gregk@cbnewsm.att.com (gregory.p.kochanski) (01/31/91)
Is there a real time unix which might allow me to read or write to a device in 50 microseconds or less? To be more specific, I do experimental physics, and am basically poking around, looking at atoms on surfaces with a tunneling microscope. This is an experiment where the software and hardware need to be closely coupled. Basically, you read a word from a A-D converter, do a few computations, and output a word or two to a D-A converter. The computations are not stylized enough or simple enough to convenientlyy put in a device driver (and I'm not that masochistic, anyway). Are there unixes where this is possible? Perhaps allowing a user program direct access to some hardware, and locking processes in the CPU?
rstevens@noao.edu (Rich Stevens) (01/31/91)
Just yesterday I received a flyer in the mail for a new book "Real-Time UNIX Systems: Design and Application Guide". It's published by Kluwer Academic Publishers (617-871-6600) and written by 6 people at Modular Computer systems. The table of contents looks interesting, but I haven't seen the book. Unfortunately it's expensive ($59.95), which to me is kind of high for a 352-page book by unknown authors. Rich Stevens (rstevens@noao.edu)
wwm@wa8tzg.mi.org (Bill Meahan) (02/01/91)
In article <1991Jan30.235536.13416@cbnewsm.att.com> gregk@cbnewsm.att.com (gregory.p.kochanski) writes: >Is there a real time unix which might allow me to read or write to a device >in 50 microseconds or less? To be more specific, I do experimental physics, >and am basically poking around, looking at atoms on surfaces with a tunneling >microscope. This is an experiment where the software and hardware need to >be closely coupled. Basically, you read a word from a A-D converter, >do a few computations, and output a word or two to a D-A converter. > >The computations are not stylized enough or simple enough to convenientlyy >put in a device driver (and I'm not that masochistic, anyway). >Are there unixes where this is possible? Perhaps allowing a user >program direct access to some hardware, and locking processes in the CPU? Hewlett-Packard's flavor of UN*X, HP-UX, does indeed have many real-time extensions. The 50 microsecond requirement, however, is difficult for even dedicated RTOS systems (e.g. MTOS, VRTX, etc.) to handle without the appropriate hardware. Check with your local HP rep. There is supposedly a "real-time coprocessor" board available for the 800 (RISC) boxes. There is also a VME extension kit for the 300 (680[34]0) series. "I have no connection with HP other than being a satisfied (well, MOSTLY satisfied) customer." -- Bill Meahan (WA8TZG) | Programming is simple: wwm@wa8tzg.mi.org OR | uunet!mailrus!sharkey!wa8tzg!wwm | All you have to do is put the right "Home for Cybernetic Orphans" | numbers in the right memory locations!
m14167@mwvm.mitre.org (Larry Lawrence) (03/14/91)
gregk@cbnewsm.att.com (Gregory Kochanski) inquired on 30 Jan about the existence of real time unix that can read/write in 50 micro sec. According to an article (Comm. ACM Dec. 1990, vol 33, no 12, "Acquisition at Mission Control" by Muratore et al), there is a product called (surprise!) "Real-Time Unix" (RTU)(TM) by Concurrent Corporation. The authors don't state how "real time" it is, but describe their use of it for acquiring weather data during shuttle landings. Let me know if this is useful. If anyone else has experience with RTU or equivalent products I would like to know how it works. Call at 703 883-6661, or email. I am at MITRE Corp., McLean VA. * * Larry
dave@bigguy.ocpt.ccur.com (David F. Carlson) (03/21/91)
In article <26266@adm.brl.mil>, m14167@mwvm.mitre.org (Larry Lawrence) writes: > gregk@cbnewsm.att.com (Gregory Kochanski) inquired on 30 Jan about > the existence of real time unix that can read/write in 50 micro sec. > > According to an article (Comm. ACM Dec. 1990, vol 33, no 12, "Acquisition > at Mission Control" by Muratore et al), there is a product called (surprise!) > "Real-Time Unix" (RTU)(TM) by Concurrent Corporation. The authors don't > state how "real time" it is, but describe their use of it for acquiring > weather data during shuttle landings. > > Let me know if this is useful. If anyone else has experience with RTU > or equivalent products I would like to know how it works. Call > at 703 883-6661, or email. I am at MITRE Corp., McLean VA. * (Larry I tried to call but...) Concurrent's RTU (Real Time UNIX) is Real UNIX. It supports full SVID and is SVVS complient. It allows realtime priorities and schedulers in addition to normal timesharing. It is fully symmetric multiprocessing and it allows processors to be dedicated to a particular task or set of tasks. UNIX system calls are modified to be pre-emptible so that a high priority process can run as soon as possible. To the programmer, its primary difference to regular ol' UNIX is that it has system calls to dedicate resources (CPU, memory, I/O) to the user process. Thus, when a realtime task *has* to run, it can secure the resources necessary to complete its task. RTU currently runs on Motorola 68020/030/040 and MIPS R3000 in configuations of 1, 2, 4 and 8 CPUs. Tightly integrated multi-threaded device drivers for vector accellerators, disk I/O and data acquisition are available. -- David F. Carlson, Concurrent Computer Co. dave@bigguy.ocpt.ccur.com Fairport, NY "The faster I go, the behinder I get." --Lewis Carroll