> > Neil, sure is funny how many times that dead horse gets resurrected. ;>) >
Sigh... Even *I'VE* heard this debate enough times that I'm sick of it. When you design a product, you make decisions. The reason the 3000 and 5000 use 1394 is that Flex made that decision back when they were designing the radios. Over the EONS we've been having this discussion, Gerald has told us (in detail, repeatedly) that with the information they had at the time, and the requirements they were using, 1394 appeared to be the best choice at that time to the engineers who were responsible for the decision. Search the archives... you'll find all the details. We can -- and have -- debated ad infinitum whether USB 2.0 would work. It just doesn't matter. The decision's been made. It was NOT a bad decision then, and it has not proven to be a bad decision even now... several years later. For the record (the following is pure redundant trivia -- don't read it if you don't want to read the semi-technical ramblings of an engineer): > even if the 3000 and 5000 could run on USB 2.0 it would present the same set > of latency requirement as firewire Absolutely. There's nothing about USB that would lead us to believe that the types of latency issues that we see on 1394 would be any better on USB. The latency issues are really related to the drivers involved, not the underlying bus technology. The Windows USB stack (Host Controller Driver, etc) presents its own stunningly unique set of challenges, just like the Windows 1394 stack. With USB there are other common issues as well, like hubs... that can greatly influence the throughput on a bus. > > I realize you won't get 480 megabit/sec across a USB 2.0 interface, but many > applications routinely get many tens of >megabits/sec across them. > Actually, applications routinely get tens of megaBYTES per second across USB. At least 200Mbps... maybe 300Mbps. > > IEEE 1394 (Firewire) is actually a better protocol than USB. > The protocols are sufficiently different that you have to define "better." Any X can be said to be "better" than any Y, if one defines "better" so that this is true. Firewire is more costly, less interoperable, and much more complex. This would mean that USB is "better" if better is defined based on those factors. > > It does support isochronous streaming transfers, > something that Flex is just now trying to get going on USB. I > USB inherently supports isochronous transfers. I have personally written Windows USB drivers that support isochronous endpoints. It's not difficult. The driver and the device firmware make ALL the difference. At THIS point, neither USB NOR 1394 would be the best choice, IMNSHO. The best choice would be Gigabit Ethernet. Oh, by the way: Even THEN you'll have to be concerned about DPCs and latency... Peter K1PGV _______________________________________________ FlexRadio Systems Mailing List FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/ Knowledge Base: http://kc.flexradio.com/ Homepage: http://www.flexradio.com/
