kirchner@uklirb.UUCP (Reinhard Kirchner) (04/24/89)
From article <38971@bbn.COM>, by slackey@bbn.com (Stan Lackey): > > I have a real problem with anything that includes IEEE floating point > AND calls itself a RISC. IEEE FP violates every rule of RISC; it has > features that compilers will never use (rounding modes), features that ^^^^^^^^^^^^^^ These things are NEEDED to generate accurate, verified results, not just numbers you may trust or not ( remember: specific combinations of algorithm + data may generate random numbers !! ) Therefor there are compilers which use these modes, and they are called Pascal-SC and Fortran-SC and are made by mathematitians. These language extensions actually would need not rounding modes, but instructions with rounding control since they change the 'mode' from operation to operation. Take a look at interval arithmetic to see examples. R. Kirchner
shankar@hpclscu.HP.COM (Shankar Unni) (04/28/89)
> No apologies are due to the compiler writers. Rather, criticism is due to > them for the arrogance they took in leaving out the possibility for the > programmmer to do something intelligent. The HLLs are woefully inadequate, > and I would not be surprised that the ability to do intelligent coding using > the machine capabilities may be destroyed by learning such restrictive coding > procedures first. Well, excuuuuse me! (sarcasm on) Herman, hopefully no one ever forced you to do your coding in HLL's. You sound to me like someone whose code I'd *hate* to have to maintain. If you dislike HLL's so much, why the *(*&@#) don't you go back and start coding in assembler? Oh, you want to *port* your application to another machine? Naaah, no one would possibly want to waste their precious time doing that ****ty porting stuff. Anyway, if a machine doesn't have MOVC3 and CALLG, it must be horse####. (sarcasm off) There's a reason for HLL's not supporting your favorite all-singing, all-dancing instruction at the source level, and that is that not every machine has that instruction in that form. Got it???? ---- Shankar.
gillies@p.cs.uiuc.edu (04/28/89)
From article <38971@bbn.COM>, by slackey@bbn.com (Stan Lackey): > > I have a real problem with anything that includes IEEE floating point > AND calls itself a RISC. IEEE FP violates every rule of RISC; it has ^^^^^^^^^^ > features that compilers will never use (rounding modes), features that Just 6 months ago we were arguing about "What is the definition of RISC?" Now (apprently) it is so cut-and-dried there are rules. Just what are "the rules" of RISC? 8-)
wendyt@pyrps5 (Wendy Thrash) (04/29/89)
In article <650011@hpclscu.HP.COM> shankar@hpclscu.HP.COM (Shankar Unni) writes: <some suggestions for Herman Rubin> I have this theory that people who actually use computers to do useful things can tell language designers, compiler writers, and system architects a great deal about what they actually need. If we listen to them, we may advance the state of the art. Herman has been quite clear in telling us some of what he wants/needs. He has not spared our egos in the process. We may not always agree with what he says, but can we not find inspiration in his ideas? When an intelligent person tells me that HLLs are inadequate for what s/he is trying to do, I suspect that it's time to investigate, not time to flame. Of course, Herman's suggestion that >> the ability to do intelligent coding using the machine capabilities may >> be destroyed by learning such restrictive coding procedures first seems fairly risible, but who knows: I seem to recall a fairly respectable computer scientist* saying something similar about coding in BASIC and its effect on coding in modern HLLs, and compiler writers didn't dump on him. :-) *(This is certainly not meant to imply that Herman is not a respectable computer scientist; skill with statistics, which his net address leads me to ascribe to him, does not preclude knowledge of other disciplines.)