Intel adopts 40Gb per SECOND USB-C plug for Thunderbolt 3.0 Intel's sort of given up on the Thunderbolt interface - but it's also found a way to help increase its relevance by making it a superset of USB 3.1. Thunderbolt has always been fast, and has the nifty trick of connecting peripherals, carrying data and bringing video to monitors over a single cable — but it never took off. …
This sounds like a recipe for confusion to me.
A new Thunderbolt device using a USB Type C connector could perhaps be converted back to an old Thunderbolt connector, but what about the backward compatibility to DisplayPort and FireWire?
A new USB device using a USB Type C connector could be converted back to a USB 2.0 or 3.x connector, but not to a Thunderbolt/DisplayPort/FireWire connector.
Would it not make more sense to include the Thunderbolt feature set in USB 3.1, but not call it Thunderbolt at all and avoid any confusion.
@Davidmb the usb-c on the current macbook doesn't support thunderbolt and won't unless a Firmware update of some sort is released.
Its blindingly obvious that Apple knew this was coming though and maybe the next macbook refreshes in Oct will get it across the range.
I think the only intention of the fibre option is for long runs of cable, hence why the article mentions runs of up to 60 metres.
Fibre has some advantages, not least of which is massively increased range compared to copper. However the one problem it always has is latency, because at the end of the day the devices at each end have to convert photons on fibre to electrons on copper, and that's always going to take time not necessarily huge amounts but why bother unless you really need to.
"
Fibre has some advantages, not least of which is massively increased range compared to copper. However the one problem it always has is latency, because at the end of the day the devices at each end have to convert photons on fibre to electrons on copper, and that's always going to take time not necessarily huge amounts but why bother unless you really need to.
"
There is no significant latency converting between electrical signals & light signals. If you think about it you will realise that there cannot be, because neither the photo-transmitter nor the photoreceptor can "store" more than 1 bit, and so they must both convert between light and electrical signals within 0.5 bit-time at most. There is far more than a bit-time of latency involved in the system as a whole whatever the transmission medium, because each byte (or more usually word) that is sent must be serialised and then de-serialised before it will be passed to and hence be processed by the destination CPU.
"Wasn't that the plan originally way back when, Intel were pushing optical fibre cables as a new Ethernet standard but due to the cost dropped the idea, so now they are back trying again?"
Intel's original intent was a cable technology called Light Peak. It had enormous 'gee whiz' appeal at the time, to the point of euphoria. But it wasn't realistic, at least at that time, to combine an optical cable with a wire cable. Optical provided fast data transfer over long lengths. But the wire cable was wanted for device control and transferring power to external devices. Despite the hyper-hype, Light Peak was killed off and a compromised called 'Thunderbolt' was created instead, an entirely wired solution with at least an improvement in speed.
Meanwhile, using optical for fast data transfer has remained useful. This new compromise again integrates the ability to use optical cable.
Here are a couple articles of interest. The first is about Light Peak from 2010:
http://www.anandtech.com/show/3930/intel-light-peak-lighting-up-in-2011
The second article is about the resulting Thunderbolt compromise in 2011:
http://www.anandtech.com/show/4194/intels-codename-lightpeak-launches-as-thunderbolt
Thunderbolt is the new Firewire. It's superior to USB3 in most ways, but it's also insanely expensive - and who cares, when USB is 'good enough?' There is no consumer application that might require more bandwidth than USB3 can provide, so thunderbolt is stuck in the same niche as Firewire once occupied: High-end AV gear and super-fast external drives.
I keep re-testing Thunderbolt bridging for IP networking.
Tested again this weekend using OS X 10.10.3.
Whilst I can get speeds as high as 500MB/s between a MacBook Air and MacBook Pro Retina,
it fluctuates wildly, getting as low as 50Mb/s.
It would be awesome if I could finally replace 10Gb ethernet and 8Gb Fibre for basic storage networks.
Looks like a dream that will never be realised.
"...if I could finally replace 10Gb ethernet and 8Gb Fibre..."
Congratulations, you not only re-invented FCoE but added a display protocol to further overcomplicate the mess. Given the price difference between the FCoE switches and the "vanilla" 10 GbE switches one is left to wonder how pricey a VESA-n-FC-over-Ethernet port would be...
Tried Infiniband? The switches cost a fortune, but you can pick up 8Gb/s interfaces dirt cheap. Connect a cable between them and you've a point-to-point link. It does a decent bitrate, and it can do RDMA too, so if you get it configured right (which is a bit tricky) you can get some impressive performance out of it. The prices do get high if you want more than two computers on your network though.
"Chipzilla's now come to the conclusion that Thunderbolt can't stand alone, so has adopted the USB-C port for the standard, replacing the current proprietary port."
Something I don't understand here, Why is USB-C not considered proprietary port whereas the thunderbolt port was considered proprietary?
I am pretty sure that you have to pay intel to add USB-C capability to any devise, and part of that is a fee for the USB-C port.
I think the real thing going on is that USB has such a market following as a connection protocol, but Thunderbolt didn't have much of a following. Since,
Unlike ChunderBolt (which was a brilliant idea, until Intel crippled it by deciding to gouge people in licensing fees, in order to access the specs detailing the Secret Sauce that they added to garden-variety DisplayPort, and PCI Express); USB's "non-proprietary", or "open", in the sense that anyone can obtain the specs for it for no-charge, without signing an NDA, or otherwise having to register; in order to develop drivers (and probably host interface controllers).
However, if you want to actually manufacture a slave device (i.e. a peripheral), you've got to pay the price of a small car to reserve a non-transferrable Vendor ID, and a block of Product IDs - unless you're doing a small run of something like a USB-to-RS-232 converter, or something else based on throwing a fancy case on a generic reference design (in which case, some vendors will let you share their VID/PIDs, under certain conditions, like not using it with competitive chipsets).
"Thunderbolt has always been fast, and has the nifty trick of connecting peripherals, carrying data and bringing video to monitors over a single cable — but it never took off."
No shit Sherlock. The only mainstream manufacturer using thunderbolt is Apple. They've been putting on their consumer gear since 2011, and only on their, ahem, pro gear since early 2014 (yes, I know the mac pro was released late 2013, but nobody could get hold of one until 2014)
You can pick up a 1TB usb spinning drive for, i dunno, going by the prices here in Swtzerland, for around £50. An equivalent thunderbolt drive would be topping £300, and that would still be spinning rust at it's heart. No consumer is going to go for the thunderbolt alternative. In fact I believe the vast majority of pro users would also baulk at the cost of TB preipherals, with perhaps the exception being RAID storage.
It was, still is, a damned good interface, but ruined by price gouging and licensing costs. I have 3 macs with Thunderbolt interfaces and never once has a thunderbolt device graced the appropriate orifices.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
