ASMedia to Launch USB4 Host-Controllers This Year

[TheLostSwede]

Taiwanese ASMedia appears to become the first peripheral chipset company to launch a USB4 host-controller chipset, based on reports from an event the company held this week. ASMedia is at least the first company to get certified by the USB-IF, as its ASM4242 host-controller is the first of its kind to get certified. The ASM4242 was certified alongside the ASM2464PD device controller, although it's unclear exactly what type of device controller this is, but the PD in the model name indicates that it has native support for USB PD power delivery, without the need of any additional chips. This should allow for simpler implementation, as well as it would be saving some PCB space that the extra components no longer take up.

The ASM4242 is said to have a PCIe 4.0 x4 interface and does as such support up to 64 Gbps of bandwidth. As this is a two port controller, it's still not quite enough to enable both ports to operate at the full 40 Gbps that USB4 Gen 3x2 supports, but it should be enough for most consumer implementations. ASMedia has also added support for DP Alt Mode and USB4 is of course also backwards compatible with Thunderbolt 3. ASMedia is already sampling its partners. We should hopefully get a better look at what ASMedia is working on next week, but DigiTimes mentions that ASMedia has also developed PCIe 5.0 controllers of some kind and have finished the tape-out of said products.



View at TechPowerUp Main Site | Source

 

[Nanochip]

Excellent news to see a non-Intel USB4 / TB-compatible controller come to market. This should put downward pressure on pricing and allow the price of pcie-based NVME enclosures to fall.

 

[AsRock]

Excellent news to see a non-Intel USB4 / TB-compatible controller come to market. This should put downward pressure on pricing and allow the price of pcie-based NVME enclosures to fall.

I hope it's better than the last time experiences with ASMedia.

 

[Tek-Check]

If it's one port solution then it could deliver UHBR10 DP 2.0 Alt Mode at 40 Gbps, but if it's two ports, that would be a challenge. Interesting to see whether both ports share x4 connection

 

[THU31]

USB4 Gen 3x2.

http://imgur.com/gTtpqkm

 

[eidairaman1]

I hope it's better than the last time experiences with ASMedia.

I miss Via Technologies

 

[AsRock]

I miss Via Technologies

Well last time i had to deal with there crap was back in 1998-2002 time, the first warning that me and Gigabyte never get along haha.

EDIT: i guess ASMedia always worked just not very well for me were as VIA i found them very troublesome.

 

[TheLostSwede]

USB4 Gen 3x2.

http://imgur.com/gTtpqkm

Hey, it's what the 40 Gbps implementation is called.

I miss Via Technologies

They're still around, their division that does USB stuff is called VLI now. A lot of the people that founded ASMedia are ex VIA staff. There was a bit of a legal kerfuffle a few years ago. As you might know, Asus pitched in to have ASMedia started.

 

[tpu7887]

I hope it's better than the last time experiences with ASMedia.

What was your last (bad) experience with ASMedia?

I've only had good experiences recently (and previously to the best of my memory)
I have a SATA RAID card and USB controller card, and both couldn't work better

About this article: I have a card with their 3242 on it, plugged into the x4 slot supplied by the chipset my z390 motherboard. Gives my machine USB 3.2 Gen 2x2 support!
It works quite well - perfectly in fact. I use it for an external NVMe drive
If the 3242 is any indication of how the 4242 will be when released, it will be GREAT!

edit: I wonder if they will support USB 3.2 Gen2x2 (20Gbps) on the 4242, because it's optional to.

It is mandatory for all USB 4 controllers to support 10Gbps single lane USB (plain USB 3.2 Gen 2), but since USB 4 doesn't use both sides at once for data (it's reserved for display and such), it's something that would take a little bit of work to implement. A very little bit of work, not much work at all really, no reason why all USB 4 controllers shouldn't support 3.2 Gen 2x2. When USB-IF were putting the standard together, they should have just made it mandatory for ALL USB4 controllers to support 20Gbps USB 3

If the ASM4242 supports 20Gbps USB 3.2 Gen 2x2 I will definitely buy it and replace the 3242 in my z390 (which I plan to keep as my primary system for the next 3 years and a secondary for 4 or so after). The 3242 would go into another system.
If the ASM4242 only does 10Gbps USB 3.2, I don't know how I'd add it to my system. My 3080 takes up all 16 PCIe 3.0, and it's 3.0 so running at x8 is would probably affect performance

 

[TheLostSwede]

What was your last (bad) experience with ASMedia?

I've only had good experiences recently (and previously to the best of my memory)
I have a SATA RAID card and USB controller card, and both couldn't work better

About this article: I have a card with their 3242 on it, plugged into the x4 slot supplied by the chipset my z390 motherboard. Gives my machine USB 3.2 Gen 2x2 support!
It works quite well - perfectly in fact. I use it for an external NVMe drive
If the 3242 is any indication of how the 4242 will be when released, it will be GREAT!

edit: I wonder if they will support USB 3.2 Gen2x2 (20Gbps) on the 4242, because it's optional to.

It is mandatory for all USB 4 controllers to support 10Gbps single lane USB (plain USB 3.2 Gen 2), but since USB 4 doesn't use both sides at once for data (it's reserved for display and such), it's something that would take a little bit of work to implement. A very little bit of work, not much work at all really, no reason why all USB 4 controllers shouldn't support 3.2 Gen 2x2. When USB-IF were putting the standard together, they should have just made it mandatory for ALL USB4 controllers to support 20Gbps USB 3

If the ASM4242 supports 20Gbps USB 3.2 Gen 2x2 I will definitely buy it and replace the 3242 in my z390 (which I plan to keep as my primary system for the next 3 years and a secondary for 4 or so after). The 3242 would go into another system.
If the ASM4242 only does 10Gbps USB 3.2, I don't know how I'd add it to my system. My 3080 takes up all 16 PCIe 3.0, and it's 3.0 so running at x8 is would probably affect performance

I should be seeing them next week, so I'll ask for you.

 

[LabRat 891]

The only thing attractive to me about USB 4 is the PCIe externalization.
IIRC, PCIe encapsulation (like ThunderBolt) is 'optional' in USB 4. I didn't see a direct mention in the article of the feature, so I assume this will be a rather underwhelming USB controller.
Wasn't X570 was supposed to have USB4 as an option? I think *ONE*, maybe 2 ultra-premium workstation boards had it; not a good start for USB 4. (Not to mention the (now-popular) non-sense naming schemes: Wi-Fi 6, USB # Gen# #x#, HDMI and DP also now have non-standardized 'standards' too.)

"THE NUMBERS MASON, WHAT DO THEY MEAN?!"
they mean nothing....

 

[Xajel]

The only thing attractive to me about USB 4 is the PCIe externalization.
IIRC, PCIe encapsulation (like ThunderBolt) is 'optional' in USB 4. I didn't see a direct mention in the article of the feature, so I assume this will be a rather underwhelming USB controller.
Wasn't X570 was supposed to have USB4 as an option? I think *ONE*, maybe 2 ultra-premium workstation boards had it; not a good start for USB 4. (Not to mention the (now-popular) non-sense naming schemes: Wi-Fi 6, USB # Gen# #x#, HDMI and DP also now have non-standardized 'standards' too.)

"THE NUMBERS MASON, WHAT DO THEY MEAN?!"
they mean nothing....

According to appledaily, it does support TB3, but thats only according to google translation, so a pinch of salt.

 

[TheLostSwede]

The only thing attractive to me about USB 4 is the PCIe externalization.
IIRC, PCIe encapsulation (like ThunderBolt) is 'optional' in USB 4. I didn't see a direct mention in the article of the feature, so I assume this will be a rather underwhelming USB controller.
Wasn't X570 was supposed to have USB4 as an option? I think *ONE*, maybe 2 ultra-premium workstation boards had it; not a good start for USB 4. (Not to mention the (now-popular) non-sense naming schemes: Wi-Fi 6, USB # Gen# #x#, HDMI and DP also now have non-standardized 'standards' too.)

"THE NUMBERS MASON, WHAT DO THEY MEAN?!"
they mean nothing....

Not enough public information available, hence the lack of information in the article, but Thunderbolt 3 support is meant to be present.
Going to have a meeting with them, but it got delayed until next week, so we'll be posting more details after that.

X570 was not meant to have USB4, as USB4 wasn't ratified when X570 launched, so that would've been impossible.
Some later boards had Thunderbolt 4, but Intel screwed up Thunderbolt 4, as TI couldn't supply an additional part that was required for USB4 support and there was no second source. Hence a lot of Intel notebooks and systems have Thunderbolt 4 support, but can't support USB4...

 

[LabRat 891]

Not enough public information available, hence the lack of information in the article, but Thunderbolt 3 support is meant to be present.
Going to have a meeting with them, but it got delayed until next week, so we'll be posting more details after that.

X570 was not meant to have USB4, as USB4 wasn't ratified when X570 launched, so that would've been impossible.
Some later boards had Thunderbolt 4, but Intel screwed up Thunderbolt 4, as TI couldn't supply an additional part that was required for USB4 support and there was no second source. Hence a lot of Intel notebooks and systems have Thunderbolt 4 support, but can't support USB4...

Got it.
Thank You for the info. Wasn't aware it hadn't been ratified yet.

 

[TheLostSwede]

Got it.
Thank You for the info. Wasn't aware it hadn't been ratified yet.

Oh, it was in 2019, but in September and X570 launched in July, which makes it a bit hard to have a USB4 chipset ready.
In fact, outside of Intel's TB4 chipset, that is said to support USB4 as well, yet I haven't seen a single implementation that does, the ASM4242 is the first USB4 host controller.

 

[Tek-Check]

Some later boards had Thunderbolt 4, but Intel screwed up Thunderbolt 4, as TI couldn't supply an additional part that was required for USB4 support and there was no second source. Hence a lot of Intel notebooks and systems have Thunderbolt 4 support, but can't support USB4...

This does not make sense at all.
TB4 ports ARE compliant with USB4 specification. If you have TB4 port, you have USB4 port. TB4 is simply more featured than USB4.
Minimum requirements for USB4 port are: 20 Gbps speed, USB 10 Gbps, one display and 7.5W of power. This is far less than TB4 requirements.

In fact, outside of Intel's TB4 chipset, that is said to support USB4 as well, yet I haven't seen a single implementation that does, the ASM4242 is the first USB4 host controller.

TB4 chip does not need separate "USB4 host controller" to function as USB4 port. TB4 chip contains USB 10 Gbps, which is a minimum spec for USB4 ports on their own.

 

[TheLostSwede]

This does not make sense at all.
TB4 ports ARE compliant with USB4 specification. If you have TB4 port, you have USB4 port. TB4 is simply more featured than USB4.
Minimum requirements for USB4 port are: 20 Gbps speed, USB 10 Gbps, one display and 7.5W of power. This is far less than TB4 requirements.
View attachment 248880


TB4 chip does not need separate "USB4 host controller" to function as USB4 port. TB4 chip contains USB 10 Gbps, which is a minimum spec for USB4 ports on their own.

You clearly read something I didn't write.
Please see below.

Temporary stop for Thunderbolt 4 and USB 4 on Intel’s Tiger Lake? What the new chip shortage really means | Exclusive | igor'sLAB

Sometimes it's the really small things that fail and even such a small, quite unknown cog in the gearbox can stop a whole machine in its tracks. The chip shortage is not a new topic, but now the whole…

www.igorslab.de

 

[Tek-Check]

You clearly read something I didn't write.
Please see below.

Temporary stop for Thunderbolt 4 and USB 4 on Intel’s Tiger Lake? What the new chip shortage really means | Exclusive | igor'sLAB

Sometimes it's the really small things that fail and even such a small, quite unknown cog in the gearbox can stop a whole machine in its tracks. The chip shortage is not a new topic, but now the whole…

www.igorslab.de

Yes, this is true. Read about it a few months ago. It's a practical difficulty and stop gap solution for several EOMs, as they mentioned above.
It's terrible for consumers who cannot easily check which TI chip is inside.
It'd be good to see which models were affected and whether anyone created a list.

 

[TheLostSwede]

Yes, this is true. Read about it a few months ago. It's a practical difficulty and stop gap solution for several EOMs, as they mentioned above.
It's terrible for consumers who cannot easily check which TI chip is inside.
It'd be good to see which models were affected and whether anyone created a list.

So in other words, we are on agreement that there was a snafu on Intel's side, causing most early Thunderbolt 4 devices to lack USB4 support?
You're obviously correct about how the controllers from Intel work, but I never mentioned anything about the controller lacking USB4 support.

 

[tpu7887]

I should be seeing them next week, so I'll ask for you.

Lol I don't work there! I was just wondering what AsRock's issue was with an ASMedia product, possibly to avoid it lol.

What did you have written before? I was just busy this past week

 

[Prima.Vera]

USB4 Gen 3x2

Are they still doing this nonsense?????? WTF!?? Somebody should fire that dude(s) making up this shit. Seriously...

 

[TheLostSwede]

Are they still doing this nonsense?????? WTF!?? Somebody should fire that dude(s) making up this shit. Seriously...

You're going to have a hissy fit if you click on this link...
There's no less than four different USB4 variations, of which only have marketing names. They even use different data encoding...

USB4 - Wikipedia

en.wikipedia.org

Lol I don't work there! I was just wondering what AsRock's issue was with an ASMedia product, possibly to avoid it lol.

What did you have written before? I was just busy this past week

I mean, I will be seeing ASMedia, which got pushed to next week, so I'll ask them about your question.

I wonder if they will support USB 3.2 Gen2x2 (20Gbps) on the 4242, because it's optional to.

 

[TheLostSwede]

edit: I wonder if they will support USB 3.2 Gen2x2 (20Gbps) on the 4242, because it's optional to.

I can confirm that yes, they will. More details in the piece I'm working on about USB4.

 

[Tek-Check]

So in other words, we are on agreement that there was a snafu on Intel's side, causing most early Thunderbolt 4 devices to lack USB4 support?
You're obviously correct about how the controllers from Intel work, but I never mentioned anything about the controller lacking USB4 support.

Yes, we are alligned. My impression is that laptops with TB4 were more affected than rare desktop motherboards with TB4. Out of curiosity, I will check my ProArt B550 MB with TB4/USB4. That TI chip is not usually installed in desktop MBs.
Has there been any follow-up on TI chip supply?

I can confirm that yes, they will. More details in the piece I'm working on about USB4.

Out of curiosity, since Zen 4 CPUs support PCIe 5.0 on all lanes from the processor, why is it that chip vendors produce chipset for AM5 and USB4 chips with PCIe 4.0 only? Did they not have time to develop PCIe 5.0 solution with wider bandwidth? Or perhaps we should expect such enhancements in the next gen of peripheral controllers?
I'd imagine a successor of 4242 chip to have 5.0 x4 interace and TB5 the same.

 

[TheLostSwede]

Yes, we are alligned. My impression is that laptops with TB4 were more affected than rare desktop motherboards with TB4. Out of curiosity, I will check my ProArt B550 MB with TB4/USB4. That TI chip is not usually installed in desktop MBs.
Has there been any follow-up on TI chip supply?

Having spent some time talking to VLI and ASMedia, it seems like the situation is far more complex than it should be. So we were both partially correct here and both partially wrong.
The issue in reality, is that until Intel certifies a different PD chip, or TI fixes their supply problem (which they might not), Intel can't certify its Thunderbolt 4 controllers as USB4, but they are technically compatible. However, as Intel hasn't certified them for USB4, they can't market or sell them as USB4 and it's possible that some edge cases, for example a USB 3.2 2x2 device, will end up only working at 10 Gbps speeds rather than 20 Gbps, as Thunderbolt only supports 10 Gbps USB 3.2 Gen 2. Then again, so does USB4, but at least in ASMedia's case, the ASM4242 will support 20 Gbps speeds natively. What a mess...

Out of curiosity, since Zen 4 CPUs support PCIe 5.0 on all lanes from the processor, why is it that chip vendors produce chipset for AM5 and USB4 chips with PCIe 4.0 only? Did they not have time to develop PCIe 5.0 solution with wider bandwidth? Or perhaps we should expect such enhancements in the next gen of peripheral controllers?
I'd imagine a successor of 4242 chip to have 5.0 x4 interace and TB5 the same.

Time, complexity, heads up, cost, lots of factors. Keep in mind that PCIe 5.0 is a lot more complex to design chips and PCBs for. Also, Intel doesn't even have a PCIe 4.0 Thunderbolt controller as yet. PCIe 5.0 will take time, but ASMedia is working on products that supports PCIe 5.0, but none of them will arrive in the market until next year.
Keep in mind that ASMedia is a comparatively small company with only about 250 employees.

I'm working on a longer piece about USB4, that should be up next week, but it's a lot more complicated than you'd think, of which some is politics, even if it's between different companies.