vixie@decwrl.dec.com (Paul Vixie) (11/07/88)
# Anybody with DSP experience know if that TI DSP chip in the T1000 # is up to the task of implementing v.32? I don't have any DSP experience, but at the SLIP BOF at Interop '88, Mike Ballard from Telebit said that the echo cancellation in V.32 was a very hard problem to solve digitally and that it looked like it was going to take two 32-bit DSP chips vs. the one 16-bit DSP chip they use today. In addition to a 68020 or '030 instead of the '010 they use now. It sounds like it's going to be expensive compared to a V.32-only modem made with the new (Rockwell?) V.32 modem chip set. But it'll do more, of course -- stuff like SLIP framing and UUCP/Kermit/Xmodem spoofing and of course we'll expect it to do PEP (at 28KB/sec!) and whatever USR HST's use. Whether all that extra stuff will make the Telebit V.32 entry desirable remains to be seen. Personally, I'm not holding my breath for V.32. The TB+ does everything I want it to do; full-duplex at a lower speed solves no problem I happen to have. -- Paul Vixie Work: vixie@decwrl.dec.com decwrl!vixie +1 415 853 6600 Play: paul@vixie.sf.ca.us vixie!paul +1 415 864 7013
james@bigtex.cactus.org (James Van Artsdalen) (11/07/88)
In <23@jove.dec.com>, vixie@decwrl.dec.com (Paul Vixie) wrote: | [...], but at the SLIP BOF at Interop '88, Mike Ballard from Telebit | said that the echo cancellation in V.32 was a very hard problem to | solve digitally and that it looked like it was going to take two | 32-bit DSP chips vs. the one 16-bit DSP chip they use today. In | addition to a 68020 or '030 instead of the '010 they use now. A recent hire at my company used to work in the DSP group at Motorola and claims that the NeXT DSP is indeed capable of V.32 with minimal help. oakhill!bryant, Motorola DSP manager, made the same claim in comp.sys.next. Future Moto DSPs are supposedly much more powerful yet. Since the DSP is a real uProcessor in its own right, it's not difficult to see the DSP handling both the signal *and* the data link to the host computer (at least when no compression is involved), eliminating the need for a 680x0 class chip altogether. -- James R. Van Artsdalen james@bigtex.cactus.org "Live Free or Die" Home: 512-346-2444 Work: 338-8789 9505 Arboretum Blvd Austin TX 78759
wtm@neoucom.UUCP (Bill Mayhew) (11/09/88)
I wonder if the DSP used in the Trailblazer modems could be pushed hard enough to manage v.32. Several people who seemed to know emailed me that the Motorola 56K DSP should be able to process a 9600 bps v.32 link. The Moto chip is quite a bit more potent than the TI DSP chip in the Trailblazer, I believe. One compromise might be to offer a spoofing mode in the Trailblazer that implements v.32 up the 4800 bps fallback speed. When negotiating the connection, the Trailblazer could be set up to ask for a fallback even if the line is good. This, of course, presumes that 4800 is easier to process than 9600 (which may or may not be true). While not the ultimate pizza, such a mode would allow the rest of us souls with TBs and TB+s to at least hold a coffee clatch with the modems that a certain party at portal says we should be interested in talking to. --Bill