VIA Labs VL832 USB4 Device Achieves USB-IF Certification

[TheLostSwede]

VIA Labs, Inc. (VLI), a leading supplier of USB4, USB 3.2, and USB Power Delivery Controllers, today announced that the VL832 USB4 Endpoint Device Controller achieved USB4 certification from the USB Implementor's Forum (USB-IF). VIA Lab's VL832 supports USB 40 Gbps operation and is now listed on the USB-IF Integrator's List under TID: 10033.

VIA Lab VL832's USB-IF Certification represents a key milestone in the USB4 ecosystem. The certified USB4 device controller with an integrated USB 3.2 USB 10 Gbps Hub, USB 2.0 Hub, and DisplayPort output, provides essential connectivity for peripheral devices such as multi-function adapters and docking stations. In USB4 40 Gbps mode, VL832 supports full DisplayPort HBR3 bandwidth (32.4 Gbps), and the USB 20 Gbps hub can enable multiple USB 10 Gbps devices to operate at full performance on supported host platforms. Both metrics are double what was previously possible using DisplayPort Alternate Mode, which is limited to 2-lanes of DisplayPort when supporting USB 3.2 functionality.




The increased display bandwidth perfectly aligns with today's display trends, including transitioning to QHD (2560 x 1440) and higher resolutions and a growing emphasis on providing users with more detailed visuals and richer, more accurate colors through wide color gamut and HDR technologies. It also caters to the demands of gaming enthusiasts for more fluid and immersive experiences by supporting higher refresh rates, such as uncompressed 4K at 120 Hz or even 240 Hz+ with Display Stream Compression (DSC). Additionally, the capability to connect and efficiently run up to four 4K displays at a consistent 60 Hz refresh rate is a significant advancement for creators and professional and business environments.

"The VL832 builds on the momentum of the highly successful VL830, which has been adopted in nearly 50 customer designs worldwide, powering a diverse range of products from travel docks to monitors," said Terrance Shih, Business Development Director, VIA Labs, Inc. "While VL830 targeted projects with a captive upstream port cable, VL832 features enhanced signal integrity, giving us the flexibility to support more designs with detachable cables."

"USB-IF's distinction as the leading interconnect standard extends beyond familiarity and usability—it lies in the robust testing and certification program that guarantees consumers a seamless computing experience," said Jeff Ravencraft, USB Implementers Forum (USB-IF) President and COO. "Certified USB-IF products, such as those from VIA Labs, signifies a strategic investment in the success of the technology, and significantly contributes to the sustained growth and widespread acceptance of the USB ecosystem."

VL832 Availability
The VIA Labs VL832 features five downstream USB ports and a DisplayPort output, and is offered in a compact FCCSP 10x10mm package. When paired with a USB PD controller solution such as the VIA Labs VL108 with Extended Power Range (EPR) support, compatible host systems can enjoy charging rates of 140 W or more. The VL832 USB4 Endpoint Device Controller is available now and is shipping in quantity.

View at TechPowerUp Main Site | Source

 

[Tek-Check]

Where is controller supporting DP 2.1?

 

[Assimilator]

Just in time for Zen 6 - this might actually make it a viable upgrade compared to the extremely anemic Zen 5, assuming VIA has managed to not implement this terribly.

Where is controller supporting DP 2.1?

Gonna be a little difficult for 40Gbps USB4 Gen 3×2 to support 80Gbps DP 2.0+, we'd need USB4 Gen 4 and I don't know of anything that natively supports the latter yet, so supporting controllers are going to be waaay further out. And honestly, the ~26Gbps of DP 1.4 already offers a massive amount of display bandwidth unless you're trying to drive multiple 4K monitors at high refresh rates.

 

[LabRat 891]

Admittedly, I'm probably a little too excited to put VIA in a modern computer...
Still, it's neat seeing VIA make ways back into the PC sector.

 

[Ferrum Master]

Admittedly, I'm probably a little too excited to put VIA in a modern computer...
Still, it's neat seeing VIA make ways back into the PC sector.

Don't forget to install Hyperion driver with PCI latency fixes.

 

[FreedomEclipse]

Article: "Via labs..."

Me:

 

[Ferrum Master]

Article: "Via labs..."

It seems we have fond memories of time when market wasn't boring, we had a choice having VIA, SIS, nForce, Uli and only then AMD and Intel motherboards from myraid of motherboard OEMs... EPOX, ABIT, DFI, SOLTEK, AOPEN, Chaintech, Soyo etc. Healthy open market. Now what? Look at the prices...

 

[Assimilator]

It seems we have fond memories of time when market wasn't boring, we had a choice having VIA, SIS, nForce, Uli and only then AMD and Intel motherboards from myraid of motherboard OEMs... EPOX, ABIT, DFI, SOLTEK, AOPEN, Chaintech, Soyo etc. Healthy open market. Now what? Look at the prices...

The north bridge was what allowed those vendors to differentiate themselves, e.g. by providing a better moemory controller or additional PCIe lanes. Once that moved into the CPU you really only had the south bridge, which is just IO and thus not enough to provide a compelling value prospect. Today, even the SB is being subsumed into the CPU more and more.

Personally I see a market for third-party south bridges that provide additional PCIe lanes, but my guess is that the CPU manufacturers will never allow this because they're more interested in control over innovation. And the actual motherboard manufacturers aren't interested in innovating either - to them "innovation" is RGB and metal PCIe slots.

 

[TheLostSwede]

The north bridge was what allowed those vendors to differentiate themselves, e.g. by providing a better moemory controller or additional PCIe lanes. Once that moved into the CPU you really only had the south bridge, which is just IO and thus not enough to provide a compelling value prospect. Today, even the SB is being subsumed into the CPU more and more.

Personally I see a market for third-party south bridges that provide additional PCIe lanes, but my guess is that the CPU manufacturers will never allow this because they're more interested in control over innovation. And the actual motherboard manufacturers aren't interested in innovating either - to them "innovation" is RGB and metal PCIe slots.

Anyone could in theory make an I/O hub for AMD CPUs, since the interface is PCIe, but clearly no-one is interested, as they're not cheap to develop.

 

[Ferrum Master]

Anyone could in theory make an I/O hub for AMD CPUs, since the interface is PCIe, but clearly no-one is interested, as they're not cheap to develop.

With exception of Zhaoxin and maybe Sony and Microsoft for their consoles.

 

[Assimilator]

Anyone could in theory make an I/O hub for AMD CPUs, since the interface is PCIe, but clearly no-one is interested, as they're not cheap to develop.

Not sure why it's not cheap, considering it's literally just an aggregation of IO.

 

[Ferrum Master]

Not sure why it's not cheap, considering it's literally just an aggregation of IO.

Licensing?

 

[TumbleGeorge]

Licensing?

The licensing fee is US$3,500 for a two year term (this fee is waived for USB-IF members).

I think that also has small tax per unit.

 

[TheLostSwede]

With exception of Zhaoxin and maybe Sony and Microsoft for their consoles.

Huh? Zhaoxin is more or less VIA and is making its own CPUs and chipsets that are as far as I'm aware, not supported by AMD.
As for consoles, I presume AMD and/or ASMedia made those designs for the consoles.

Not sure why it's not cheap, considering it's literally just an aggregation of IO.

Do you have any idea what it costs to design somehting like that, just in terms of the engineering team you need to have?
There's a reason why everyone else have let ASMedia take over this space, as neither VIA or Realtek have even bothered with a USB4 host controller for starters, nor any of the other companies that made USB 3.0 host controllers. The cost vs the reward is simply not there and with AMD apparently having picked ASMedia as their prefered partner, it's going to be really hard for someone else to break into this market and they'd obviously have get certified by AMD as well.
My understanding is that AMD isn't opposed to the idea of a second source for chipsets for the AM4/AM5 platforms, but no-one has been willing to jump onboard.
You'd need at least a dozen engineers that are familiar with 20 or so different high-speed and low-speed interfaces outside of PCIe to make a decent chipset and they'd also have to work out how to communicate with AMD's CPUs/APUs over various interfaces such as GPIO, SPI, SMBus, I²C and so forth to make it all work as intended.
Most of these guys have retired over the years or moved on to other positions in the companies that used to make these products and although other SoCs also rely on the same interfaces, these types of engineers are in high demand and are not easy to poach if they're any good at what they're doing.

This doesn't even take into account how much it costs to tape out and manufacture a chip at a somewhat recent node. A tape out costs millions of US dollars on modern nodes, a couple of millions on something like 12-16 nm. Based on estimates of TSMC wafer pricing, a wafer on the N5 node three years ago used to cost around US$13,500, but you obviously would get a ton of chips out of that, if you've done your job right. As companies like to recouperate their investments within a year or two, the first bunch of chips are going to cost more than the market is likely going to want to pay for them just to get a second source to ASMedia.

Licensing?

Nah, there's not much of that needed, but there are a few industry standards consortiums that the company making a chipset would have to be part of, but it's peanuts compared to the overall cost of making a chipset as per above. However, if it would be a startup, it would most likely take them a couple of years to join all the various industry standards groups and get access to all the documentation and standards.

 

[LabRat 891]

Not sure why it's not cheap, considering it's literally just an aggregation of IO.

I'm not either, as just bundling an off-the-shelf ASM PCIe Switch, ASM PCIe-SATA and PCIe-USB bridge together, would constitute a "Southbridge".
Even ASrock has made bundled add-in cards with a 'Southbridge' on-board, for expanded capabilities. The same kind of device, is fairly common in devlabs too.

I've used newer gen PCIe switch cards on old PCIe 1.1 systems, almost precisely the way a Southbridge once would've.

 

[Ferrum Master]

Huh? Zhaoxin is more or less VIA and is making its own CPUs and chipsets that are as far as I'm aware, not supported by AMD.

At this moment we don't know nothing about it. The key is in the fact that in their vision it is like what your technologies... you mean our comrade? So... it could be anything.

ITE and ENE if they wish could make a bridge as they are still present pretty much on any board. There are some other less popular.

 

[TheLostSwede]

ITE and ENE if they wish could make a bridge as they are still present pretty much on any board. There are some other less popular.

They don't make anything using PCIe/NVMe, USB, SATA or Ethernet related though, so how they're going to magically make a chipset, I don't know.
USB4 is apparently a real PITA to make compliant and as I've mentioned before, there's a reason why ASMedia was a year late and most of it has to do with poor documentation from you know who.
Yes, those companies can handle all the other lower speed interfaces easily enough and most likely make all the singal syncing work no problem but where's the monetary incentive for them? They'd still have to spend a good bit of money on hiring new engineers, tape out a new chip and then make a few hundred thousand units for a market that isn't really there due to no-one asking for a second source.

 

[Ferrum Master]

They don't make anything using PCIe/NVMe, USB, SATA or Ethernet related though, so how they're going to magically make a chipset, I don't know.

But they actually do exactly things that make south bridge special, they have the firmware hub that actually wakes and manages they whole board, sensors, legacy I/O and then we have our set USBs, Azalia, SATA, NIC or other special stuff like that cnvio proprietary crap intel makes. I think there is some licensing conflict as no other makers have tried to integrate it inside the bridge and get rid of firmware hub makers.

 

[TheLostSwede]

But they actually do exactly things that make south bridge special, they have the firmware hub that actually wakes and manages they whole board, sensors, legacy I/O and then we have our set USBs, Azalia, SATA, NIC or other special stuff like that cnvio proprietary crap intel makes. I think there is some licensing conflict as no other makers have tried to integrate it inside the bridge and get rid of firmware hub makers.

Nah, there are no issues there, there's simply no money for more players in that market. Most of these peripheral chipset makers are trying to find a niche for themselves.
Why do you think that there are no other real options to Realtek in the PC audio market? Simply because no-one else can compete with them on cost, so the companies that used to compete with them withdrew to other niches.
Same for Ethernet chips, today we pretty much only have Intel or Realtek for the stuff that isn't PCIe based, unless it goes into a router or it's faster than 2.5 Gbps.

There has to be a certain amount of revenue to be had from doing something like this for there to be any kind of interest.

 

[Ferrum Master]

no-one else can compete with them on cost, so the companies that used to compete with them withdrew to other niches.

I smell simply a price fixing scheme and agreement between those players to screw us all over... prove me wrong

 

[qcmadness]

I smell simply a price fixing scheme and agreement between those players to screw us all over... prove me wrong

It is the cost of design and tape-out that prohibits further players in the industry.

And you will not need that "chiplet" or "I/O hub" in a notebook / SFF builds.

 

[TheLostSwede]

I smell simply a price fixing scheme and agreement between those players to screw us all over... prove me wrong

That would be impossible, but I'm this case it doesn't make sense to have too many players.

 

[Ferrum Master]

That would be impossible

Remember 2010 and Cartel case between Micron, Samsung, Hynix, Infineon, NEC, Hitachi, Mitsubishi, Toshiba, Elpida and Nanya? And LCD case same year with Samsung Electronics and LG Display of Korea and Taiwanese firms AU Optronics, Chimei InnoLux Corporation, Chunghwa Picture Tubes and HannStar Display Corporation... IMHO the last one ain't settled yet.

They are all scheming bastards and drinking beer in the same bar after work... .

Either way ENE and ITE does make those bridges for laptops, there is nothing that would stop them to do it for ATX boards too, they have it all, actually motherboard vendors could make a board without any ASMEDIA or AMD/INTEL South Bridge at all already, you won't loose anything really. Look how fierce mobile SoCs are competing, in any areas actually. We have shit SoC like UniSoc(ex Spreadtrum), Mediatek as middle, Qualcomm, Kirin(Broadcomm?), Apple Ms... going top offering a full package, and fun part there are even more makers. PC reeks of weird stagnation... it contradicts if we compare crowded mobile with PC, we are being bullied actually(imagine smiling Jensen Huang).