klm@munsell.UUCP (11/19/87)
In article <7730@g.ms.uky.edu> sean@ms.uky.edu (Sean Casey) writes: > >Where I work I do programming on a Masscomp 5500 running Real Time Unix >version 3.0A. RTU supports real time applications, and does it quite well, >thus showing it can be done. All one really has to do is give the user >system calls to lock memory and a priority system that allows a process >to take over the machine. All I can say is that Masscomp must have made some really good improvements to their Real Time Unix. I haven't touched a Masscomp machine in over 3 years, but when I did, it was for high speed data acquisition. RTU's native facilities for doing Real Time stuff were completely inadequate. In order to have the machine keep up with the data bus we ended up not only having to dork with the system quite a bit, but we also had to write our own disk driver that would do raw burst DMA without a filesystem. I currently know of another situation with some friends of mine who are trying to do high speed data collection with a Sun 3 and they're going through the same nightmare. Strip down the Kernel, no display, no file system, raw disk i/o at 10 tracks a whack, and they're barely cutting it. In other words, yes, Masscomp, VAX/VMS, <whatever> will do realtime stuff, but you have to dork with it so much that you end up wondering why you ever bothered. Why not start off with a 'real' realtime system and don't expect it to do very much more than that. The Amiga definitely needs memory protection to keep naughty programs from being anti-social, but if you put VM on it, it's gonna be SLOOOOOOOOW! The Sun/3 can cut it with VM because it's CPU goes about 3 times faster than the Amiga. -- Kevin McBride, the guy in the brace // | It's the end of the world Eikonix - A Kodak Co. // | as we know it, Billerica, MA \\ // Amiga | {encore,adelie}!munsell!klm \X/ Rules! | And I feel fine...
sean@ukma.UUCP (11/20/87)
In article <1390@atkins.munsell.UUCP> klm@munsell.UUCP (Kevin [Being Weird Isn't Enough] McBride) writes: >All I can say is that Masscomp must have made some really good improvements >to their Real Time Unix. I haven't touched a Masscomp machine in over 3 >years... > >In other words, yes, Masscomp, VAX/VMS, <whatever> will do realtime stuff, >but you have to dork with it so much that you end up wondering why you >ever bothered. Why not start off with a 'real' realtime system and don't >expect it to do very much more than that. RTU has changed a == WHOLE LOT == in 3 years. Version 4.0 meets the SysV interface definition and still has the fastest real time response of any other real time Unix by a factor of three. PLEASE DO NOT tell the world your "facts" about something you haven't touched in three years. Three years is an awfully long time in the computing hardware world. Sean -- -- Sean Casey sean@ms.uky.edu, {rutgers,uunet,cbosgd}!ukma!sean -- (the Empire guy) sean@ms.uky.csnet, sean@UKMA.BITNET -- University of Kentucky in Lexington Kentucky, USA -- "Inconceivable!"
sims@stsci.EDU (Jim Sims) (11/25/87)
In article <1390@atkins.munsell.UUCP>, klm@munsell.UUCP (Kevin [Being Weird Isn't Enough] McBride) writes: <> In article <7730@g.ms.uky.edu> sean@ms.uky.edu (Sean Casey) writes: <> > <> >Where I work I do programming on a Masscomp 5500 running Real Time Unix <> >version 3.0A. RTU supports real time applications, and does it quite well, <> >thus showing it can be done. All one really has to do is give the user <> >system calls to lock memory and a priority system that allows a process <> >to take over the machine. Agreed. <> <> All I can say is that Masscomp must have made some really good improvements <> to their Real Time Unix. I haven't touched a Masscomp machine in over 3 <> I currently know of another situation with some friends of mine who are <> trying to do high speed data collection with a Sun 3 and they're going Let's be specific, what is "high-speed"? I've done ~56Kbyte data acquisition and display and file creation on VAX 750 (~.7 MIP) under full VMS. No Problem. Is that fast enough? <> through the same nightmare. Strip down the Kernel, no display, no file <> system, raw disk i/o at 10 tracks a whack, and they're barely cutting it. <> <> In other words, yes, Masscomp, VAX/VMS, <whatever> will do realtime stuff, <> but you have to dork with it so much that you end up wondering why you <> ever bothered. Why not start off with a 'real' realtime system and don't <> expect it to do very much more than that. <> Because some of us LIKE a REAL development environment. <> <> The Amiga definitely needs memory protection to keep naughty programs from <> being anti-social, but if you put VM on it, it's gonna be SLOOOOOOOOW! <> The Sun/3 can cut it with VM because it's CPU goes about 3 times faster <> than the Amiga. So, lets build an Amiga with a 68030, like I originally suggested!!! VM != SLOW VM only provides a good means for running a program bigger than the memory you have physically. There are several ways around paging problems, there is not anything inherent in the concept of VM that precludes real-time. -- Jim Sims Space Telescope Science Institute Baltimore, Md. 21218 sims@stsci.edu