home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.realtime
- Path: sparky!uunet!stanford.edu!nntp.Stanford.EDU!kellyj
- From: kellyj@leland.Stanford.EDU (Kelly Johnson)
- Subject: Re: novel idea?
- Message-ID: <1993Jan12.020252.23939@leland.Stanford.EDU>
- Sender: news@leland.Stanford.EDU (Mr News)
- Organization: DSG, Stanford University, CA 94305, USA
- References: <erempel.726720401@sol.UVic.CA> <1993Jan11.065941.21342@leland.Stanford.EDU> <1993Jan11.212259.16530@noao.edu>
- Date: Tue, 12 Jan 93 02:02:52 GMT
- Lines: 55
-
- In article <1993Jan11.212259.16530@noao.edu> cole@noao.edu (Lonnie Cole) writes:
- >In article <1993Jan11.065941.21342@leland.Stanford.EDU> kellyj@leland.Stanford.EDU (Kelly Johnson) writes:
- >>In article <erempel.726720401@sol.UVic.CA> erempel@sol.UVic.CA (Evan Rempel) writes:
- >>>I haven't read very much in the realtime environment, but I have
- >>>a thought about an OS that has what I think is a fairly unique
- >>>idea.
- >>>
- >>>I am looking for an OS (real time obviously) that has the
- >>>ability to swap a task from one cpu to another. This is merely
- >>>a multiprocessing system. The catch is that I want to be
- >>>able to mix cpu's. Say a 68040 as the general, but then a i860
- >>>running on the side for complex graphics. If the i860 becomes
- >>>bogged down (I don't know what I would be doing to bog down the
- >>>i860, but lets say I do) and the 68040 is sitting idle (or close
- >>>to it) can one of the tasks that is running on the i860 be switched
- >>>over to the 68040 to run there.
- >>>
- >>>I know that the single task on the 040 would run slower than on the
- >>>i860, but if the i860 was bogged down, the net result would
- >>>be a system speedup.
- >>>
- >>>Is there any system out there that provides this type of
- >>>functionality?
- >>>
- >>
- >>The problem with the idea is that the executable code for the 68040
- >>and i860 are totally different. It is not uncommon to have a
- >>homogeneous multi-processor system that swaps tasks between cpu's
- >>for load balancing, but doing so between totally different types of
- >>cpu's would not be a simple task of just swapping tasks.
- >>I fail to see how this could ever be done effectively.
- >>Anyone else agree/disagree?
- >>
- >
- >Why does each processor have to execute the same binary code. Suppose
- >that each processor had its own binary copy of the code to execute and
- >some mechanism to share the environment (local data, stack information
- >etc.) then the process is not so far fetched.
- >
- >I don't think it's such a silly idea, kind of interesting.
- >
-
- I agree that it is interesting, but it sure has some significant problems.
- I took the original post literally when I should not. I figured he was
- looking to do the same type of thing you would do when transfering context's
- between homogeneous machines (ie. just transfer simple state information like
- the PC, data space/code space addresses, etc). In a heterogenous system
- like that you would (like many have said) need separate executables on each
- system. This could easily introduce problems with data object sizes, alignment
- restrictions, etc. Also, what do you do in the case of instructions that have
- "faulted" (like due to a page swapped out). Theoretically it sounds interesting
- and MAY have merit, but pragmatically it seems a bit out there, but who knows
- it may be closer than I think. I don't like to limit things more than the
- laws of physics allow and even those can "appear" to be broken at times.
-
-