4 ms·
> Perhaps in case of storage devices (although it would be far less efficient to do it that way). That's in fact how it works for USB bulk transfers: The host
by lxgr 3y ago
> Perhaps in case of storage devices (although it would be far less efficient to do it that way).
That's in fact how it works for USB bulk transfers: The host learns about the checksum error having occurred and it will retry until it passes.
That doesn't seem inefficient to me, given the low error rates involved (many flavors of Ethernet do the same, for example) – are you saying there should be FEC?
> But this approach doesn't apply to e.g. high speed industrial cameras.
These usually use isochronous transfers, where errors are indeed only detected (and reported), but no retransmission is attempted.