wichert@quando.UUCP (Peter Wichert) (09/29/88)
Hi folks ! We're using SCO Xenix 2.2 with parallel line printers . There are two things we don't understand: 1. The problem which occured is that the printer would not work as fast as possible . The printer is connected to the parallel port of the Hercules-Card. The spooler seems to work correctly. First we thought, that the terminal processes accessing the graphics card would slow down the supplement of the printer. But connecting the printer to a serial/parallel interface card had the same effect; also 'cat file > /dev/lp[0,1]' . The manual does not give us any hints. What's wrong ?? 2. What is the standard number of NBUF and SABUF ? Are there optimal values differing from the default values generated by the kernel during startup ? AND: what are SABUFs ?? Please reply directly to : (If anyone is also interested in the answers, let us know. We'll forward the useful ones .) ________________________________________ | Usenet: wichert@quando.uucp | | UUCP : ..!unido!quando!wichert| ________________________________________ Thanks, _ __ _ ' ) ) _/_ ' ) / / _/_ /--' _ / _ __ / / / o _. /_ _ __ / / </_<__</_/ (_ (_(_/ <_(__/ /_</_/ (_<__
jfh@rpp386.Dallas.TX.US (The Beach Bum) (10/01/88)
In article <644@quando.UUCP> wichert@quando.UUCP (Peter Wichert) writes: > > >Hi folks ! > >We're using SCO Xenix 2.2 with parallel line printers . >There are two things we don't understand: > >1. The problem which occured is that the printer would not work as fast > as possible . there are update kits available for slow line printers. the problem has several causes. i suggest you get the updates for parallel printers from SCO. then, if that doesn't fix it, there is a polling parallel driver out there someplace. use it instead. >2. What is the standard number of NBUF and SABUF ? NBUF should be about 10% of your physical memory. for a 2MB machine, i have NBUF == 302 and SABUF = 30. i don't have to have any SABUFs but configure put them in anyhow. > AND: what are SABUFs ?? system addressable buffers. these buffers are addressable from the kernels data segment and don't need to have the MMU changed to be addressed. the 386 can address all of physical memory, whereas the 286 can't. so, CPUs which can't address their entire physical address space need to have some SABUFs. i've seen all manner of guidelines, but in general i suggest having as many SABUFs as your kernel data segment will hold. they are much faster than far buffers. -- John F. Haugh II (jfh@rpp386.Dallas.TX.US) HASA, "S" Division "Why waste negative entropy on comments, when you could use the same entropy to create bugs instead?" -- Steve Elias
tif@cpe.UUCP (10/04/88)
Written 6:04 pm Sep 30, 1988 by rpp386.UUCP!jfh in cpe:comp.unix.xenix >In article <644@quando.UUCP> wichert@quando.UUCP (Peter Wichert) writes: >>1. The problem which occured is that the printer would not work as fast >> as possible . >SCO. then, if that doesn't fix it, there is a polling parallel driver >out there someplace. use it instead. This is an investigation still underway but here are two bits fer ya... 1. Polled printing can be accomplished by mknod'ing a device with a minor node of (n + 64) where n is your current (not polled) printer device. This (like most USEFUL information) is in the Release Notes. 2. There is a global variable, pabusycnt, in the kernel which determines (for a polled device) how long to wait for the printer to become un-busy. Default is 64 I think. This information was in one of the DiSCOver magazines. Note that both of these steps are likely to decrease system performance. This information is only a summary of otherwise available information. Further disclaimers apply and are available upon request :-) Paul Chamberlain Computer Product Engineering, Tandy Corp. {convex,killer}!ninja!cpe!tif