dann@bmcg.UUCP (Dann McCreary) (07/30/85)
Here is a collection of responses to my request for sources of C compiler and 80*86 assembler alternatives to Intel's outrageous prices. Thank you all very much for your responses to my inquiry. -Dann McCreary ------------------------------------------------------------------------------ From: sdcsvax!dcdwest!ittvax!decvax!seismo!utah-cs.UTAH-CS!utah-cs!schimpf (James Schimpf) You are not going to like this but.... The assembler and linker on an IBM PC or clone is usable for this purpose. (The '186 has only a few more instructions than the 8088 and you can EQU these.) We are doing development on a '186 engine using Lattice C and the IBM PC assembler. Also for ROM code we are using the 2500AD assembler as the PC Linker will not put code to an absolute memory location. The C is about $500, the assembler I don't know but Big Blue will gladly quote you a price and the 2500AD assembler is about $100. --jim schimpf schimpf@utah-20 ARPA From: sdcsvax!ihnp4!gargoyle!sphinx:sphinx!shor (Melinda Shore,Staff 002,53971,,staf) [] For a C cross-compiler, try Manx (better known as ``Aztec C'') . They have an extensive set of compilers and cross-compilers for bezillions of chips. Their compilers don't always generate the tightest code (though they're cleaning that up), but they're reliable, professional products. There are several sources of cross-assemblers, but I don't know how good they are. Frankly, I think the best source of information about what software is out there and how good it is is the Programmer's Shop. I've bought all my software from them, and have been very pleased with their service. They've got the only software salespeople I've dealt with that can answer questions and give evaluations. They have several special programs, including a pioneer program for trying out software that's new to them. Also, if they recommend a piece of software and you don't like it, you can *return* it. They have a newsletter and several literature packages. Their telephone no. is 800-421-8006. (I have no affiliation with the Programmer's Shop, other than as an utterly satisfied customer ... ) Call them and ask about cross-development tools. Melinda Shore University of Chicago Computation Center From: sdcsvax!ucbvax!nbires!mccallum (Doug McCallum) You might try Intermetrics. They have cross-compilers/assemblers for 8086/80186 as well as a few other processors. The C compiler has a few idiosyncracies, but does have most of the K&R constructs as well as most of the newer things such as passing/returning structures. I don't know if they have a SysV version, but I am using a 4.x bsd version. Probably the strangest of the problems I found is that C expects the called function to pop it's first parameter from the stack on exit and the caller will pop the rest. The code generated isn't too bad. The same compiler will generate 8086 code or 80186. If told it is targetted to a 186, it uses the new instructions. There is a linker and a locator and a final tool to format the module into Intel HEX or Tektronix format, etc. Doug McCallum NBI, Inc. {allegra, ucbvax, ut-sally}!nbires!mccallum From: sdcsvax!dcdwest!ittvax!decvax!seismo!philabs!sbcs!root (SBCS Systems Staff) We used Tannenbaums Compiler Kit (available from UniPress) to produce 80186 code. The Kit consists of a suite of compilers/assemblers for common machines (incl. 808x) -- to add 80186 code generation, I simply hacked added a couple of lines to the code generation table (movb, popa, pusha, etc). By the way, the Kit `links' at the symbolic assembly language level (essentially, the libraries are supplied in source form and assembled every time). Another alternative is to purchase Xenix for your Intel box. Xenix includes a C compiler, assembler and linker for 8086. Given that the instruction differences between 8086 and '186 are not too important, you can run 8086 code on your machine. To get compiled code to proms is a simple matter of writing a conversion program (~100 lines C) to crack a.out format into Intel hex (we have one of you choose this method). From: sdcsvax!decvax!utzoo!aesat!rwh Unfortunately, those prices are most of what there is, particularly for UNIX cross tools. As soon as you mount it on a multi-user station, the prices go to the multi-kilobuck range. Speedware has a hopped-up assembler that runs on the PC that is advertised as 80186 compatible - another department here may be acquiring it (it's only $99). Systems and Software sells cross-development tools for VAX. Call them at (714) 241-8650. They seem to be the only game in town for cross-assemblers. Call Unipress for leads to other tools, including their VAX port of Phoenix's PLINK-86. ______ Russ Herman / \ {allegra,ihnp4,linus,decvax}!utzoo!aesat!rwh @( ? ? )@ (416) 821-9190 ( || ) AES Data Inc. ( \__/ ) 1900 Minnesota Ct. \____/ Mississauga, Ont. CANADA L5N 3C9 From: sdcsvax!decvax!cca!mirror!prism!matt (Matt Landau) We have had fairly good success with the Lattice C compiler running as a cross compiler. It's supported by Lattice, and published by Unipress in New Jersey. We're running it on a Pyramid 90x under 4.2 BSD, but I believe they also publish a Sys V version. It's fairly reliable, except that we've found some bad floating point bugs in the latest release. Source code has been sent to Lattice and they're working on fixes. Lattice is a reasonble compiler, supporting V7 C (no enums, no structure passing or assignment). It generates 8086 or 80186 code, and seems reasonably good at it. Reliable rumor has it that the missing C features will be added in a completely new release this fall. As far as an assembler goes, I wish I could help you there. We're doing our assembly work on AT's and porting the objects to the Pyramid for linking. That's an awful way to work, and we're actively searching for an 80x86 Unix based cross assembler. If you find anything, I'd be interested in hearing about it. For information about the compiler, contact Unipress Software, Inc. at (201) 985-8000 and ask to speak to Vicky Anto. ----------------------------------------------------------------------------- Matt Landau {cca, datacube, ihnp4, inmet, mit-eddie, wjh12}... Mirror Systems, Inc. ...mirror!prism!matt ----------------------------------------------------------------------------- "Spontaneity has its time and place..." DISCLAIMER: My only connection with Unipress is as a usually-but-not-always satisfied customer. From: sdcsvax!dcdwest!ittvax!decvax!decwrl!amdcad!cae780!ubvax!lance (Lance Keigwin) We've had good experience with the Intermetrics cross-compilers running on our 750's (4.2). They sell 8086 family, 68000, and z80's. Have used them for about 2 years. Good support. Few bugs. Relatively efficient code. A little "pricey." Intermetrics, Inc. 733 Concord Avenue Cambridge, MA 02138 (617)661-0072 Our salesman: Paul Kusiak From: scgvaxd!kontron!cramer.ttyh0 (Clayton Cramer) Perhaps you have other requirements which preclude this, but the obvious choice is to use the Microsoft C V3.0 compiler (which can produce 186 and 286 code), and the Microsoft V2.0 Macro Assembler (which doesn't seem to have a specific 186 mode), under PC-DOS. Both of them combined cost about $550, and you should be able to buy an XT or XT-clone and still be cheaper than Intel's software only price. -- / Dann McCreary Burroughs ===/=== Corporation / Advanced Systems Group MS-703 (619) 485-4585 10850 Via Frontera San Diego CA 92128 {decvax || ucbvax || ihnp4 || philabs}!sdcsvax!bmcg!dann
lotto@talcott.UUCP (Jerry Lotto) (08/01/85)
A recent PC Week RUMOR Central quote: "...Staggered by repeated body blows to the semiconductor business, Intel has quietly ported the assemblers and compilers it uses for development work with its microprocessors to the PC to sell as end-user products. Until now, the compilers and assemblers have only been used in conjunction with a $30,000 Intel super-micro system for manufacturers of systems using Intel processors. The PC products, which are said to be more powerful than any on the market, will be available by October for about $700 per package. Spencer expects Intel to release assemblers for its 8087, 80287, 8086, 80286 and 80386. He expects compilers for PLM, possibly C, and FORTRAN." The usual disclaimers apply here, I do not work for Intel, PC Week (or any one else for that matter) and the opinions expressed herein are a direct quote and belong to someone else. -- Gerald Lotto - Harvard Chemistry Dept. UUCP: {seismo,harpo,ihnp4,linus,allegra,ut-sally}!harvard!lhasa!lotto ARPA: lotto@harvard.ARPA CSNET: lotto%harvard@csnet-relay