Keep in mind, that 1394 has one simple benefit not mentioned here: There are no switches necessary and you can use it as a bus if you have many devices, like in a studio (mixer, mics, effects,...). USB has the disadvantage only to be a point 2 point connection, ethernet needs switches and switches cost time, see AVB and limited number of 7 hops (endpoints & switches)... also regenerating clock from USB devices isn't that easy if not impossible to reach firewire quality in this point. In the technical point of view, firewire is a better solution than ethernet or USB.
As an interesting side note to this article, the entire music score and music production for the movie "Avatar" was done on 6 Pro Tools HD hardware and software systems and some Mac computers. The HD interface was PCI Express. Earlier, non "HD" products used Firewire and USB2.0.
Although the author's company has a financial interest in 1394's continued use in pro audio, his points are still valid. 1394's isochronous capability, peer-to-peer capability, and carrying timing information with the data are huge benefits in a recording studio or in a live sound production environment. USB and Ethernet were historically inadequate to meet the demands of multi-channel pro audio production. That could be overcome with the sheer power of huge bandwidth, but it's difficult to imagine audio engineers abandoning a solution that already works, on the promise that something newer and cheaper has more bandwidth, so don't worry about the protocol limitations.
Having lived through the various Ethernet/token ring/100VG LAN wars, not to mention Internet Protocol vs IPX vs ISO, and other contenders like DECnet and SNA, let me say I am skeptical about how IEEE 1394's specific attributes will ensure its longevity. Every scheme has its strong suits, but what history shows time and again is that the winner almost always find a way around its weaknesses. The winner is usually the simplest and cheapest solution, which overpowers the competition by "throwing bandwidth" at the problem. Which it can afford to do thanks to its lower cost. And beyond that, the winner usually finds a way, in its upgrade cycle, to introduce features that were once the sole province of the fancier schemes.
The first two versions of USB might have required the PC bus master to poll any connected device, but USB 3.0 is supposed to permit any connected device to initiate transfers. So the advantage of 1394 peer to peer communications becomes not so unique after all. The original USB might have been restricted to mouse and keyboard connections, but USB 2.0 can reach 480 Mb/s, and 3.0 has a path to at least 5 Gb/s. So honestly, how difficult can it be to establish multiple audio patyhs over such an interconnect?
The initially more "elegant" solutions are certainly not the ones that prove to be the ultimate winners.
IEEE 1394 is being removed from its last nich market, professional audio. I can understand the problem of Mr Lave and his desire to see IEEE1394 being the super champion of professional audio and have TCTechnologies which bet everything on IEEE1394 succeed. But if there is no doubt IEEE1394 is a better technology than USB from a latency point of view and from P2P transfer it's a doomed technology which will not live the next 5 years.
USB3 is there with all its default but it will conquer the professional audio by its ubiquity and cheapness. Present in all new motherboards and chipset it's free for the consumer and the raw transfer speed will compensate for lack of P2P transfer. CPU+GPU processing power is cheap (AMD Fusion processor for example), the fact handset manufacturer will not deploy usb3.0 does not seem an heavyweight argument.
And even ethernet with realtime capacity and AVB should unify on stage all the different bus used (midi/DMX/analog transport).
IEEE1394 is a superior technology as was token ring, it wont prevent it to disappear soon. I hope TCTechnologies is going to do USB and Ethernet products soon.
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...