Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!nic.csu.net!csun!kithrup!sef From: sef@kithrup.COM (Sean Eric Fagan) Newsgroups: comp.arch Subject: Re: Offloading I/O [was:Incremental sync()s ...] Message-ID: <1991Mar18.194229.21422@kithrup.COM> Date: 18 Mar 91 19:42:29 GMT References: <1088@spim.mips.COM> <18MAR91.09033243@uc780.umd.edu> Organization: Kithrup Enterprises, Ltd. Lines: 24 In article <18MAR91.09033243@uc780.umd.edu> cs450a03@uc780.umd.edu writes: >line-orientation is an approximation of what you want, but not exactly. This kind of thing is done all the time for pc's running unix. The ones I have most experience with are the anvil/stallion boards. There, they literally download code onto the board, which has its own processor, to handle the line discipline. Then, the "real" line discipline and device driver in the kernel cooperate to do things correctly. Basicly, everything gets buffered, until one of the "special" characters comes acros (intr, swtch, or quit; the others are handled on board [no pun intended]). Using a smart board, I've seen 20MHz '386s running with a couple of telebits and 5 users. Not bad. However, there is a problem with this. Specifically, sco added job control to their version of unix, but the onboard card doesn't recognize it (yet). As a result, if I want to use job control, I cannot use the neato nifty-keen device driver (since all it does is echo the control-Z back at me, which, on a wyse-60, clears the screen). -- Sean Eric Fagan | "I made the universe, but please don't blame me for it; sef@kithrup.COM | I had a bellyache at the time." -----------------+ -- The Turtle (Stephen King, _It_) Any opinions expressed are my own, and generally unpopular with others.