bwong@ihwpt.ATT.COM (bruce wong) (05/19/88)
Do general purpose commercial multiprocessor systems *without* shared-memory exist ? I know there are "general purpose" shared-memory machines like: recent VAXen recent Crays Sequents I know there are "application specific" non-shared-memory machines like: hypercubes the Connection Machine Transputer arrays Are there any general purpose non-shared-memory machines out there ? I heard that DEC has a version of its OS that hides the multiprocessor nature of its new machines; that is, you can treat them like bigger uniprocessor machines. Are there non-shared-memory machines with such an OS ? Please EMAIL all replys. I will post a summary at a future time. -- Bruce F. Wong 1A-111 312-416-5111 ATT Bell Laboratories ihnp4!ihwpt!bwong 200 Park Plaza Naperville, Ill 60566-7050
bwong@ihwpt.ATT.COM (bruce wong) (05/19/88)
In article <2552@ihwpt.ATT.COM> bwong@ihwpt.ATT.COM (bruce wong) writes: > >Do general purpose commercial multiprocessor systems *without* shared-memory >exist ? From one response I got to yesterday's posting I realize that I should clarify my request. By general purpose multiprocessor system I mean that the machine looks like a uniprocessor machine to the user at the shell level and at the application programming level. The user is not concerned at all about processor allocation and application partitioning. This probably means that parallelism occurs at the process level. I read in EE Times that DEC has a version of their OS that makes their multiprocessor shared-memory VAXen look like a uniprocessor machine. Please tell me if this is incorrect. NOTE: I'm interested in similar machines *without* shared-memory. Please repy via EMAIL. I will post a summary after sufficient verbiage has been received. -- Bruce F. Wong 1A-111 312-416-5111 ATT Bell Laboratories ihnp4!ihwpt!bwong 200 Park Plaza Naperville, Ill 60566-7050
rwa@auvax.UUCP (Ross Alexander) (05/20/88)
Bruce Wong @ somewhere.att asks about non-shared memory general purpose multiprocessor timesharing-type machines (see subject line). There is such a beast - the AT&T 3B4000. Runs Unix Sys V3.x, too; but shared memory stuff is not supported between processes running on disjoint processors. So this machine's Unix is _not_ strictly SVID compliant. I wonder what Mr. Gwynn @ smoke.brl has to say about that ? I personally was not teribly impressed. As an aside, it seems to me that the hardware would run a 4.2BSD kernal without any rude violations of the BSD interprocess primitives ;-) Ross Alexander, Athabasca University
res@ihlpe.ATT.COM (Rich Strebendt, AT&T-DSG @ Indian Hill West) (05/21/88)
In article <646@auvax.UUCP>, rwa@auvax.UUCP (Ross Alexander) writes: > Bruce Wong @ somewhere.att asks about non-shared memory general purpose > multiprocessor timesharing-type machines (see subject line). There > is such a beast - the AT&T 3B4000. Runs Unix Sys V3.x, too; The current release of the 3B4000 runs SVR3.1.1. > but shared memory stuff is not supported between processes running on > disjoint processors. So this machine's Unix is _not_ strictly SVID > compliant. To pick a nit and correct a misimpression: Shared memory is identified in the SVID as a machine-dependent feature "and may not be present on all systems." So the 3B4000 does adhere to the letter of the SVID. Shared memory, incidently, is available for processes on the SAME processor, but is not available for processes running on different processors. Rich Strebendt [iwsl6|ihlpe|ihaxa]!res [cuuxf|cuuxg]!iw1res!res