jordan@aerospace.aero.org (Larry M. Jordan) (01/16/91)
I am considering creating benchmarks that will exercise processor features that endeavor to "efficiently support the execution of Ada." I admit that I am quite ignorant in this area. So, I pose the question: "What does a processor (RISC,CISC) need in order to efficiently support the execution of Ada?" Another way to pose the question is, "What would be the Ada compiler writer's dream target architecture?" One avenue I thought I would pursue is to acquire the source code for an Ada run-time system and examine how such services are implemented. (Apart from relying upon services of the RTS, is there anything about the code generated for an Ada application that might not be basically similar to that generated by a C compiler?) Then, knowing what I know about the intended target, reimplement the RTS (or abstraction of it) in the target assembly language. Does anyone know where I can get my hands on an Ada RTS that is in the public domain (and/or is of minimal cost)? HOL or assembly are fine. It doesn't have to be the best implementation thereof out there, but one that serves to demonstrate what services are provided and how they are implemented. --Larry
BARRYTE@CASPER.dnet.hac.com (01/18/91)
Once upon a time didn't Intel believe the i432 to be an Ada machine? I believe there were articles in NCC & elsewhere. I vaguely recall that Western Digital was working on enhancing their Pascal engine into an Ada engine... & they, too, waxed poetic.
carters@ajpo.sei.cmu.edu (Scott Carter) (01/18/91)
In article <1991Jan16.190417.1@elcgl.epfl.ch> madmats@elcgl.epfl.ch writes: >> Another way to pose the question is, "What would be the Ada compiler >> writer's dream target architecture?" >> --Larry > >You may find an answer to that question by looking at Rational's architecture, >which was made specially for the development of Ada programs. You may find >Rational's address in a previous post containing the list of all validated Ada >compilers. Note that an optimal [however you want to define optimiality :} ] architecture for _development_ is by no means necessarily an optimal environment for target _execution_. <this is not at all a denigration of the [very impressive, IMHO] Rational environment, but I don't think anyone has suggested putting in on the ATF> In the embedded target, speed and memory optimality are generally the driving factors, and the development environment is more of a maintenance environment, where the tradeoffs are different. > >Another interesting RTS is BiiN's, which is entirely written in Ada. I don't >know what happened to BiiN, since I haven't heard of it for a year or so. BiiN is defunct. I hear various stories about what happened to the compiler (chip production, marketing, etc. have reverted to Intel), but it doesn't seem to be on the market (960 compilers are available from Tartan and Irvine Compiler Corp.). We had a couple of low-end BiiN workstations (Model 6160, as I recall - att any rate only a single self-checking pair CPU unit), and I don't remember anything particularly remarkable about the Ada environment. The performance was pretty lously, but since our applications were largely compute-bound I imagine that it was simply due to the general slowness of the 960 rather than anything about remarkable either way about the RTS. >Mats Weber >Swiss Federal Institute of Technology >EPFL DI LGL >1015 Lausanne >Switzerland > >E-mail : madmats@elcgl.epfl.ch >phone : +41 21 693 52 92 >fax : +41 21 693 39 09
braun@CTC.CONTEL.COM (Chris Braun x4475) (02/01/91)
Someone asked about whether Intel has at one time considered the i432 to be an Ada machine? The answer is yes. An article on this may be found in the June 1981 (yes, I mean 1981) issue of IEEE Computer. It is titled "Ada for the Intel 432 Microcomputer" and is by Stephen Ziegler, Nicole Allegre, Robert Johnson, James Morris, and Gregory Burns. Also, regarding acquiring source and documentation for an Ada runtime system: It should be possible to obtain this material for any Ada compilers developed under government contract. These include ALS compilers done by SofTech and by CDC under contract to the Army and the Navy, and compilers done by Intermetrics under contract to the Air Force. You need to contact an organization called NTIS (National Technical Information Service, or some such) for further information. (Or perhaps SofTech or Intermetrics could provide information.) Chris Braun braun@ctc.contel.com