mcdaniel@uiucdcsb.CS.UIUC.EDU (01/22/86)
Thanks in advance. (As advanced as possible, in fact.) I'm going to evaluate the possibility of "parallelizing C": adding language features for parallel computers and/or cobbling together an automatic parallelizer. I need numeric application programs coded in C (matrix manipulations, as one example). Also, I need programs which manipulate various data structures: queues, lists, graphs, trees, et cetera. If you would indicate how often your program(s) are run and how much importance they have in relation to your site's workload, I'd appreciate that too. Since I will primarily be looking at the code, it doesn't have to be adapted to any particular system (working code is greatly preferred, though!). If you have a choice, however, I'd prefer code for BSD 4.2. ------------------ Tim McDaniel; CSRD at the Silicon Prairie (Center for Supercomputing Research and Development at the University of Illinois at Urbana-Champaign) Internet: mcdaniel@uicsrd.csrd.uiuc.edu (Or try: mcdaniel%uicsrd.csrd@uiuc.edu) Arpa, for old mailers: mcdaniel@Uiuc.arpa Usenet: ...{pur-ee|ihnp4|convex}!uiucdcs!mcdaniel Bitnet: MCDANIEL AT UIUCVMD
larry@jpl-vlsi.arpa (01/24/86)
Before you do your own "parallelizing C" you may want to contact Flexible Computer Corp. in Dallas, Texas. They sell a computer family (the Flex/32) which contains both tightly coupled (shared memory) and loosely coupled (message based) 32-bit microprocessors. For parallel processing they provide a parallel-C, parallel-ForTran, and Ada. For actual code samples a good contact point might be the Cosmic Cube people at CalTech (Pasadena, Calif) who did a fair amount of C programming for their hyper-cube "mini-supercomputers." Larry @ jpl-vlsi
brooks@lll-crg.ARpA (Eugene D. Brooks III) (01/26/86)
In article <1892@brl-tgr.ARPA> larry@jpl-vlsi.arpa writes: >For actual code samples a good contact point might be the Cosmic Cube >people at CalTech (Pasadena, Calif) who did a fair amount of C programming >for their hyper-cube "mini-supercomputers." These codes are good examples of what you can do with a message passing system, without any modifications to the serial languages themselves. They are not however, good examples of what you can do or would want to do with the shared memory systems now appearing on the market.