home *** CD-ROM | disk | FTP | other *** search
- Path: darkstar.prodigy.com!davidsen
- From: davidsen@tmr.com (bill davidsen)
- Newsgroups: comp.os.linux.networking,comp.dcom.modems
- Subject: Re: ISDN for Linux
- Date: 18 Jan 1996 23:06:04 GMT
- Organization: TMR Associates, Schenectady NY
- Message-ID: <4dmjos$n0s@usenety1.news.prodigy.com>
- References: <4betj8$qq9@nntp.Stanford.EDU> <30F93B47.2DF7@ubet.com> <slrn4fj488.f93.schmadde@schmadde.isdn.fu-berlin.de> <4de2k4$9r@grebble.oau.org>
- NNTP-Posting-Host: darkstar.prodigy.com
- Originator: davidsen@darkstar.prodigy.com
-
- In article <4de2k4$9r@grebble.oau.org>,
- Eric Wampner <eww@grebble.oau.org> wrote:
-
- | Well, I agree, getting the device to go faster than 115K is not useful
- | if you are using a serial port which can't go faster. My ZyXEL can do
- | 400K, still looking for a serial port. :-) (Of course, the ZyXEL has
- | a parallel interface, currently useless)
-
- You might find that a parallel port buys you nothing...
-
- Unless you have a buffered parallel port you will have to grab avery
- byte with the CPU before the next one comes in, somewhat like using
- a 16450 UART. And even with buffering the speed only rises to the
- limit of how much CPU you will spend moving bytes.
-
- Ideally a fast serial port would be on a card which delivered the
- data to memory via DMA or bus master access, used the full bus width
- of (at least) 32 bits, and could be programmed to generate an IRQ
- N byte times after the first byte was received.
-
- There's no problem getting data in via serial at high speed,
- Ethernet (10mbit), Token Ring (16mbit) and FDDI (100mbit) are all
- serial. The idea of using 8 bits at a time instead of one sounds
- great, but the truth is that the bottleneck is not from the modem to
- the UART, but between the UART and the memory.
- --
- -bill davidsen (davidsen@tmr.com)
- "Anarchy does not scale well." -Dave Welch
-