Re: Has any one tried the Lucid 'C' compiler?

markg@urchin.csc.ti.com (Mark Greenberg)
Tue, 2 Jul 1991 16:13:51 GMT

          From comp.compilers

Related articles
Has any one tried the Lucid 'C' compiler? trdlnk!psm@uunet.UU.NET (1991-06-25)
Re: Has any one tried the Lucid 'C' compiler? markg@urchin.csc.ti.com (1991-07-02)
C compiler comparisons (Lucid vs. Sun) David.Chase@Eng.Sun.COM (1991-07-03)
| List of all articles for this month |

Newsgroups: comp.compilers
From: markg@urchin.csc.ti.com (Mark Greenberg)
Keywords: C, benchmarks, Lucid
Organization: TI Computer Science Center, Dallas
References: 91-07-006
Distribution: usa
Date: Tue, 2 Jul 1991 16:13:51 GMT

In article 91-07-006 trdlnk!psm@uunet.UU.NET (Paul Stephen Murphy) writes:


      Has any one tried the Lucid 'C' compiler? If so, how does it stack up
      against gcc? I'd like to know before I invest in a new compiler.


I've just returned my "trial" copy after having done a little
benchmarking. Here are some results:-


                nfib sim netsim dry pqueens Xserver
--------------------------------------------------------
lcc O4 45.5 88.4 808.0 32822 85.8
lcc O3 46.5 88.5 772.5 32432 86.8
lcc 45.9 108.5 840.0 18714 88.9


sun O4 46.4 82.0 888.0 22779 143.4
sun O3 46.9 83.2 847.0 22831 143.0
sun 85.9 151.0 1365.0 12684 234.2


csc O2 44.1 88.3 698.0 24449 141.6
csc 51.7 96.5 945.0 14829 145.5
--------------------------------------------------------
lcc = Lucid C
sun = Sun bundled compiler
ticc = TI in-house C compiler (no optimisation done in the past year)


nfib: The fibonacci function modified to produce a result equal to the
                  times the function is called recursively. The argument supplied
                  is 35, resulting in 29860703 function call.


sim: Simulation of a 16 processor reduction machine running a
                  program to compute the 20th fibonacci function in parallel.


netsim: Low level simulation of 100,000 64 byte messages flowing through
                  a 128x128 multistage network.


dry: Drystone benchmark in dhrystones/sec.


queens: Divide-and-conquer computations of n-queens problem using list
to represent chess broards. n is 15.


Xserver: X11 server running a series of benchmark programs. It was
taking 6hrs to compile, so stopped experimenting.
--------------------------------------------------------
The compiler has some nice feature like flags for switching between K&R
and ANSI C, for inlining etc, but most of these features are available
under gcc. Compilation speed good, but not as good at ticc!


Mark.
--
Mark I Greenberg, Computer Science Centre, Texas Instruments,
13510 N. Central Exprway, PO.Box 655474, MS-238, Dallas, Tx 75265
Fax: 214 995-0304 Tel: 214 995-0778 Email: markg@csc.ti.com
--


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.