brucec@shark.UUCP (The Traveller in Black @ The Old Phoenix) (05/28/85)
--------------------------------- [this is the first line of the rest of this message] I've seen several articles in the last week or two on floating point benchmarks. Up til now, people have been telling me that no one really needs floating point, and that there's no reason to spend effort on improving performance or functionality. The recent interest in benchmarks makes me think that may be wrong, so I'm posting this questionnaire. Please reply by mail to the address in the signature block, and I'll summarize in this newsgroup. 1) Are you planning on using floating point computation in your software? 2) Do you expect performance of floating point operations to be critical to the success of your program? 3) If you use a compiler, does it use the SANE library which Apple supplies? 4) If you use assembly language, do you use SANE? 5) Would floating point hardware be an attractive alternative, assuming it were available? 6) If fp hardware were available, would compatible software be an important consideration? What level of compatibility is important (source code, requiring different versions of the program, or object, detecting the hardware at run-time)? 7) Would elementary functions be useful? Which ones? 8) Please include comments, especially details of what you want to do with your program if they are not proprietary. If nothing else, it should be interesting to see if there is a real desire for floating point, or if people are just into benchmarking their compilers. Bruce Cohen UUCP: ...!tektronix!shark!brucec CSNET: shark!brucec@tektronix ARPA: shark!brucec.tektronix@rand-relay USMail: M/S 61-277 Tektronix, Inc. P.O. Box 1000 Wilsonville, OR 97070 Bell: (503) 685-2597