🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: comp.sys.atari.st
Article #35873

Turbo-C

#35873
From: Jos Vermaseren
Date: Mon, 09 Jan 1989 17:56
69 lines
3155 bytes

I wrote a review of 5 C compilers for the Dutch magazin start (no connection
with the US magazin) a few months ago. In it we compared (Daan Jitta and me)
Laser C, Lattice, MarkWilliams, Aztec and Turbo C.
Nearly all categories were won by Turbo. The category that it does worst
in is the reliability of its libraries. This is explained by the fact that
it is still a rather new product.
Here are the results of some benchmarks:

Savage:
            compil  exec    error
Aztec       10.3    68.04   -3.177E-07
Laser        9.5    36.27   +1.737E-10
Lattice     32.7    40.01   -3.177E-07
MWC         27.1    85.14   -1.188E-07
Turbo       12.6    31.82   +1.737E-10

Dhrystone:
            comp    Dhrystones/sec
Aztec       17.2    1137
Laser       12.1     845
Lattice     38.6     786
MWC         29.5    1048
Turbo       17.6    1702    (compilation delayed by two pages warnings (ANSI))

Next I compiled a larger program, consisting of about 20 files (> 300K together)

            comp        exec     size,size during execution
Aztec       573         97      74924,83230
Laser       263         99      91006,96444
Lattice    1230        111     103668,100554
MWC         633         84      78079,83222
Turbo       201         74      62180,68688

This last program didn't use floating point and used hardly anything of
the libraries.

With other programs MWC usually beats Aztec in the length of the file.
The running time difference between MWC and Aztec in the last example
is mainly due to the fact that Aztec cannot use predecrement and post
increment.

The above compilations were from ramdisk (Mega 4). The compiler was sitting
on hard disk and had to be called each time by make.
registers were used as much as possible.

This last program was designed with MWC and fairly much optimized to produce
efficient code with it.

In my programs typical rates are that code is about 20% shorter and faster.


By the way:
Turbo C was made by Heimsoeth with much help from Borland.
When it is a hit in Germany they will sell it in the US under the name Borland.
I haven't seen the official release of the assembler and the debugger yet.
The assembler (in beta version) uses motorola instructions and looks in its
use like what the DRI assembler should have been. Because it hasn't been
released yet I don't know about the macro facilities. My manual is a file
that contains about 50 lines or less about the assembler. Its a good 
workhorse though.
The debugger is at assembler level (the source level debugger will come
later. If it looks like the one on the PC it will be very good) and looks
a little bit like a very much improved SID.

Price of turbo is less than DM200 in Germany which is a little over $100.

Jos Vermaseren
Technical Details
Message-ID: <601@nikhefh.hep.nl>
Path: arc1-ca.pugleaf.net!utzoo!utgpu!watmath!clyde!att!osu-cis!killer!ames!mailrus!cornell!batcomputer!itsgw!steinmetz!uunet!mcvax!nikhefh!t68

Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!utzoo!utgpu!watmath!clyde!att!osu-cis!killer!ames!mailrus!cornell!batcomputer!itsgw!steinmetz!uunet!mcvax!nikhefh!t68
From: t68@nikhefh.hep.nl (Jos Vermaseren)
Newsgroups: comp.sys.atari.st
Subject: Turbo-C
Message-ID: <601@nikhefh.hep.nl>
Date: 9 Jan 89 17:56:43 GMT
Organization: Nikhef-H, Amsterdam (the Netherlands).
Lines: 69