Path: utzoo!attcan!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!mailrus!purdue!mentor.cc.purdue.edu!l.cc.purdue.edu!cik
From: cik@l.cc.purdue.edu (Herman Rubin)
Newsgroups: comp.arch
Subject: Re: Architecture questions
Summary: How much, and what would it get?
Message-ID: <2531@l.cc.purdue.edu>
Date: 9 Sep 90 18:49:37 GMT
References: <10057@goofy.Apple.COM> <2516@l.cc.purdue.edu> <6838.26e7f109@vax1.tcd.ie>
Organization: Purdue University Statistics Department
Lines: 56

In article <6838.26e7f109@vax1.tcd.ie>, rwallace@vax1.tcd.ie writes:
> In article <2516@l.cc.purdue.edu>, cik@l.cc.purdue.edu (Herman Rubin) writes:

			........................

> Consider the following kinds of operations:
> 
> 1. Scientific and engineering work (floating-point)
> 
> 2. Text processing (byte addressing)
> 
> 3. Transaction processing (floating-point, bytes and fixed-length integers)
> 
> 4. Graphics (best done in chunks as large as the CPU can handle, using barrel
>    shifters and masking)
> 
> 5. Compiling (integers and pointers)
> 
>  - in no particular order. These make up something over 99% of all computation.
> Hence hardware is designed to make them fast. This means dealing with
> floating-point, integers and bytes fast. To be sure you could probably list 50
> different kinds of algorithms that would benefit from bit addressing but these
> make up less than 1% of the computation that gets done, and the extra hardware
> would slow down other operations and increase costs. Why should 99% of users
> pay in money and time for something that will improve processing speed for 1%?

Every trigonometic and exponential routine starts out with a division, getting
an integer quotient and a remainder.  Ususally, these are heavily kludged,
because of the inadequacy of hardware.  Even the multiplication by the quotient
is kludged.

Text processing, for anyone who needs more than ASCII, is in a very primitive
state.  Even ASCII processing is limited.  Again, hardware instructions would
help.

There are quite a few situations in transaction processing which could use a
few intelligent instructions, like reading from a buffer with an automatic
interrupt when the buffer becomes empty.

People producing graphics software have complained loudly about the lack of
fixed-point arithmetic.

Compilers and languages are weak, and those who live by them seem incapable of
understanding that anything else can and should be done.

What about simulation (Monte Carlo)?  I have methods which I would not suggest
as economical without certain fast bit operations.

The question of cost is relevant.  With the cost of the entire ALU being a
rather small part of the cost of the machine, and also that the units wanted
could have a great deal of overlap with the present units, I would doubt that
a more flexible hardware would cost 1% more.
-- 
Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907
Phone: (317)494-6054
hrubin@l.cc.purdue.edu (Internet, bitnet)	{purdue,pur-ee}!l.cc!cik(UUCP)