Ryzen 3000: Asus opens up PCIe 4.0 support for selected X470 and B450 boards

asturur

2019-07-12 21:45

Clouseau:

The x470 Prime-Pro is nothing special. reason being that PCIe 4.0 x8 is the same as 3.0x16. Other than that is the same as the other boards that only support the 4.0 connection to one M2 drive. The boards to have would be the B450 boards.

I think that being not special, is probably simple, and so it works. I wonder if the special pci express managing that the crosshair 7 does ( x16, x8/x8 or x8/x4/x4) is the indeed the extra layer on that bus that make the compatibility with v4 not possible.

#5690719

Athlonite

2019-07-12 22:06

Beats me why they even bothered with the TUF X470 Plus Gaming the boards a piece of shite and tem minutes into gaming with an 3900x installed would probably cook the tiny little VRM on it

#5690722

Fox2232

2019-07-12 22:12

Athlonite:

Beats me why they even bothered with the TUF X470 Plus Gaming the boards a piece of shite and tem minutes into gaming with an 3900x installed would probably cook the tiny little VRM on it

CPU will not exceed set up power limits as long as it is physically possible. And that minimal power is like 20~25W for 3900X. For 2700X it is 16~17W.

#5690726

Alessio1989

2019-07-12 22:24

Outside BIOS settings manipulations, the only issue with VRM can be heat. Some MB have bad or not VRM dissipation at all. Although the BIOS should assure the system never reach the maximum VRM temperature allowed, having always VRM over 100-105° is not good.

#5690744

Falck

2019-07-12 23:38

Stoked about X470/B450 gen 4 support! But for me, that has C6HWIFI (x370) I couldn't help notice that there's a picture also in the article about BIOS versions. But as I can't read anything other than latin letters, what is the title of that image? Since 7106 doesn't wan't to boot my 3900x I'm curius why Asus as shared that image over BIOS versions.

#5690754

HARDRESET

2019-07-13 00:16

Only 5 min from Microcenter $294.47 Tax: 25.57 New Grand Total: $320.04 ASUS TUF B450M-PLUS Gaming AM4 mATX AMD Motherboar...[/paste:font] SKU: 80250410+ In Stock https://c773974.ssl.cf2.rackcdn.com/0510035_802504.jpg Price $49.99 Total $49.99 AMD Ryzen 5 3600 3.6GHz 6 Core AM4 Boxed Processor...[/paste:font] SKU: 95197025+ In Stock Available for In-Store Pickup Only. Price $199.99 G.Skill Aegis 16GB 2 x 8GB DDR4-3000 PC4-24000 CL1...[/paste:font] SKU: 2171829 in Stock Available for In-Store Pickup Only. https://c773974.ssl.cf2.rackcdn.com/0474187_217182.jpg Price $59.99 Total $59.99

#5690975

Evildead666

2019-07-13 22:29

Evildead666:

Its not a standard chipset, its the IO chip from the Ryzen CPU's. Its on a 12nm process, which I would have thought would have brought down the TDP, but since it was designed with massive throughput and low latency (being a CPU IO chip) it might not have much in the way of efficiency. I mean, its meant to be cooled with a CPU cooler, along with the CPU itself, but a large cooler nonetheless. edit : Cost would be the 12nm node i expect, much higher than the previous chipsets. Also, the fact AMD has to have them made, maybe in a bit of a rush, due to ASMedia's not being ready yet ? (New substrate fabbed just for the sole IO chip, validation and testing, putting the IO chip on the specific substrate...) edit2: The pinout is probably completely different from the previous chipsets also, so previous mobo designs can't be used..... It must have been a right clusterf**k for AMD when they found out the ASMedia chipsets weren't going to be ready in time.

i know its replying to me, but: Buildzoid sort of confirmed in one of his vids that the X570 doesnt have any power efficiency. Its either on or off, so thats one of the reasons for the active heatsink. As i said before, it wasn't meant to be passively cooled, or power gated that much.

#5690977

signex

2019-07-13 22:30

My GPU-Z is saying mine is running at PCI-E 4.0, but i suppose that's wrong?

#5691006

waltc3

2019-07-14 00:27

Should be interesting...can't see how these PCi4.x after-the-fact B4xx motherboards can compare to x570 chipset boards--the discrete chipset from AMD. There has to be a reason that the x570 chipset (reportedly) demands 11-15W, depending on the load--x470 should have difficulty on that basis alone. I hope it's true for everyone who has a qualifying Asus mboard (why on Earth the TUF?...;)) That would be most interesting! AMD really didn't have a problem with jerry-rigging these older boards to do some PCIex4.x, apparently, so they have given the OEMs permission to do it for a certain number of their mboard models-- leaving it up to the individual mboard OEM! I don't understand why they did not do this in the beginning--just leave the entire matter up to the respective mboard vendors!

signex:

My GPU-Z is saying mine is running at PCI-E 4.0, but i suppose that's wrong?

Well, The 5700's are PCIex4 cards, so do you have a PCIe4.x motherboard? I have a PCIe4.x motherboard, but unlike you my GPU is PCIe3.x, and GPU-Z shows it running @ 3.x performance.

#5691032

Clouseau

2019-07-14 05:57

The reason why so much wattage is that it is the exact same chip as the io chip in the 3000 series cpu. All the 4.0 success is from the io chip in the cpu. Only thing that matters on the motherboard side are how the traces are laid out. Given that the main M2 drive slots were directly connected to the cpu is why the M2 drives all have 4.0 connectivity. The interesting answer of all this would be what are the differences from the high end boards and the B450 boards that does not allow for 4.0 connectivity regarding the first PCIe graphics slot.

#5691091

Evildead666

2019-07-14 13:15

Clouseau:

The reason why so much wattage is that it is the exact same chip as the io chip in the 3000 series cpu. All the 4.0 success is from the io chip in the cpu. Only thing that matters on the motherboard side are how the traces are laid out. Given that the main M2 drive slots were directly connected to the cpu is why the M2 drives all have 4.0 connectivity. The interesting answer of all this would be what are the differences from the high end boards and the B450 boards that does not allow for 4.0 connectivity regarding the first PCIe graphics slot.

Maybe there's just not so much extra stuff on the B450 boards, the traces actually ended up better signal quality ? I would expect Asus to have actually tested these boards at PCIe 4 speeds. Either that, or maybe the X470 boards have a second x8 slot, and the physical drivers for splitting the primary slot are not compatible with PCIe 4 ?

#5691095

Mundosold

2019-07-14 13:30

HARDRESET:

Only 5 min from Microcenter $294.47 Tax: 25.57 New Grand Total: $320.04

That is really tempting.. I wonder if the mobo is pre-flashed for Ryzen support or not. Difference is I am 80 miles from Microcenter but I make the drive occasionally anyway.

#5691246

Vananovion

2019-07-15 09:25

Evildead666:

i know its replying to me, but: Buildzoid sort of confirmed in one of his vids that the X570 doesnt have any power efficiency. Its either on or off, so thats one of the reasons for the active heatsink. As i said before, it wasn't meant to be passively cooled, or power gated that much.

Der8auer actually measured the power consumption and it peaked at about 10W. While he said it is quite high, he was very much able to cool it with a cheap passive heatsink with no airflow peaking at 74C with no speed degradation. Although he didn't test a RAID situation, I bet with a beefier passive cooler and some airflow you'd be able to cool that as well. Der8auers video (interesting stuff starts at about 12:00): [youtube=qk3PD-4zPN0]

#5691251

Evildead666

2019-07-15 09:49

Vananovion:

Der8auer actually measured the power consumption and it peaked at about 10W. While he said it is quite high, he was very much able to cool it with a cheap passive heatsink with no airflow peaking at 74C with no speed degradation. Although he didn't test a RAID situation, I bet with a beefier passive cooler and some airflow you'd be able to cool that as well. Der8auers video (interesting stuff starts at about 12:00): [youtube=qk3PD-4zPN0]

The main problem, is that this solution has to work for everyone, even the guy/gal sitting in an apartment in the middle of Summer with 35°C ambient temps. They need to be able to cater for all possibilities..... For most enthusiast cases, where there is pretty good airflow, a good passive cooler will do fine. I would have thought one or more of the ThermalRight ones would have fitted...

#5691266

Astyanax

2019-07-15 13:04

Evildead666:

i know its replying to me, but: Buildzoid sort of confirmed in one of his vids that the X570 doesnt have any power efficiency. Its either on or off, so thats one of the reasons for the active heatsink. As i said before, it wasn't meant to be passively cooled, or power gated that much.

buildzoid has no credentials to his name.

#5691269

Evildead666

2019-07-15 13:27

Astyanax:

buildzoid has no credentials to his name.

“It's better to keep your mouth shut and appear stupid, than open it and remove all doubt” ― Mark Twain

#5691275

Astyanax

2019-07-15 14:33

Evildead666:

“It's better to keep your mouth shut and appear stupid, than open it and remove all doubt” ― Mark Twain

Buildzoid has no credentials to his name

#5691293

Hilbert Hagedoorn

Administrator

2019-07-15 15:54

Astyanax:

Buildzoid has no credentials to his name

You just got your last warning for continuously arguing, one more and you are out.

#5691483

Athlonite

2019-07-16 05:45

Fox2232:

CPU will not exceed set up power limits as long as it is physically possible. And that minimal power is like 20~25W for 3900X. For 2700X it is 16~17W.

until you want it to actually do something then it gets dicey the VRM is so crap I wouldn't put anything over a 65W CPU in it

#5691486

Fox2232

2019-07-16 07:06

Athlonite:

until you want it to actually do something then it gets dicey the VRM is so crap I wouldn't put anything over a 65W CPU in it

As I wrote. Take for example 2700X (105W) TDP chip. Set limit in BIOS to 65W and it will not exceed it. (As long as BIOS has this option since RM on runtime allows minimum of 70W for given chip.) And since people operated 2700X on those boards, it is really not that bad. But you can cook many boards by having no fans in case. I have seen test of HW Unboxed where they tested different boards and one of tests was case w/o running fans... ~120°C on VRMs, then one fan and something like 60~80°C. Surely it can handle itself running chip that will not exceed desired limit for average power.