cs161agc@sdcc10.ucsd.EDU (John Schultz) (12/28/88)
In article <403@nth.UUCP> loyd@nth.UUCP (Loyd Blankenship) writes: [lots of stuff about TDI M2 nuked] > I was using the expensive 'Professional' version of the thing, too! > I finally just bit the bullet, considered it $225 down the tube, and >bought the Lattice C Professional kit. I am so much happier with it I am >like a new man! I don't know anything about Benchmark Modula-2, and wasn't >willing to risk another couple hundred dollars on an unknown product. Lattice >is a fantastic product, they have good customer support, and if you aren't >dead set to work in Modula-2, I'd recommend you go with Lattice. >Loyd Blankenship cs.utexas.edu!nth!loyd (UUCP) We use both Lattice 5.0 and Benchmark. For speed in development time, where runtime code speed/size are not hypercritical, Benchmark still has the nicest, fastest development system. The emacs editor for Benchmark is much nicer than LSE, and it has no bugs. Compile, link and run, all from the editor, at 10,000 - 30,000 lines per minute. The only drawbacks to Benchmark are: awkward assembly language interface, no control over data placement (near/far/register), the linker includes all code in modules imported (it's not smart), global data is allocated in file space (making huge files). Even with these disadvantages, Benchmark runtime speed it quite good, but not as good as Lattice. Lattice on the other hand (or even Manx), has excellent control over everything: code, data, assembly. Lattice has a Global Optimizer, direct library calls, your choice of register or stack function parameters, etc., giving it the current runtime edge over the competitors (hand-tuned assembly will shred the rest for some time). Only two major drawbacks to Lattice: the integrated environment using LSE is weak; solution: CygnusEd Professional with an Arexx port, compile times are slow, even with precompiled headers, residencies, and libraries in ram; solution: 1. Drop in an '020 or '030 with 32 bit ram, this may increase speed 2 to 4 times. 2. Lattice comes out with a "Turbo" version of C. I talked to Lattice at AmiExpo; they stated that their 5.0 compiler ran at "about 2000 lines per minute." So, if you can afford an '020 or '030 and a good glob of 32 bit ram, you might get 8000 lines a minute from Lattice. Same case with Benchmark would be 40,000-120,000 lines per minute. A "Turbo" C compiler looks like the way to go (incremental/one pass). Heck, they've got LightSpeed for the Mac, Turbo, Quick, and Zortech for ibm, porque no por la Amiga, eh? John Schultz I predict the first "Turbo" C compiler for the Amiga will nuke the competition. Hands down. Bar none.