From:	Michael Castle <mcastle AT umr DOT edu>
Subject:	Re: DJGPP Speed
To:	eliz AT is DOT elta DOT co DOT il
Date:	Tue, 29 Mar 1994 18:27:02 -0600 (CST)
Cc:	djgpp AT sun DOT soe DOT clarkson DOT edu

>   Using the ``topline'' parameter of the GO32 environment variable, it can be
> seen that  CPP and AS programs together take about 15% of the total compilation
> time, so that (a frequently heard) argument about GCC being multi-pass as
> oppposed to a single-pass BCC, is probably *not* an explanation to the speed
> difference.
> 
> 		Eli Zaretskii
> 

Try running TCC the same way that GCC is run:  

Run source through CPP first, then through TCC, then through TASM.  Compare
that with the time that GCC takes.  

preprocessed and assembler files tend to be MUCH bigger then the C source
(I've had *.c files approx 30k blow up to 1.5M *.s files).  Reading and
writing by cc1 CAN be a big impact.  

Also, are you using 32bit or 16bit version of gcc.exe?  That is, are you
spending time swapping gcc in and out of memory?


*shrug*  If you don't like the speed... get the source and tweak it...I
imagine most of the bottleneck is still in the I/O.  

Remember:  GCC is designed to be PORTABLE first; speed of the compiler and
speed of the generated code is secondary.  

If you're concerned that it takes 10 minutes to compile code rather than 2
minutes... then you're probably spending too much time in the compiler and
less time designing and verifying code.  

- Raw text -