v128ll9e@ubvmsd.cc.buffalo.edu (Richard B Wicks) (11/09/90)
I was thinking late at night on how I could make a brand new and really fast transfer protocol. This is (I think) a completely new idea. It occured to me that the only way to really speed up a transfer is the have the receiving computer send NO information back to the sending computer. The way that it would work (for error checking) is have the sending computer would send a CRC check every 1024 bytes. The receiving computer would independently compute the CRC and compare them, if they are different, then and only then would the receiving computer send anything back. My question: would this speed up transfers? It seems to me that it would have to if you programmed to protocol to take advantage of the multitasking ability of the Amiga. The only problem is getting some poeple to adopt the protocol. -Rich
mcmahan@netcom.UUCP (Dave Mc Mahan) (11/09/90)
In a previous article, v128ll9e@ubvmsd.cc.buffalo.edu writes: > > I was thinking late at night on how I could make a brand new and really >fast transfer protocol. This is (I think) a completely new idea. It occured to >me that the only way to really speed up a transfer is the have the receiving >computer send NO information back to the sending computer. Not quite true. To take advantage of the maximum rate of your modem, you have to ensure that the transmitting end is always full and there are no pauses in the outbound character stream. Transmission rate in one direction does not depend on transmission rate in the other direction (for modems that are true full duplex). > The way that it would work (for error checking) is have the sending >computer would send a CRC check every 1024 bytes. The receiving computer would >independently compute the CRC and compare them, if they are different, then and >only then would the receiving computer send anything back. I think this is pretty much the concept behind Z-modem. If you get a fast host and a fast term program that are optimized for Z-modem speed, you will see the back channel light flicker only occasionally and you should never see the outbound data light stop and wait for it. I use JR-Comm for the amiga (nice job on the software, Mr. Radigan!) and a Sun-4 host. I regularly get 235 characters per second coming in (out of a possible 240 cps with a 2400 baud modem) even with heavy usage of the host computer. I have seen this number go up to 239 or 240 cps when little else is taking CPU time or disk accesses from my host. I almost always get 240 cps when I send files to my host. >-Rich -dave -- Dave McMahan mcmahan@netcom.uucp {apple,amdahl,claris}!netcom!mcmahan
ckp@grebyn.com (Checkpoint Technologies) (11/09/90)
In article <45220@eerie.acsu.Buffalo.EDU> v128ll9e@ubvmsd.cc.buffalo.edu writes: > > I was thinking late at night on how I could make a brand new and really >fast transfer protocol. > [description of protocol deleted] You have described ZMODEM fairly accurately. Yes, it's a great idea, it really is a good, fast transfer method. And people are adopting it. I only wish it dealt with errors more quickly. Don't feel bad if you didn't know this. A friend of mine also "invented" ZMODEM himself... -- First comes the logo: C H E C K P O I N T T E C H N O L O G I E S / / \\ / / Then, the disclaimer: All expressed opinions are, indeed, opinions. \ / o Now for the witty part: I'm pink, therefore, I'm spam! \/
joseph@valnet.UUCP (Joseph P. Hillenburg) (11/10/90)
A VERY easy way to make a transfer protocol with max thoroughput would be to have NO error checking, and restrict the protocol to people with MNP, HST, V.32/42, etc. Sort of like supercharged versions of Zmodem/Overdrive or Ymodem-G. Joseph Hillenburg, Secretary, Bloomington Amiga Users Group (BAUG) joseph@valnet.UUCP ...!iuvax!valnet!joseph "Only Apple could slow down a 68000 chip." -Computer Shopper
jprad@dynalogic.UUCP (Jack Radigan) (11/11/90)
In article <45220@eerie.acsu.Buffalo.EDU>, Richard B Wicks writes: > I was thinking late at night on how I could make a brand new and really >fast transfer protocol. This is (I think) a completely new idea. It occured >me that the only way to really speed up a transfer is the have the receiving >computer send NO information back to the sending computer. ZMODEM (full streaming) and YMODEM-g both operate in this fashion. With ZMODEM, it only sends a reposition header to the sender when a data frame is recieved in error. YMODEM-g requires an MNP or similar error-corrected connection since it only checks for errors, it does not correct for them. It also requires full 256 byte transparancy as do all XMODEM technology protocols. > My question: would this speed up transfers? It seems to me that it >would have to if you programmed to protocol to take advantage of the >multitasking ability of the Amiga. The only problem is getting some poeple >to adopt the protocol. YMODEM-g will give you some great throughput figures, 98+% is quite common. ZMODEM, since it is compatible with packet-switched networks, is somewhat slower due to the necessary character escaping it has to do to prevent network problems. >-Rich -jack-
king@motcid.UUCP (Steven King) (11/13/90)
In article <45220@eerie.acsu.Buffalo.EDU> v128ll9e@ubvmsd.cc.buffalo.edu writes: > > I was thinking late at night on how I could make a brand new and really >fast transfer protocol. This is (I think) a completely new idea. It occured to >me that the only way to really speed up a transfer is the have the receiving >computer send NO information back to the sending computer. > The way that it would work (for error checking) is have the sending >computer would send a CRC check every 1024 bytes. The receiving computer would >independently compute the CRC and compare them, if they are different, then and >only then would the receiving computer send anything back. > My question: would this speed up transfers? It seems to me that it >would have to if you programmed to protocol to take advantage of the >multitasking ability of the Amiga. The only problem is getting some poeple >to adopt the protocol. > >-Rich Excuse me, but did somebody mention Zmodem? What you've described is essentially the inner workings of Zmodem. It *is* faster than other protocols that regularly acknowledge the packets. The speedup is especially noticable over packet-switched networks, less noticable for normal dialups. Zmodem has been around since before 1986; that's the date of the revision on the spec sheets I have. (Chuck Forsberg, Omen Technology Inc., "The ZMODEM Asynchronous Inter Application File Transfer Protocol", rev051486 Printed 5-16-86) Multitasking isn't necessary, but it might be handy if you wanted to do simultaneous bidirectional transfers. There's a "Bimodem" protocol for IBM PCs that I think is bidirectional Zmodem. (Don't quote me on that.) The closest the Amiga world ever got was with a term program called Diga that would let you transfer one direction and chat with the other person at the same time. Don't think it did bidirectional transfers, though. (Again, don't quote me.) -- ---------------------------------------------------+--------------------------- Nothing is more conducive to peace of mind than | Steven King not having any opinions at all. | Motorola Cellular (Georg Christoph Lichtenberg [1742-1799]) | ...uunet!motcid!king