26 Jul
2015
26 Jul
'15
1:30 p.m.
On 7/26/15 12:15 PM, Chuck Guzis wrote:
On 07/26/2015 06:12 AM, tony duell wrote:
The first USB devices were utterly terrible--I've got a few that work
only with certain controllers--modern controllers are absolutely blind
to the things.
It seems to me (and I'll defer to more experienced hands) that USB is
not wonderful for single- or low-count byte transfers; that the
negotiation overhead makes short transfers rather problematic,
particularly where the topology involves multiple hubs. In other
words, USB is well-suited to block transfers.
Is this a fair assessment?
And the proliferation of unofficial VID and PIDs seems to be an issue,
particularly with Chinese-origin devices. "Squatting" seems to be a
practice also: VID 0001 = Fry's Electronics; 0004 = Nebraska
Furniture Mart--really? I've found Chinese USB devices squatting on
the Linksys VID, with a non-Linksys PID.
Really, it's a mess.
Saw the same thing with ethernet controllers that were "clones" of DEC's
2114x chips. Some of them had PCI vendor and device codes that
indicated they were DEC but weren't. They also didn't behave properly
in all cases.
The whole VID & PID works because of "gentleman's agreements" (ie use
your own VID and don't pretend to be someone else's). It's not clear
how to solve this for folks that don't follow the rules.
TTFN - Guy
Remember when USB was referred to as the Useless Serial Bus after
it was introduced? I think it was a solid 1-2 years after it was
introduced that I began to notice peripherals designed for it.
I still call it 'Useless Serial Botch' most of the time. It's not a
bus, after all.