and they piss all over backwards compatible as very clear business move born out of just being able to do it bacause of their dominant position.
We don't know the level of 'anger' AMD has towards motherboards manufacturers doing this that or the other... But we can guarantee who will look like SHIT to everyone when shit doesn't work as advertised on old mobos.. And that is going to be AMD, no Asus, not anybody else.
 
As far as I know AMD has never said that 370 or 470 boards were supposed to have PCI-e 4 support. I'm pretty damn sure it was never announced except for 570 boards. I see no problem with AMD flat out stating no PCI-e 4 on anything but what they officially support and I see no problem with locking out that ability. The fact of the matter is that on the multitude of boards which can't handle it and people try to make it work, AMD is going to catch the flak when their boards die and take who know what other components die with the board.

Allowing motherboard makers to put in the option to run PCI-e 4 on old boards not meant to run it sets AMD up for massive public backlash when many of those boards die. AMD disabling the option from ever seeing the light of day saves them so much trouble and bad press in the long run. I guarantee you the number of people complaining about AMD stopping this is going to be a tiny fraction of those complaining about dead hardware trying to run a specification the hardware was never designed for. This is not difficult to understand.
 
Here's my expectation from my b450, PCI-e 3 works.... If I score an advantage, cool, if not then I know where to buy new mobos
 
it may seem like forcing x570, But the stink people would raise if a board manufacturer released it on a board that couldn't handle it and corrupted data would be a lot bigger issue
 
It's not like Ryzen 2 was something AMD had no idea would be like. They knew full well where their roadmap was going and what the differences would be like. AMD should have worked with board partners to make sure that their own promise of full comparability was fulfilled.

But they didn't, it really is that simple.

If any other company would have done this they would have been lambasted 7 ways from Sunday, and not a single person can deny that.

And let's not even get into the fact that tons of their CPUs aren't even hitting their advertised boost clocks and there hasn't even been any criticism of any notable worth on this regard. This is yet another aspect that other companies would have been burned at the stake for.

Look at the article date - https://www.anandtech.com/show/11967/pcisig-finalizes-and-releasees-pcie-40-spec

OK, the X470/B450 boards were in manufacturing at his time. I am pretty sure that the boards need some kind of certification to meet this spec officially. Hence unofficial support from AMD... you know to avoid those class action suits. A roadmap is a roadmap, that's why there is always a little asterisk at the bottom saying it is "Subject to change". Not defending AMD in any ways as this is the same as Intel on many fronts. Intel took advantage of it with Z170 and caught red handed.
 
What really surprises me about this.....

Is that current SSD’s can benchmark 4K random at faster than SATA3 speeds. That is truly amazing.
 
I can confirm its working fine on my 3700x/x470 Prime Pro/5700XT combo. This is on the latest bios (but was released on June 16th).
 
Back
Top