eugene@orville.nas.nasa.gov (Eugene Miya) (10/12/88)
I was sent mail and asked to summarize my thoughts on this workshop. I think it was generally pretty good for a first workshop. In the past, I attended the 1st and 2nd Usenix Graphics Workshops, both had attendence less than 100 (the usual maximum set by Usenix). This conference had 183 attendees. We discovered problems of the physical layout of the hotel within the first hours, but the attendees took it in stride. Contents appear below (excepting work in progress reports (The Fujitsu UTS on their VP line was perhaps most significant as they had planned to announce next year, a "scoup!"). The general atmosphere was not as free flowing as most Usenix meetings, in fact, there were times (people asked themselves if they were attending a CUG [Cray User Group] meeting. There were no Hypercube papers (possibly some in the future, few Crayette papers, and I thought the technical content to be a bit light (comments returned to Lori and Melinda indicated most thought it was just right). There were a fair number of people attending the /usr/group/supercomputing POSIX standards meeting [following the workshop, I was asked to attend, but could not due to prior commitments]. I think future meetings will have a better idea of how to organize and have fun, share software, etc. There is clearly big money in supercomputers, and the attendence surprised all in a year when there are too many supercomputer meetings. --eugene miya %A Jan Edler %A Jan Lipkis %A Edith Schonberg %Z NYU Ultracomputer Project %T Process Management for High Parallel UNIX Systems %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 1-17 %K MIMD, RP3, fork, spawn, scheduling group ID, meta-system call, temporary non-preemption, IPC, signal, %A Dennis Ritchie %Z AT&T Bell Labs %T A Guest Facilities fpr Unicos %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 19-24 %K virtual machines, exguest(XP), testing, simulator, Cray-XMP, %A H. Stehpen Anderson %Z OSCP, OSC %T Distributed Supercomputer Graphics Using UNIX Tools %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 25-32 %K distributed visualization, network, animation, apE, chimp, remote shells, filters, Convex, Cray X-MP, %A Jonathan Brown %Z NMFECC LLNL %T The CTSS/POSIX Project %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 33 %K Cray, %X Abstract only. %A Robert M. Panoff %Z Physics, Clemson U. %T Real Producitivity for Real Science without Real Unix %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %X Abstract only. "increased productivity of current supercomputer users appears doubtful..." poor compilers, inadequate debugging, process proliferation as problems. %A C. A. Stewart %T Numerical Applications Interprocess Communication Protocol: RPCODE: RPC server to solve ODEs %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 37-42 %K Cray, SUN XDR, %X Interesting JSR quote RPC1057. %A Kenneth Bobey %Z Myrias %T Monitoring Program Performance on Large Parallel Systems %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 43-49 %K profiling, parallel programmer's workbench (PPWB), %A E. N. Miya %T Some Observations on Computer Performance Characterization: Supercomputer and Mini-Supercomputer Clocks and Compilers %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 51-65 %K optimization, Cray, Convex, Cydra 5, amdahl, alliant, IBM, %A John Renwick %Z Cray %T High-speed Network with Supercomputers %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 67 %K Ultrabus, HSX, HSC, %X Abstract only. %A Ray Bryant %Z IBM TJW %T The RP3 Parallel Computing Environment %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 69-92 %K Mach, AIX Transparent computing facility (Locus), ROMP, EPEX/CMS, tasks threads (TTM), seamless system image, %X Project survey, no new information, emphasis on batch rather than interactive execution. %A Brewster U. Kahle %A William A. Nesheim %A Marshall Isman %Z Thinking Machines %T UNIX and the Connection Machine Operating System %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 93-107 %K CMOS, CM file systems (CMFS), virtual processor, %X Good paper %A Y. Langue %A T. Muntean %Z U. Grenoble %T PARX: A Unix-like Operating System for Transputer-based Parallel Supercomputers %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 109-120 %K Occam, Minix, supernode, MIMD, SIMD, SPMD, %A Martin Fouts %T Multitasking under Unicos: Experiences with the Cray 2 %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 121-131 %K microtasking, Cray, %X Elliptical Gauss-Seidel block diagonal solver. %A Ralph Knag %Z AT&T Bell Labs %T Unicos Fair Share Scheduler %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 133 %X Abstract only. %A Kevin Wohlever %T UNICOS System Administration at the Ohio Supercomputer Center Tuning Considerations %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 135-136 %X Extended abstract only. Copies of the paper exist (not bad). %A Patrick Clancy %Z Multiflow %T Virtual Memory Extensions on TRACE/UNIX %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 137-150 %K VLIW, copy on write, block allocation, %A Jan Edler %A Jan Lipkis %A Edith Schonberg %Z NYU Ultracomputer Project %T Memory Management in Symunix II: A Design for Large-Scale Shared Memory Multiprocessors %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 151-168 %K Symunix I (Ultra I and II), cache, paging, %X Several added systems calls. %A E. C. Pariser %Z AT&T Bell Labs %T Reduction of Static and Dynamic Memory Requirements on the Cray X-MP %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 169-182 %K solid-state disk (SSD), %X The memory management is static or dynamic, not the memory technology (a somewhat unfortunate title). A significant I/O performance table. %A Michael John Muuss %A Terry Slattery %A Donald F. Merritt %T BUMP: The BRL/USNA Migration Project %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 183-214 %K Cray, %X What to do on "no space on disk." %A Alan Poston %T A High Performance File System for UNIX %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 215-226 %K HPFS, amdahl UTS, %X Proposal, work in progress. %A Douglas E. Engert %Z Argonne NL %T Attaching IBM Disks Directly to a Cray-XMP %J Proc. Workshop on UNIX and Supercomputers %I Usenix Assoc. %C Pittsburgh, PA %D Sept. 1988 %P 227-229 %K XIOP, Block Multiplexer Controller (BMC), %X Low cost, modest performance, software only changes (no added hardware).