home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Info-Mac 3
/
Info_Mac_1994-01.iso
/
Development
/
General
/
GCC 1.37.1r15
/
Documents
/
Statistics
< prev
next >
Wrap
Text File
|
1990-08-02
|
2KB
|
38 lines
Some Statistics Comparing MPW C To MPW GCC
This is not a formal benchmark study, but a brief summary of what we've observed in
working with MPW C and GCC.
Experimental Setup
The numbers are all based on a Mac IIci running Multifinder 6.0.4/5, with a Daystar
cache card, external video, 8M RAM, and the internal 80M hard disk. No virtual
memory, and the RAM cache is set at 96K. MPW partition is 3000K, with a 128K stack.
The C compilers are out-of-the-box MPW 3.1 C and MPW GCC 1.37.1r5.
Compiler Size
MPW C is 338K, while GCC's cpp and cc1 together are 604K (self-compiled).
Compiler Speed
GCC is much slower to compile than MPW C, partly because it runs three separate
programs, and partly because it does more optimization. Typical GCC compilation
times are 3-4 times longer than MPW C. For instance, MPW C takes 16 min to
compile and link all of cpp and cc1, while the MPW-compiled GCC takes 58 min,
and the self-compiled GCC takes 48 min.
Code Size
GCC output code is from 5% to 25% smaller than for MPW, with an average of 10%.
For instance, self-compiled cc1 is 13% smaller than MPW C-compiled cc1, while cpp
is 9% smaller.
Code Speed
GCC output code is from 5% to 32% faster than MPW code, with an average of 10%.
Self-compilation is 18% faster with a self-compiled GCC than with an MPW-compiled
GCC. Dhrystone is 25% faster, going up to 32% if C versions of the string functions
are used (in fact, dhrystone with C-coded string functions is faster than with the
library functions, for both MPW C and GCC).