4 and 5 fans! Why?
Rear I/O looks really disappointing, without Thunderbolt option for monitors.

Tek-CheckRear I/O looks really disappointing, without Thunderbolt option for monitors.

1. There's no integrated graphics capabilities, so why would there be display outputs?
2. For Thunderbolt, it's an Intel spec and if anything, there should've been USB4 support, but maybe ASRock felt the boards were expensive enough as they are and have enough PCIe slots that a USB4 add-in card is the way to go.

TheLostSwede2. For Thunderbolt, it's an Intel spec and if anything, there should've been USB4 support, but maybe ASRock felt the boards were expensive enough as they are and have enough PCIe slots that a USB4 add-in card is the way to go.

With how many PCIe lanes TR and TR Pro 7000 series have, there really should be a whole lot more USB4 ports. At least there's 10GbE connectivity.

wNotyarDWith how many PCIe lanes TR and TR Pro 7000 series have, there really should be a whole lot more USB4 ports. At least there's 10GbE connectivity.

No native support. Talk to AMD about it, it's not really the board makers fault.
Also, I don't know if ASMedia has cranked up the chip production enough as yet, since only MSI appears to have announced an add-in card.

Which idiot approved a board design with active fans?

TheLostSwede2. For Thunderbolt, it's an Intel spec and if anything, there should've been USB4 support, but maybe ASRock felt the boards were expensive enough as they are and have enough PCIe slots that a USB4 add-in card is the way to go.

What you mean is: ASRock cheaped out. Just like every motherboard manufacturer has cheaped out on USB4/TB4 on AM5.

Tek-Check4 and 5 fans! Why?
Rear I/O looks really disappointing, without Thunderbolt option for monitors.

Asrock seems to have lazy bums for engineers when it comes to VRM Cooling design, they seem to have gone down the route of slapping fans on VRMs since Z690 launch. Love them or hate them atleast Gigabyte has started with proper finned heatsinks on their motherboards.

What is so special about USB4 that the industry has been unable to cope with its large-scale introduction for so many years?

AssimilatorWhich idiot approved a board design with active fans?

Same ones who approved fans on Z690 and Z790.

AssimilatorWhat you mean is: ASRock cheaped out. Just like every motherboard manufacturer has cheaped out on USB4/TB4 on AM5.

Well, as I said, I'm not sure ASMedia has produced enough USB4 chips to enable the board makers to add USB4 to their boards as yet, since ASMedia was a year late as it was.

TumbleGeorgeWhat is so special about USB4 that the industry has been unable to cope with its large-scale introduction for so many years?

Nothing special at all.

However, there are multiple issues as to why it hasn't taken off.
Besides the complexity and cost of designing the chips themselves, which apparently only ASMedia was willing to do, it turned out that the documentation from the USB-IF was lacking due to a certain company who donated the specs and the documentation to the USB-IF, as the documentation made a lot of assumptions based on how what company does things. This has made it very hard to 1. use the documentation to make a product and 2. get certified.
Now that both AMD (inside some of their mobile SoCs) and ASMedia has managed to do it, other might follow, but I very much doubt we'll see a bunch of third party host controllers as we did with USB4, as there's no real financial incentive of doing it.
Instead, companies like VLI and Realtek seem more interested in focusing on peripheral ICs with native USB4 support.

There's more to it, but I can't go into any more details without stirring up shit for various USB-IF members, which is not someting I'm inclined to do.

Here's the block diagram btw.


via

Man, I wish AMD would at least up the lane count on X#70 boards to provide similar I/O options on PCIe, or offer a more budget TR with 7 or even 9 (like old XL-ATX) PCIe lanes.

It would be nice to see them follow up with a M-ATX powerhouse.
What ever you can get on the to handle the 7995, and max out the options.
Clean products thus far.

TechLurkerMan, I wish AMD would at least up the lane count on X#70 boards to provide similar I/O options on PCIe, or offer a more budget TR with 7 or even 9 (like old XL-ATX) PCIe lanes.

Most of the PCIe lanes are from the CPU, so would be impossible without a new socket.

I miss my TR system. The boards were not expensive and there was uber flexibility. For me it was how much different options you had for PCIe with all those lanes and that was only 64. Since Multi GPU (Unless MIning) is no longer a thing I have turned to Storage as an option for their I/O. It is too bad they abandoned us for Disney and Netflix content creators with no $999 chip to use with TRX40 but I had to come back to AM4 when the 5900/5950X launched as those chips made my 2920X look slow.

I was not until X570S that we got the ultimate I/O on AM4 but the board vendors made that a luxury that you have to pay for. Most X470 boards supports lane splitting on the top 2 connected to the CPU by the time X570 launched that went to the top M2 and PCIe lanes only.

For me it is laziness. The board on AM4 that shows what they could do if they put some effort in is the Asus B550-XE Wifi. If you doubt me look it up and drool.

These boards have my interest but V-Cache are the only CPUs I am buying going forward so I will probably have to wait or be out of luck.

ChaitanyaSame ones who approved fans on Z690 and Z790.

Dude. Just because Intel sells nuclear reactors don't blame ASRock.

TheLostSwedeWell, as I said, I'm not sure ASMedia has produced enough USB4 chips to enable the board makers to add USB4 to their boards as yet, since ASMedia was a year late as it was.


Nothing special at all.

However, there are multiple issues as to why it hasn't taken off.
Besides the complexity and cost of designing the chips themselves, which apparently only ASMedia was willing to do, it turned out that the documentation from the USB-IF was lacking due to a certain company who donated the specs and the documentation to the USB-IF, as the documentation made a lot of assumptions based on how what company does things. This has made it very hard to 1. use the documentation to make a product and 2. get certified.
Now that both AMD (inside some of their mobile SoCs) and ASMedia has managed to do it, other might follow, but I very much doubt we'll see a bunch of third party host controllers as we did with USB4, as there's no real financial incentive of doing it.
Instead, companies like VLI and Realtek seem more interested in focusing on peripheral ICs with native USB4 support.

There's more to it, but I can't go into any more details without stirring up shit for various USB-IF members, which is not someting I' inclined to do.

Here's the block diagram btw.


via

https://twitter.com/i/web/status/1717536399707287917

This is interesting and rather different to what I expected. We've got the dual x16 slots I wanted, but then an x8 slot instead of the two extra NVMes I anticipated. "Only" two NVMe drives, "only" one of which is PCIe 5.0, but that's rather trivially overcome by a PCIe add-in card. Asrock has repurposed the eight PCIe 3.0 lanes to be split half and half between a USB 20Gbps port and a 10GbE network chip.

Could be worse, could be better. I personally have no use for the MCIO or SlimSAS ports, but there's enough lanes allocated to more useful places that it won't kill me. The price on the other hand...

TheLostSwedeWell, as I said, I'm not sure ASMedia has produced enough USB4 chips to enable the board makers to add USB4 to their boards as yet, since ASMedia was a year late as it was.


Nothing special at all.

However, there are multiple issues as to why it hasn't taken off.
Besides the complexity and cost of designing the chips themselves, which apparently only ASMedia was willing to do, it turned out that the documentation from the USB-IF was lacking due to a certain company who donated the specs and the documentation to the USB-IF, as the documentation made a lot of assumptions based on how what company does things. This has made it very hard to 1. use the documentation to make a product and 2. get certified.
Now that both AMD (inside some of their mobile SoCs) and ASMedia has managed to do it, other might follow, but I very much doubt we'll see a bunch of third party host controllers as we did with USB4, as there's no real financial incentive of doing it.
Instead, companies like VLI and Realtek seem more interested in focusing on peripheral ICs with native USB4 support.

There's more to it, but I can't go into any more details without stirring up shit for various USB-IF members, which is not someting I' inclined to do.

Here's the block diagram btw.


via

https://twitter.com/i/web/status/1717536399707287917

I think also diminishing returns might be starting to kick in, I think USB 3.2 is plenty of bandwidth for most use cases. Plus I wonder if its a challenge to make these ports work at full speed on consumer grade boards as on my board e.g. only the front headers are the max bandwidth the ports at the back are half the speed (also only type C headers). Rear I/O also limited to just one type A port at 3.2 gen 2.

1 USB 3.2 Gen2x2 Front Type-C
2 USB 3.2 Gen2 (Rear Type A+C)
8 USB 3.2 Gen1 (4 Rear, 4 Front)

www.asrock.com/mb/Intel/Z690%20Steel%20Legend/index.asp

Surprised no one has made a front USB port panel to use front headers and slot into a 5inch bay at front of case, would be more useful than a USB PCIe addon card.

AssimilatorThis is interesting and rather different to what I expected. We've got the dual x16 slots I wanted, but then an x8 slot instead of the two extra NVMes I anticipated. "Only" two NVMe drives, "only" one of which is PCIe 5.0, but that's rather trivially overcome by a PCIe add-in card. Asrock has repurposed the eight PCIe 3.0 lanes to be split half and half between a USB 20Gbps port and a 10GbE network chip.

Could be worse, could be better. I personally have no use for the MCIO or SlimSAS ports, but there's enough lanes allocated to more useful places that it won't kill me. The price on the other hand...

Don't forget the MCIO and two SlimSAS ports that are all PCIe as well.

chrcolukI think also diminishing returns might be starting to kick in, I think USB 3.2 is plenty of bandwidth for most use cases. Plus I wonder if its a challenge to make these ports work at full speed on consumer grade boards as on my board e.g. only the front headers are the max bandwidth the ports at the back are half the speed (also only type C headers). Rear I/O also limited to just one type A port at 3.2 gen 2.

Front headers, yes, not going to work well with USB4, since it builds in Thunderbolt 3. Cable lenghts also seem to top out at about 1.5 meters, so it might not be super useful for a desktop, unless you can afford optical cables. That said, USB 3.2 is 1.8-2 meters, so not a big difference.

chrcolukSurprised no one has made a front USB port panel to use front headers and slot into a 5inch bay at front of case, would be more useful than a USB PCIe addon card.

It would be better to use one of those MiniSAS connectors for PCIe 4.0 x4 to the drive bay unit in this case and put the USB4 or Thunderbolt chip in that unit, but not likely to happen.

TheLostSwede1. There's no integrated graphics capabilities, so why would there be display outputs?
2. For Thunderbolt, it's an Intel spec and if anything, there should've been USB4 support, but maybe ASRock felt the boards were expensive enough as they are and have enough PCIe slots that a USB4 add-in card is the way to go.

You are aware that the Gigabyte TRX50 Aero D uses a Maple Ridge (Intel JHL8540) Thunderbolt 4 controller and offers two USB4 ports, which most likely also provide TB4-capabilities since it comes with a DP-in port as well?
This is similar to the older creator focused Gigabyte motherboards from the TRX40, B550, B650, X570 and X670 series that also support Thunderbolt via an add-in card with the older Titan Ridge, although not all of them have DP-in.

TheLostSwedeHere's the block diagram btw.

With such a new platform, it would actually be nice to include block diagrams in the original article. Especially since this motherboard seems to be a fairly weird ASRock-typical mix of both an extreme overclocker and an OEM-style workstation board.
And unless I missed some lanes while counting, there should actually be four more Gen 4 lanes available from the CPU that could be used for a TB3/4 controller. :D

Maybe you guys could try to get the block diagrams for the other TRX50 motherboards from Gigabyte and Asus as well?

kapone32I miss my TR system. The boards were not expensive and there was uber flexibility. For me it was how much different options you had for PCIe with all those lanes and that was only 64. Since Multi GPU (Unless MIning) is no longer a thing I have turned to Storage as an option for their I/O. It is too bad they abandoned us for Disney and Netflix content creators with no $999 chip to use with TRX40 but I had to come back to AM4 when the 5900/5950X launched as those chips made my 2920X look slow.

I was not until X570S that we got the ultimate I/O on AM4 but the board vendors made that a luxury that you have to pay for. Most X470 boards supports lane splitting on the top 2 connected to the CPU by the time X570 launched that went to the top M2 and PCIe lanes only.

For me it is laziness. The board on AM4 that shows what they could do if they put some effort in is the Asus B550-XE Wifi. If you doubt me look it up and drool.

These boards have my interest but V-Cache are the only CPUs I am buying going forward so I will probably have to wait or be out of luck.

Don't know why you think multi-gpu isn't going to comeback. at some point When you got from pure raytracing cards & away from anything with rasterization, you're going to need both cards type.
second there's plenty of games that support S.L.I Not so much for crossfire.
My issues right now is the Pure LACK of mGPU support any newer game have, when even a RTX 4090 can't play the game without using some kind of Band-Aid tech software like D.L.S.S & Frame generation just to get an acceptable 30-45fps. it's really annoying that developers are taking the propriety software way out of optimizing the games to run when they use DX12 support for Varible rate shading & mGPU to no need lower quality picturee.

DemonicRyzen666Don't know why you think multi-gpu isn't going to comeback.

Because it's not. NVIDIA and AMD determined that supporting mGPU generically is no longer worth it, do you really think that game devs - that lack the technical expertise of those companies - are going to be able, let alone willing, to implement mGPU? No, they aren't, because mGPU is a feature that a tiny minority of users will care about and so it makes zero financial sense to waste effort on implementing it.

mGPU is dead, it is never coming back, and your wishing is not going to make it come back.

AssimilatorBecause it's not. NVIDIA and AMD determined that supporting mGPU generically is no longer worth it, do you really think that game devs - that lack the technical expertise of those companies - are going to be able, let alone willing, to implement mGPU? No, they aren't, because mGPU is a feature that a tiny minority of users will care about and so it makes zero financial sense to waste effort on implementing it.

mGPU is dead, it is never coming back, and your wishing is not going to make it come back.

*mGPU is dead for games. It is more practical than ever if you are working with large models in ML/AI.

SamuelL*mGPU is dead for games. It is more practical than ever if you are working with large models in ML/AI.

... the person I responded to was talking about games, as was I, so I don't understand why you felt it necessary to add this comment.

RedwoodzDude. Just because Intel sells nuclear reactors don't blame ASRock.

Most competiting boards managed to cool VRMs without active cooling so it certainly was Asrock's laziness in not having a proper heatsink. Even with Threadripper 7000 its only Asrock who has added fans(atleast they are easy to replace unlike in past) while all competiting boards have passive cooling solutions.

Assimilator... the person I responded to was talking about games, as was I, so I don't understand why you felt it necessary to add this comment.

I take offense to you for laughing at me. -_______-

One, you failed to comprehend what I said.
Multi GPU will eventually comeback, because it will take the place of being able to use older cards when new cards that are pure raytracing only, won't be capable to render through rendering rasterization Methods anymore.

Two, practically all of AMD's RDNA line up currently support mGPU. It is Nvidia who doesn't support mGPU properly.

It's not my fault the PC gaming market is full of uninformed consumers that only buy Nvidia with terrible value. It's also NVidia who's priced their cards from the last generation tiers 50% to 100% more. That's why no one is really buying them & they're sitting on shelfs mostly. That's makes your third argument completely irrelevant to the cost of building games to support like mGPU, claiming it's a minority. Practically any game that is built with D.L.S.S 3/3.5 frame generation currently is exactly that. Currently built for a minority market that can afford that right now. To say that it's financially acceptable right now for games to be built right with D.L.S.S frame generation is completely wrong.

TheLostSwede1. There's no integrated graphics capabilities, so why would there be display outputs?
2. For Thunderbolt, it's an Intel spec and if anything, there should've been USB4 support, but maybe ASRock felt the boards were expensive enough as they are and have enough PCIe slots that a USB4 add-in card is the way to go.

It's weird with Asrock. They have some brilliant boards, such as Creator line across sockets. I have one of the best ITX boards on Z390 platform from them. When they put in effort, ambition and use imagination, they make most amazing boards competing with top boards from Asus. However, they are also capable of dodgy products. I/O on TRX50 in one of them. It's silly cheap on a very expensive board and this needs to be called out clear and loud.

If we compare new HEDT/WS boards with TXR40 and WRX80 from Zen 3 platform, it is clear that no major modernisation of boards took place, apart from necessary PCIe 5.0, memory and power stages upgrade. In fact, some important regressions took place. Zen 3 WS board had 14 layer PCB, dual Thunderbolt with two DP IN ports and WiFi 6E.

It does not really matter there is no iGPU. On such high-end board, there is no excuse not to provide DP IN ports, so that users can bring in video from GPU and use fast PCIe transfer for large media files in ecosystem that is already prevalent with diverse Thunderbolt solutions installed on storage and audio-video devices.

It does not really matter that Thunderbolt is Intel's solution. They did not care about it on Zen 3 boards, so why would they care now? Sure, they could have installed new ASM4242 chip, but I am not sure if it's ready for commercial use. I do not agree with USB4-AIC idea. The board is expensive enough on its own and it needs to meet higher I/O standard instead of putting even more pressure on users' wallets. Cheapening on I/O is understandable of low-end desktop boards, but this... Asrock, get a grip!

AssimilatorWhat you mean is: ASRock cheaped out. Just like every motherboard manufacturer has cheaped out on USB4/TB4 on AM5.

The paradox is that board vendors rushed with PCIe 5.0 support across GPU and NVMe slots, and almost nobody in the world uses those capabilities, while gatekeeping TB4 to top and halo designs only. Absurd.

TumbleGeorgeWhat is so special about USB4 that the industry has been unable to cope with its large-scale introduction for so many years?

Good question. USB4 native controllers are simply late. It's a total mess of partial solutions that have not even reached the market. Blame it on Covid, R&D costs, delay of DP 2.1, etc.

Also, only Intel are working on more advanced connectivity solution from USB4 v2 spec with 80 Gbps ports. AsMedia and ViaLabs have 40 Gbps solution, but ViaLabs is DP 1.4 and USB only, so no PCIe traffic, and AsMedia 4242 does have PCIe 4.0 x4, but it does not have DP 2.1 support (it's USB4 v1). I have not heard from Realtek and JMicron developing USB4 in any form or shape. It's a Wild West...

Apparently, Intel did not pass onto USB-IF "all necessary" information about Thunderbolt 3, and so USB4 engineers needed to retro design and implement several solutions. While I am open to believe that Intel enjoys playing games sometimes, I am also surprised that it has been taking so much time to release one single USB4 Gen 3x2 host controller with all features. It's still not there and no one has announced any R&D work on USB4 Gen 4 (80 Gbps solution), to the best of my knowledge.

USB4, when implemented fully, is special because it allows us to use several protocols and power solution over one single port. That's it. Its beauty is simplicity.