|
Re: Gibson Magic - A/D conversion - and fun with hacking Magic: msg#00175music.equipment.midi-guitar
--- In midiguitar@xxxxxxxxxxxxxxx, Tom May <tom@xxxx> wrote: > WRT the discussion below: > > Are you sure Gibson chose UDP simply because it has a litle less > overhead than TCP? My guess would be that they chose it because it is > stateless, and simple at both the application layer and the IP layer. > You just send data and forget about it. No timeouts. No retries. No > blocking writes. No broken connections that need to be reestablished. > No complicated TCP stack. If they really can't squeeze 24/96 data > down a wireless link with TCP (and they can't squeeze it down an > 11Mbps link, which never run that fast anyway, so already you're > looking at one of the faster systems), then the 20 bytes/packet > savings that UDP gives is not going to help much. Compare that to a > single six-channel 24-bit sample occupying 18 bytes. You don't want > to run that close to the edge. > > But bandwidth, security, and encrpytion issues aside, a wireless > signal of any kind can be interfered with. My wireless network can be > jammed by a cordless phone. Not exactly hijacking, but bad enough. > Absolutely. Both reasons. Even if overhead was no issue, the tendency of TCP to resend packets of something happens in transit could potentially lead to all sorts of havok if things go wrong. You hear double notes, things start getting delayed, etc. With UDP you just experience a quick (and maybe unnoticable) dropout. The thing about the 20 bytes is that with all of this data, you end up sending a LOT of packets, and the overhead can start to add up. Also, if things go wrong, with TCP the receiving end has to clean up the connections which can take a while. SImilarly when you UDP flood a port when the UDP flood stops, the port is immeditately free (CPU and memory issues not withstanding). With TCP, the receiver has to shut down all of those errant open connections. The result is that the receiving port can stay "blocked" for a period of time (from ms to many seconds). So, UDP was chosen for BOTH of these reasons, and I think it was a smart choice. I just think their wireless vision (until technology improves, and it probably will by the time this technology starts to permeate the market place - if it does) needs some rethinking! BTW, I was mistaken in that you can actually put **5** repeaters on ethernet, not the 2 that I mentioned. All the more reason NOT to go wireless. Still, if it's available, you know some people will go wireless anyway... :-) Alex F/Brain21 --to unsubscribe send a blank message to midiguitar-unsubscribe@xxxxxxxxxxxxxxxxx Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ |
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| Previous by Date: | Need help with my Casio PG-300!: 00175, dim_xl <dimXL@xxxxxxx> |
|---|---|
| Next by Date: | modelling mic-preamps and Gibson MAGIC (was: fun with hacking Magic): 00175, Alex F <alexf@xxxxxxx> |
| Previous by Thread: | modelling mic-preamps and Gibson MAGIC (was: fun with hacking Magic)i: 00175, Alex F <alexf@xxxxxxx> |
| Next by Thread: | Re: Gibson Magic - A/D conversion - and fun with hacking Magic: 00175, drtimllewellyn <tim.llewellyn@xxxxxxxxxxxxxxxx> |
| Indexes: | [Date] [Thread] [Top] [All Lists] |
| News | FAQ | advertise |