Asus publishes X470 and B450 PCIe Gen 4 compatibility chart

[HEXUS]

At Computex AMD denied that pre-X570 motherboards would support PCIe Gen 4.

Read more.

[azrael-]

I love the Playstation terminology.

Now, if other manufacturers would do the same... ASRock, I'm looking at you, and the X470 Tai Chi motherboard in particular.

[DanceswithUnix]

So I can use PCIe4 NVMe on my old board, that's really handy. I'm not sure of the usefulness of x8 PCIe4 vs x16 PCIe3 unless the other lines can be routed elsewhere. I only have a GPU and USB3 card plugged in so I haven't paid attention to lane usage details

[Corky34]

Have i read that correctly? It seems the cheaper boards are more likely to support 4.0.

[meuvoy]

Have i read that correctly? It seems the cheaper boards are more likely to support 4.0.

Yes, that's right!

I can't say for sure, but I'd Guess that's probably because of size. Like, x470s will mostly be ATX or EATX while there's a lot of mATX b450s. Basically smaller board equals smaller signal channel which, in turn, equals to less signal degradation and less interference.

Newer x570 boards will use higher quality materials to make up for the distance traveled by the signal. But older boards didn't need that and most of them were built using lesser quality materials as it was cheaper and good enough.

[Corky34]

The distance from the CPU socket to the PCIe slots are a constant no matter what size the board is, i think the location of NVMe slots in relation to the first PCIe slot (above or bellow) may vary, maybe it's down to all the extra stuff on higher priced boards causing interference, IDK.

[jimbouk]

Hmmm, makes an upgrade a while lot more tempting with cheaper boards supporting the latest tech!

[DanceswithUnix]

Have i read that correctly? It seems the cheaper boards are more likely to support 4.0.

Higher end boards have SLI. That will be why my X470 Pro board which ISTR can run SLI as a pair of x8 cards can only do PCIe4 x8, I guess the other 8 traces run also to the second x16 slot messing up signal integrity to the first.
Maybe the high end boards have a bridge between the CPU and GPU sockets to try and run them at x16, so a lack of direct connection would stuff PCIe4 compatibility completely.

[azrael-]

Higher end boards have SLI. That will be why my X470 Pro board which ISTR can run SLI as a pair of x8 cards can only do PCIe4 x8, I guess the other 8 traces run also to the second x16 slot messing up signal integrity to the first.
Maybe the high end boards have a bridge between the CPU and GPU sockets to try and run them at x16, so a lack of direct connection would stuff PCIe4 compatibility completely.

Damn! Why did you have to come up with such plausible reasoning. That completely stuffs up my hope for the X470 Taichi.

Although tbh, PCIe4 for graphics isn't *that* high up on my list of "must-haves".

[DanceswithUnix]

Damn! Why did you have to come up with such plausible reasoning. That completely stuffs up my hope for the X470 Taichi.

Although tbh, PCIe4 for graphics isn't *that* high up on my list of "must-haves".

Very true, tbh I expect by the time PCIe4 matters on graphics we will be on DDR5 ram and hence an AM5 motherboard and this won't matter.

The NVMe bit though, that's a damned nice bonus that I might actually use.

[Xlucine]

I'd be annoyed if I had got a ROG board, but as I'm running a TUF board which is listed as fully compatible here I'm happy to be proven wrong (compared to what I thought we'd get based on the last bit of news about this).

Higher end boards have SLI. That will be why my X470 Pro board which ISTR can run SLI as a pair of x8 cards can only do PCIe4 x8, I guess the other 8 traces run also to the second x16 slot messing up signal integrity to the first.
Maybe the high end boards have a bridge between the CPU and GPU sockets to try and run them at x16, so a lack of direct connection would stuff PCIe4 compatibility completely.

Isn't that only on X470, so wouldn't explain the B450 ROG boards not doing it? One of those is ITX as well, so path length can't be an issue

[MRFS]

I suspect it's time to start considering much more robust passive cooling for the undersides of PCIe 4.0 motherboards. Who will be the first to add an active water cooling loop to those undersides? I'm not an EE, but I do remember learning, years ago, that circuit impedance increases with temperature. All we need is a reputable manufacturer of tower and mid-tower cases to lower the motherboard try and make up the difference with longer motherboard stand-offs. VOILA! And, the result is much more room for the same kind of cooling that is proliferating on the upper sides of late-model motherboards. At least, the idea deserves some empirical testing.

[adidan]

Sounds like AMD are still planning to disable this on previous gen boards and are going to release an AGESA to manufacturers to put an end to it.

AMD gets all mean

[kompukare]

AMD gets all mean

Since almost nobody actually likes x570, this might be one way to shift new motherboards.

Wonder if if B550 will be less power hungry. Or for that mattter, if the x570 chipset = the Ryzen 3000's IO chip, what does that say about the split of power consumption between the 7nm chiplets and the IO chip?

Also, wasn't the rumour that B550 etc. were going to be ASMedia? Wonder that means in terms of power consumption and features.

Realistically, the only thing which might actually utilize PCIe 4.0 in the near term are new NVMe drives.

[Tabbykatze]

Sounds like AMD are still planning to disable this on previous gen boards and are going to release an AGESA to manufacturers to put an end to it.

AMD gets all mean

Why would they? I have to make the joke: "But they're not Intel". (Well, I hope they won't but anyway)

AMD locking out 4.0 on older motherboards is a reasonable thing to do because if mobo vendors start allowing it and there are instabilities/flake conditions meaning that users will suffer an impacted experiences with Ryzen 3k. Then they complain and moan and it sours the whole thing and then AMD gets flack when they say "well, yeah, it's not been validated for 4.0, sure it is compatible but it's not certified".

I don't want them to lock out 4.0 on older boards because it gives people an option to see it and look to upgrade (as well as the other edge of the blade being they may stay but either way AMD wins). But I can imagine AMD doesn't want to deal with the fallout of whiners, keyboard warriors and anti-AMD Intel Fascists when 4.0 doesn't work properly on older boards.

If they do allow it, I want them to stipulate that mobo vendors put a scrolling font 72 unskippable warning that the board is not validated/certified and you use at your own risk and if it causes you issues suck it up snowflake.

[adidan]

Why would they? I have to make the joke: "But they're not Intel". (Well, I hope they won't but anyway)

AMD locking out 4.0 on older motherboards is a reasonable thing to do because if mobo vendors start allowing it and there are instabilities/flake conditions meaning that users will suffer an impacted experiences with Ryzen 3k. Then they complain and moan and it sours the whole thing and then AMD gets flack when they say "well, yeah, it's not been validated for 4.0, sure it is compatible but it's not certified".

I don't want them to lock out 4.0 on older boards because it gives people an option to see it and look to upgrade (as well as the other edge of the blade being they may stay but either way AMD wins). But I can imagine AMD doesn't want to deal with the fallout of whiners, keyboard warriors and anti-AMD Intel Fascists when 4.0 doesn't work properly on older boards.

If they do allow it, I want them to stipulate that mobo vendors put a scrolling font 72 unskippable warning that the board is not validated/certified and you use at your own risk and if it causes you issues suck it up snowflake.

I don't disagree at all.