K312240@AEARN.BITNET (Klaus Kusche) (03/03/90)
Dear Mailing List: About the public domain folding editor just announced on this list: Could someone grab it and put it onto the archive server at Cornell or Oxford (please BTOA or UUENCODE the executeables for those who can't get binaries from there)? It is hard to access a US phone bullboard from Europe, academics prefer networks... About PC transputer boards and Unix: You may put such a board into an Apollo. There are even boards especially designed for that purpose (with a very fast DMA link inter= face). I don't know if anyone uses this with the Apollo DOS coprocessor and DOS software (I've heard that there are some problems), but Apollo (and other companies, too) provide Unix software for that combination: Apollo has all the Inmos stuff and some Apollo-specific things (use your T's as coprocessors from Unix processes anywhere in the Apollo net) CSA has LS C, ... (has someone done Trollius?) About Gnu C: I've heard that someone in France has done this. However, I never contacted this person directly, hence I don't want to post his email address here without his consent. I hope he listens and speaks up by himself (if not, could anyone close to him drop him a note?). Parsytec again: Well, a list of software which will *not* run on multi-transputer Parsytec hardware without changes: All Inmos and 3L products, LS C, Express,... About their incompatible hardware standards: I like their differential links (the mechanical and electrical quality of their products is very fine in general, too), but I don't like their Error/Reset/Analyze scheme: It requires that each processor cares about its neighbours, so it adds to the application code on each processor. On the other hand, it is not really superior to what can be done with the Inmos standard in any way: You can create one- processor subsystems also with Inmos. Moreover, Parsytec doesn't provide Analyze, a serious drawback! In contrast to what ENGLE@A.ISI.EDU says, you can't have an Inmos- compatible system controlled by a Parsytec board or vice versa, you can only connect the links successfully, but not Error/Reset/Analyze. It took me more than one month (and several Faxes, dozens of phone calls and letters, and a lot of my nerves, although they are just a few hundred miles away from here and speak the same native language), until they finally confirmed to us (after lots of useless statements) that their PS/2 board can't be configured in a way which allows to control an Inmos-compatible subsystem with functionality equal to the B004 standards. I know that the other transputer supercomputer companies (Meiko, Cogent, Parsys) also use control hardware completely incompatible with the Inmos standards, but * Their hardware provides improved functionality (central system monitoring and control independent of the application network, many debugging features, link switch control, ...). * Their software makes full use of these features (Parsytec's software essentially provides the same functionality as its plain PC equivalents, previous version). Moreover, Parsytec's marketing isn't fair in this respect: The other supercomputer companies always say 'It's our system, it run's our software.'. They never claim compatibility. Parsytec always announces that most of the transputer software around (for PC's or whatever) will run on their boards. They never say that you have to buy their version! Greetings ************************************************************************ * Klaus Kusche * * Research Institute for Symbolic Computation * * Johannes Kepler University Tel: +43 7236 3231 67 * * A-4040 Linz Telex: (Austria) 22323 uni li a * * Austria (Europe) Fax: +43 7236 3231 30 * * * * Bitnet: K312240@AEARN * * Arpa/CS/Internet: K312240%AEARN.BITNET@CUNYVM.CUNY.EDU * * UUCP: mcvax!aearn.bitnet!K312240 * * Janet: k312240@earn.aearn or k312240%aearn@earn-relay * ************************************************************************