PCIe 4.0 on ASUS X470 & B450

[dgsddfgdfhgs]

whats the difference in design,  B450 vs B350 in pcie lane...?

asking from a very disappointed B350F user lol

[williamcll]

On 7/13/2019 at 9:09 PM, Dabombinable said:

I'd say no as well since the first m.2 runs off the CPU and there is no SLI. As stated in my edit, the board looks terrible compared (to clarify, spec wise) to previous ones from the series.

I wonder would there be a PCIE3x16 to dual PCIE4x4 adapator.

[NumLock21]

Did Asus purposely not support PCIe 4.0 on their higher end ROG boards?! Way to screw on your customers!

[WereCat]

8 minutes ago, NumLock21 said:

Did Asus purposely not support PCIe 4.0 on their higher end ROG boards?! Way to screw on our customers!

TBH, not sure if that's true or not but I've read somewhere that the reason is the added heat from RGB LED that restricts them from doing this. 

Now, I would call BS on that but then I remembered the Jay2Cents video where he proved that the RGB on his SSD makes it throttle... a lot. 

[Odemia]

24 minutes ago, NumLock21 said:

Did Asus purposely not support PCIe 4.0 on their higher end ROG boards?! Way to screw on our customers!

A lot of the higher end boards do single x16 or dual x8/x8. That would typically be done with a PCIe switch. Makes complete sense that the silicone in the PCIe switch can't do 4.0. But the simpler boards without the PCIe switches seem to be ok.
 

Note that they are showing almost universal support for x4 lanes to the M.2. That would make sense as there are very few boards with any options for these lanes. Which means no switch or extra traces adding capacitance to the lines.

 

The one stand out is the TUF X470, curious why that one can't support 4.0x4 off the CPU? They must not have shielded those lanes well enough, had to few layers to route them properly or something.

[LAwLz]

Not sure if someone has posted this already, but here is the statement AMD gave to Anandtech:

Quote

Our plan is unchanged. For the reliability and consistency reasons cited at Computex, we still intend to disable PCIe Gen 4 for pre-X570 motherboards. That AGESA is being released to motherboard manufacturers soon.

 

 

On AMD, motherboard manufacturers has very little control over the BIOS and chipset parts, so there isn't really much they can do if AMD intend to block it. Companies like Asus can basically choose to do one of two things.

1) Ship motherboards running a beta BIOS, potentially pissing off AMD for going against their order, and potentially causing a bunch of issues for their customers (might cause issues with things like the CPPC2 interface, which is what happened with some reviews).

2) Comply with AMD and disable PCIe 4.0 on the older boards.

 

 

 

On 7/13/2019 at 3:09 PM, JoostinOnline said:

Except AMD was honest about it. A lot of where the controversy came from was Intel saying that it was impossible, when it turned out that it was just more work. Even though it sucks that AMD is pushing against any older boards getting PCIe 4.0 support, they were honest about their reasons. It also helps that, true or not, they're generally considered more "consumer friendly" compared to Intel. 

Pretty sure Intel never said "that it was impossible". 

In fact, I have been googling around a while now and I can't find any official Intel statement on the issue at all, other than their compatibility chart simply stating that Coffee Lake isn't supported on Z270.

So I think this situation is very comparable.

[mr moose]

1 hour ago, LAwLz said:

Pretty sure Intel never said "that it was impossible". 

In fact, I have been googling around a while now and I can't find any official Intel statement on the issue at all, other than their compatibility chart simply stating that Coffee Lake isn't supported on Z270.

So I think this situation is very comparable.

 

Could you imagine the forum comments and snide remarks had Intel allowed their CPU's to run on older boards causing power issues and blown parts?

 

 

 

 

[Anomnomnomaly]

I'm very interested in this news as I will be swapping out my current Gigabyte B450 & 2600X into my mediaserver and building a Ryzen 3xxx based gaming rig. I was budgeting for an X570 but now I'm considering a high end ASUS B450 again. The Tuf Pro Gaming will cover both and future proof me for a couple more years should I upgrade to PCIE4 storage... and as I'm most likely getting the 5700XT when partner boards are released.

 

 

[WereCatf]

Just now, Anomnomnomaly said:

I'm very interested in this news as I will be swapping out my current Gigabyte B450 & 2600X into my mediaserver and building a Ryzen 3xxx based gaming rig. I was budgeting for an X570 but now I'm considering a high end ASUS B450 again. The Tuf Pro Gaming will cover both and future proof me for a couple more years should I upgrade to PCIE4 storage... and as I'm most likely getting the 5700XT when partner boards are released.

No. AMD is not allowing PCI-E 4.0 on old boards, they have disabled it in AGESA now.

[Mira Yurizaki]

On 7/12/2019 at 6:48 AM, LukeSavenije said:

but how? the tracing for 4.0 is wildly different

If they designed the traces for PCIe signaling to the point of well, being overengineered for the application, then it can carry a faster speed.

 

Further reading regarding PCB design for high-speed signaling:

https://www.tempoautomation.com/blog/special-considerations-for-high-speed-board-design/

https://www.4pcb.com/blog/high-speed-pcb-design-considerations/

https://www.epectec.com/pcb/signal-integrity.html

 

[spartaman64]

i dont really see the importance in this. pcie 4.0 shouldnt be a selling point to like 99.9999999% of people in fact pcie 2.0 is more than enough for like 90% of people probably

[ARikozuM]

2 minutes ago, spartaman64 said:

i dont really see the importance in this pcie 4.0 shouldnt be a selling point to like 99.9999999% of people in fact pcie 2.0 is more than enough for like 90% of people probably

The thing is that you can have a single lane of PCIe 4.0 and still have the same bandwidth as PCIe 2.0 x4. This means more PCIe devices can fit on a board that take advantage of the CPU rather than DMI. 

[spartaman64]

1 minute ago, ARikozuM said:

The thing is that you can have a single lane of PCIe 4.0 and still have the same bandwidth as PCIe 2.0 x4. This means more PCIe devices can fit on a board that take advantage of the CPU rather than DMI. 

when are you getting a 16GB/s nvme ssd or when are you going to need 16 samsung 970 nvme ssds

[ARikozuM]

8 minutes ago, spartaman64 said:

when are you getting a 16GB/s nvme ssd or when are you going to need 16 samsung 970 nvme ssds

Considering that we have motherboards today that run only one or two slots from the CPU at the full bandwidth whereas the rest have to split from the much slower DMI 3.0. It isn't so much "what would you gain" so much as it is how much will run natively at full speed from the CPU rather than the chipset or DMI. 

[mr moose]

1 hour ago, ARikozuM said:

Considering that we have motherboards today that run only one or two slots from the CPU at the full bandwidth whereas the rest have to split from the much slower DMI 3.0. It isn't so much "what would you gain" so much as it is how much will run natively at full speed from the CPU rather than the chipset or DMI. 

I'm having trouble thinking of anything at a consumer level that would max out the current bandwidth even on limited motherboards. 

 

 

[ARikozuM]

51 minutes ago, mr moose said:

I'm having trouble thinking of anything at a consumer level that would max out the current bandwidth even on limited motherboards. 

 

 

This could fix issues with SATA and M.2 disabling each other or PCIe x16 or x8 disabling x8 and x4 slots. 

 

I'm just stating examples where this could be used in a split fashion with PCIe 3 devices. I don't know if it can, but it'd be nice for those of us with plenty of HBA and other cards.