Re: A86: What C-compilers have we got? [82/83/83+/85/86]
[Prev][Next][Index][Thread]
Re: A86: What C-compilers have we got? [82/83/83+/85/86]
On Wed, 8 Nov 2000, Adam Thayer wrote:
> Yes, David is very correct on the point about C compilers (my own Asm
> programs usually hit 1k for small ones before optimization)... they
> just produce code that is too large for the 86's RAM space, let alone
> an 83/etc. C can be done on the 89/92+ since the RAM available is
> around 1MB for the 92+, and 512k for the 89 (this includes Archive
> memory), and the Asm program limit is greater than the ~8k of the 86,
> which means C has the room it needs on those calcs (plus the addition
> of the 68k instruction set and registers to make code smaller than
> z80 equiv).
Have you ever bothered to look at the output of TI-GCC for the 68K
calculators?
While TI-GCC doesn't generate the best code possible, uses stack-based
parameter passing, and favors speed over size, vast quantities of memory
are by no means required for those programs.
In my own experience with TI-GCC, I am repeatedly surprised by how good
its code generation actually is. With the exception of the speed penalty
from using the stack to pass parameters to functions, its output is as
fast as most assembly code I've seen, and I can hardly imagine that any
reasonable segment of code of substantial size would even be near twice
the size of assembly if compiled with TI-GCC.
Follow-Ups:
References: