News Intel Shows First Thunderbolt 80 Gbps Demo

[Admin]

Intel showed the first images of a running system with a 80 Gbps Thunderbolt.

Intel Shows First Thunderbolt 80 Gbps Demo : Read more

 

[InvalidError]

When you have to do multi-cable lane bonding of ludicrously expensive interfaces and cables to reach a target data rate over copper wiring, that looks like a sign that high-speed interfaces are long overdue for an optical upgrade.

 

[-Fran-]

When you have to do multi-cable lane bonding of ludicrously expensive interfaces and cables to reach a target data rate over copper wiring, that looks like a sign that high-speed interfaces are long overdue for an optical upgrade.

Or you just need a new cable with more lanes in it xD

Regards.

 

[InvalidError]

BTW, in the spirit and tradition of horrible names by various standards organizations lately, I propose that doubling TB4's bandwidth by simply doubling the lane count be called TB4.1-gen1x2, which leaves the door open for nudging the bandwidth up some more by increasing the per-lane bandwidth from 20Gbps to 32-40Gbps further down the lane and call that TB4.2-gen2x2.

 

[peachpuff]

BTW, in the spirit and tradition of horrible names by various standards organizations lately, I propose that doubling TB4's bandwidth by simply doubling the lane count be called TB4.1-gen1x2, which leaves the door open for nudging the bandwidth up some more by increasing the per-lane bandwidth from 20Gbps to 32-40Gbps further down the lane and call that TB4.2-gen2x2.

I prefer the simple usb 4.1 4.2 4.3 and so forth naming scheme.

 

[Kamen Rider Blade]

BTW, in the spirit and tradition of horrible names by various standards organizations lately, I propose that doubling TB4's bandwidth by simply doubling the lane count be called TB4.1-gen1x2, which leaves the door open for nudging the bandwidth up some more by increasing the per-lane bandwidth from 20Gbps to 32-40Gbps further down the lane and call that TB4.2-gen2x2.

NOOOOOOOOOOOOOOOO!!!!!!!!!!!!!!

Don't give into the bad naming scheme that breaks historical naming precedence madness.

It should be called "Thunderbolt 5"

 

[InvalidError]

NOOOOOOOOOOOOOOOO!!!!!!!!!!!!!!

Don't give into the bad naming scheme that breaks historical naming precedence madness.

I failed: I forgot to throw in the extra arbitrary letter to differentiate a numerical re-branding of an existing standard (ex.: HDMI 2.0 vs 2.1) from an updated implementation of the standard (2.1a) that supports additional features

 

[edzieba]

When you have to do multi-cable lane bonding of ludicrously expensive interfaces and cables to reach a target data rate over copper wiring, that looks like a sign that high-speed interfaces are long overdue for an optical upgrade.

Good news then: they're not using cable-bonding.

 

[Kamen Rider Blade]

I failed: I forgot to throw in the extra arbitrary letter to differentiate a numerical re-branding of an existing standard (ex.: HDMI 2.0 vs 2.1) from an updated implementation of the standard (2.1a) that supports additional features

=:::::::::::::::::
Don't add in HDMI's non-sense on top!

 

[MattD-Kauai]

Does this underwhelm anyone else? I mean what is the actual point of Thunderbolt anymore. The USB 4 2.0 is already going to be the same speed and be available to anyone without having to kiss the Intel posterior for rights.

I would think most motherboard MFGs would just implement the USB config across all brands of CPUs. So is Intel just wasting R&D money to stay relevant in this interface?

 

[InvalidError]

The USB 4 2.0 is already going to be the same speed and be available to anyone without having to kiss the Intel posterior for rights.

Thunderbolt has been license-free for a few years already and some AMD X570 motherboards have TB3.

The USB 4 2.0 is already going to be the same speed and be available to anyone without having to kiss the Intel posterior for rights.

While USB may be good enough for most everyday use, Thunderbolt is basically external PCIe with much lower overhead than USB, which is kind of important for high-performance stuff like external GPUs.

IMO, Thunderbolt is what we should have had with type-C instead of USB3.

 

[ezst036]

Or you just need a new cable with more lanes in it xD

Regards.

 

[Nikolay Mihaylov]

While USB may be good enough for most everyday use, Thunderbolt is basically external PCIe with much lower overhead than USB, which is kind of important for high-performance stuff like external GPUs.

IMO, Thunderbolt is what we should have had with type-C instead of USB3.

Err, no. TB is not external PCIe. It's a generic transport protocol that can encapsulate/tunnel PCIe traffic among others. It actually has higher bandwidth than PCIe3 x4.

Also, PCIe tunneling is possible with USB4 and is even required for USB4 hubs. At this point it seems USB4 is TB4 with some of its features made optional and other made less strict (e.g bandidth, power delivery) That's probably to allow for cheaper implementations.

In any case, the only area where PCIe tunneling is indispensable is for external GPUs. Native USB4 transfer modes should not slow down NVMe devices in a way that can be felt by most consumers.

 

[MattD-Kauai]

Err, no. TB is not external PCIe. It's a generic transport protocol that can encapsulate/tunnel PCIe traffic among others. It actually has higher bandwidth than PCIe3 x4.

Also, PCIe tunneling is possible with USB4 and is even required for USB4 hubs. At this point it seems USB4 is TB4 with some of its features made optional and other made less strict (e.g bandidth, power delivery) That's probably to allow for cheaper implementations.

In any case, the only area where PCIe tunneling is indispensable is for external GPUs. Native USB4 transfer modes should not slow down NVMe devices in a way that can be felt by most consumers.

So basically, Thunderbolt 80gb is USB4 2.0 fully loaded? Which means that Intel's announcement was basically to save face and stay in the spotlight. Sounds about typical for Intel.

 

[Nikolay Mihaylov]

So basically, Thunderbolt 80gb is USB4 2.0 fully loaded? Which means that Intel's announcement was basically to save face and stay in the spotlight. Sounds about typical for Intel.

I think so, yes. However, let's not be too critical of Intel in this context. They essentially donated the Thunderbolt specification to the USB-IF for use in USB4. That's why they are almost indistinguishible. And that's a good thing.

 

[artk2219]

I think so, yes. However, let's not be too critical of Intel in this context. They essentially donated the Thunderbolt specification to the USB-IF for use in USB4. That's why they are almost indistinguishible. And that's a good thing.

You make it sound like they did it out of the goodness of their hearts , it was purely a move to stay relevant and with a hand in whatever standards will be set moving forward. This is why they basically quit charging for licensing on thunderbolt 3, USB was catching up and they wouldn't be able to dictate the standard since very few companies wanted to pay for it, even though at this point it was "only" like $5. Thunderbolt used to be very expensive to implement, its first version added an extra $25 to every bill of materials, it was non negotiable, and not that many took them up on the offer, this is compared to USB 3 that at the time was something like $1 (or less) to implement, way way wayyyyyy less.