Contribute
Register

Asus Z690 ProArt Creator WiFi (Thunderbolt 4) + i7-12700K + AMD RX 6800 XT

Maple Ridge Thunderbolt users: a new competitor had emerged: the ASMedia ASM4242 chip, a USB4 PCIe4 chip.

Here is a video (in Dutch I believe) from computex:

There was a demo of an NVMe drive in an ASMedia USB4 NVMe enclosure connected to the MSI ASMedia USB4 PCIe card. Disk transfer of 4 GB/sec. Compared to approx 2.8 GB/sec that we get on Maple Ridge.

Hopefully, the emergence of this USB4 solution will drive down the price of USB4 peripherals in the market place… 4 GB/sec over a USB4 wire looks beautiful to me. This may even inspire Intel to release TB5 sooner rather than later.
 
… Disk transfer of 4 GB/sec. Compared to approx 2.8 GB/sec that we get on Maple Ridge.
If this is an apples-to-apples comparison (i.e., only difference is the USB4 controller) then it indicates some significant throughput issues with Maple Ridge.
 
If this is an apples-to-apples comparison (i.e., only difference is the USB4 controller) then it indicates some significant throughput issues with Maple Ridge.
Well, the ASMedia chip is using PCIe4x4 vs PCIe3x4 for Maple Ridge.
 
Maple Ridge Thunderbolt users: a new competitor had emerged: the ASMedia ASM4242 chip, a USB4 PCIe4 chip.

Here is a video (in Dutch I believe) from computex:
Definitively Dutch. (This expatriate cannot resist a snark, and shall point out that, in typical Dutch fashion, the video puts all emphasis on a single feature: "cheaper".)

But a quick search reveals that it is neither a new product (it was announced last year) nor a Maple Ridge competitor:
From its official product page, the ASM4242 is strictly a USB4 controller, not Thunderbolt 4. (The video got it right.)
As such it is compatible with Thunderbolt 3, not 4; it might be a Titan Ridge competitor.
ASM4242 was reportedly used on some AM5 boards (X670 overview from Sept. 2022):
On some boards, you’ll find USB 4 support via an ASMedia ASM4242 controller, which outputs two USB 4 ports from four PCIe 4.0 lanes via the CPU, while also supporting DisplayPort v1.4 (thus video) through the interface.
There have been mentions (here and here) of a Gigabyte X670E Aero D, which appears not to have been actually released.
So maybe this MSI card will be the first actual product to ship with an ASM4242 after all…

Well the asmedia chip is using pcie4x4 vs pcie3x4 for maple ridge.
PCIe 3.0 x4 has (almost) 4 GByte/s of potential throughput, so getting 2,8 GByte/s out of it indeed indicates some serious bottleneck.
 
But a quick search reveals that it is neither a new product (it was announced last year) nor a Maple Ridge competitor:
From its official product page, the ASM4242 is strictly a USB4 controller, not Thunderbolt 4. (The video got it right.)
I kindly disagree.

Intel considers Maple Ridge to be USB4 compliant, no? Therefore, Maple Ridge doubles as both a Thunderbolt 4 controller and a USB4 controller, no?

Remember Thunderbolt 4 is a spec that guarantees a minimum level of performance. The problem with USB and even Thunderbolt 3 is that many of the features are optional and end users don’t necessarily know what features a so-called USB/Thunderbolt product will support. For example, with Thunderbolt/USB4, does the port support PCIe tunelling or not? Who knows?

So Thunderbolt 4 guarantees a minimum baseline spec of PCIe tunneling enabled at up to 32 Gb/sec, backwards compatibility with USB3 10 Gb/sec, DisplayPort/Ethernet tunneling, support for 2 4K monitors @ 60 Hz etc. Essentially, Thunderbolt 4 is USB4 with all the optional features enabled.

With Thunderbolt 3, some vendors only used 16 Gb/sec, some didn’t support more than 1 display output, some didn't enable PCIe tunneling, some Thunderbolt 3 devices didn’t have any backwards compatibility with USB3 (e.g., some Alpine Ridge based NVMe enclosures must be plugged into a Thunderbolt port, if you connect them to a ‘regular’ USB3 USB-c port, the device will not power on… that is confusing!).

The other thing with Thunderbolt is that devices with the logo must be certified by Intel. Which adds to cost. Before USB4 came out, Thunderbolt 3/4 cables from vendors such as Anker and cable matters and others cost $50-$60. With the introduction of USB4, costs of cables have dramatically declined. A certified USB4 cable is thunderbolt compliant, and many cost 1/3 of what thunderbolt4 cables used to cost. For example, I just bought a Cable Matters USB4 cable with support for USB-pd3.1 240W power delivery, and 40 Gb/sec data for only $20. A more fully featured cable than Thunderbolt 4 (which maxes out at 100W power delivery) at a fraction of the cost. That’s what I mean when I say prices will decline when more USB4 devices come to market.

Finally, given that the ASMedia ASM4242 appears to have all the optional usb4 features enabled, for all we know, ASMedia could eventually apply for Thunderbolt 4 certification. But even if it does not, the chip is still USB4 certified, and by definition will be compatible with at least some Thunderbolt 3 and 4 end devices such as nvme enclosures. By the way, there aren’t a lot of Thunderbolt 4 end devices (such as eGPUs or nvme enclosures or audio interfaces) on the market. Of course, as we have learned with maple ridge, the firmware version of both the host and the endpoint device will dictate compatibility when the two devices are connected to each other. Maple ridge nvm36 has removed compatibility with thunderbolt 1 and 2 devices, and earlier versions of Maple Ridge’s firmware has hot plugging incompatibilities with Alpine Ridge jhl6240 devices. It will be interesting to see what level of compatibility exists between the ASMedia chip and Alpine Ridge devices.

Finally, the reason why I say it is a Maple Ridge competitor is this. Several Ryzen am5 x670e motherboards from Asus, ASRock, etc., claim to have USB 4 support. But they are using Maple Ridge JHL8540! The emergence of this ASMedia chip means that the vendors will have a choice of USB4 suppliers for the next generation of motherboards! The demo showed nvme devices transferring at 4 GB/sec, faster than Maple Ridge. Asm4242 also supports USB 3 20 Gbps, whereas Maple Ridge tops out at only USB3 10 Gbps. The ASM chip also operates at PCIe4.0 which theoretically could mean faster Egpu speeds and higher frame raes… and be a better occulink competitor. Thus, the ASM chip has inherent advantages over Maple Ridge.

I think this will put natural pressure on Intel, the co-developer of Thunderbolt and who donated the Thunderbolt 3 spec to the USB-if. Because what is the benefit of a motherboard vendor going with Maple Ridge if it is slower than the asm4242? Not a good look for the primary Thunderbolt/USB4 sponsor to have a slower solution than ASMedia.

But I think intel welcomes this competition because it donated Maple Ridge Thunderbolt 3 to the USB-if…why open source Thunderbolt if you don’t welcome competition and choice in the market place?
 
Last edited:
The ASMedia USB4 controller is very interesting because of the data transfer speeds cited in a previous post.

I’m not concerned whether it’s Thunderbolt 4 or Thunderbolt 3 or any other semantic difference. Both TB3 and TB4 are theoretically maxed out at the same MT/s.

What fascinates me is whether PCIe 4.0 is responsible for the improved throughput of the ASMedia controller, or whether we would get the same great results by installing the card in a PCIe 3.0 x4 slot.

How much of the improvement, in other words, is due to PCIe version and how much is due to just a better implementation…
 
The ASMedia USB4 controller is very interesting because of the data transfer speeds cited in a previous post.

I’m not concerned whether it’s Thunderbolt 4 or Thunderbolt 3 or any other semantic difference. Both TB3 and TB4 are theoretically maxed out at the same MT/s.

What fascinates me is whether PCIe 4.0 is responsible for the improved throughput of the ASMedia controller, or whether we would get the same great results by installing the card in a PCIe 3.0 x4 slot.

How much of the improvement, in other words, is due to PCIe version and how much is due to just a better implementation…
I like how you think Casey.
 
This must mean it’s coming any day now…

Appears to have smaller bezels than the Apple Studio Display. Most importantly, a smaller price tag.

Viewfinity_S9_vs_Studio_Display.jpeg
 
Appears to have smaller bezels than the Apple Studio Display. Most importantly, a smaller price tag.
It looks very Apple-esqe and has two video inputs (TB4 and Mini-DP), but a matte finish. Waiting patiently for the reviews and a chance to see it in person...

From the Samsung product page:
Screenshot 2023-06-03 at 1.05.05 PM.png


Screenshot 2023-06-03 at 1.10.30 PM.png

Screenshot 2023-06-03 at 1.11.40 PM.png

Screenshot 2023-06-03 at 1.18.13 PM.png
 
Last edited:
Back
Top