Can lightning strike twice? Intel has another crack at Thunderbolt It was a bolt from the blue that failed to electrify the computer market. But despite Thunderbolt's very limited success, Intel is building a new version of the high-speed data-transfer interface - and it's persuaded at least two manufacturers to start using the new tech. Code-named Falcon Ridge, Thunderbolt 2 will run at …
Sigh...
Gb/s (Gbit/s) != GB/s
http://en.wikipedia.org/wiki/Data_rate_units#Gigabit_per_second
A gigabit per second (Gbit/s, or Gb/s) is a unit of data transfer rate equal to:
1,000 megabits per second or
1,000,000 kilobits per second or
1,000,000,000 bits per second or
125,000,000 bytes per second.
http://en.wikipedia.org/wiki/Data_rate_units#Gigabyte_per_second
A gigabyte per second (GB/s) is a unit of data transfer rate equal to:
8,000,000,000 bits per second, or
1,000,000,000 bytes per second, or
1,000,000 kilobytes per second, or
1,000 megabytes per second, or
8 gigabits per second.
This is one of those cases where a technology is popular in Content industries and not massively relevant apart from that. Thunderbolt allows high volumes of data that USB3 can't. It is useful for portable HDDs and Video and Audio gear that needs extremely fast access speeds and wide bandwidths. Very specific use cases.
Also, someone mentioned that the cables are expensive. This I because they have active components in the cable. Basically the transmitter is in the cable.
Given how successful this (hasn't) been, I'd have to clarify if the above statement should have "models of" inserted near the end, or if it's accurate as-is?
In any case it's not USB3 that they have to worry about, it's Dock Port (aka AMD Lightning Bolt). That can do a lot more (including power the PC and transmit data to the monitor) and doesn't need the stupidly expensive cabling that Thunderbolt 1 (and presumably 2 as well) does.
I think part of the problem is that USB is everywhere and cheap to implement. Thunderbolt isn't everywhere and is relatively expensive in terms of licensing for manufacturers, more than USB anyway. Most people don't need the bandwidth that Thunderbolt offers so USB 3 is perfectly adequate.
IIRC it was the relatively high licensing costs of IEEE 1394 (Firewire) that helped its downfall, despite having higher bandwidth than USB at the time.
The thing that interested me vaguely about Thunderbolt was the potential for longer distance video connections using optical Thunderbolt cables. It would have been useful when I was in the outdoor events industry and often needing to rig up dirty 100m+ links between projectors and laptops, often looking at solutions like cat5 baluns or wireless HDMI and never getting them working to any sort of satisfactory level. We never had the budgets for pro audio/video transmission gear for things like that but Thunderbolt could have been great. I don't think they ever actually formulated the optical spec though.
A couple of other hiccoughs in the way - Thunderbolt as a standard has to incorporate a video signal, so PCIe add-on cards can't work without weird pass-thru cables. USB3 doesn't have that problem, and also has the bonus of 15 years worth of historical peripherals it can leverage.
Unless Intel have relaxed the standard recently?
I see a very simple solution to that, actually. Just put the thunderbolt port on the video card. It shouldn't add much to the cost, if the card has displayport anyway - most of the electronics are the same. And video cards are usually sixteen-lane - plenty of capacity. I see no reason high-end video cards can't come with thunderbolt capability.
A simple but very expensive solution. So to get Thunderbolt you'd have to replace your (potentially expensive) video card (and I think motherboard has to be compatible too). Compared with a twenty quid USB3 PCIe card.
Plus nVidia or AMD haven't created any such cards yet, which strikes me that it's not possible, or depends too heaviliy on the motherboard to be worthwhile creating.
The original name of Thunderbolt was Lightpeak and it was supposed to use fibre optics to deliver up to 100GBps speeds over 10 meter lengths of cable. It was fast enough that virtually every other IO bus could travel over the top of it conjuring up all kinds of mad possibilities.
Instead they delivered a copper based solution which offered 10GBps but required error correcting active 1m cables. It was gimped and expensive and wasn't hugely better than USB3 which is at least backwards compatible. It's not surprising it didn't take the computing world by storm.
This post has been deleted by its author
My thought too. For now, USB3 is good enough for almost everything, and eSATA good enough for the exception.
But maybe Intel are looking a long way ahead. When memristor technology arrives (dirt-cheap randomly addressible nonvolatile RAM - OK I'm an optimist), how are you going to connect your 1Tb memory stick in the year 2020? Hint: USB3 will offer only a small fraction of the attainable speed.
Also think about why USB2 beat Firewire (which was faster).
> '...allow petabytes of data to be backed up in "minutes, not hours".'
Oh, dear, who fired up the marketing hyperbole machine? Petabytes, only if you've got a *shedload* of interfaces. And presumably are backing up to the bit bucket.
By coincidence I've just done the sums for backing up 125TB in 8 hours; a tad over 32Gb/s sustained. So, assuming I could sustain 16Gb/s and I'm backing up 2 petabytes (their plural, so it's got to be at least 2!), play fast 'n loose with decimal/binary PBs to keep the sums simple and that's what, 32 ports give or take a couple?
'Chipzilla claimed that Thunderbolt (2) would allow petabytes of data to be backed up in "minutes, not hours".'
Um, what? Even if you're charitable and assume no other sources of latency, at 20gb/sec, you can only transfer 72 terabytes of data per hour. Backing up the minimum petabytes to merit the plural will take more than a day, right? 4K video isn't trivial to compress, so that can't be what they're thinking.
That PCIe part means you can easily leverage it for NICs, docking stations and displays. It's strengths are in it's ability to connect a wide range of devices (including USB by the way), as well to daisy chain devices. Intel has been careful to position both technologies, at it wants to keep the huge USB base happy, and is more than content to have it's cake selling USB PHYs and eat the cake to by selling Thunderbolt into a new space. The engineers on the lightpeak team might have had other ambitions until the channel started grumbling about their investments in USB.
Note that there are also still way more bog standard USB2 devices shipping than all of USB3, Firewire and Thunderbolt combined. Firewire looks like it is on it's way out, and Intel gets its wish of another monopoly market (Go back and read old articles here on the Reg to relive that old fight, IMHO Intel beat Firewire the same way as AMD, with bribery and leverage). Intel will rake in the benefits of having USB2 and 3 and Thunderbolt all on the same devices.
On the price front, Thunderbolt cables aren't any worse than HDMI or DVI cables were, and due to the simpler design they should drop rapidly with uptake. OEMs also seem to like the fact they can push out the HDMI licensing to a 3rd party (as the DisplayPort interface is apparently cheaper)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
