Lpc2000

I've been working with AVR MCUs for a while but the higher performance
of the ARM parts has recently captured my attention.  So far I've
built my own binutils-2.16.1, gcc-3.4.4, newlib-0.12, gdb-6.3
toolchain (from scratch) and I'm trying to compile some test code.

Using a simple:
arm-elf-gcc -o test -Os test.c
yields binaries in the 44K range (stripped).  My test code is
basically just a function that returns a constant and a main() --- so
this seems a bit big.  I also tested the GNUARM toolchain with the
same command and got very similar results.

From my reading so far, I gather that I'll need to make an inital
startup asm file and a linker command file as well.  What other things
will I need for 210X/213X development?

More importantly though, what is the correct set of compiling, linking
and stripping commands to produce code that can be loaded into the MCU?

I'm using linux, emacs and the command line for my IDE if thats
relevant.  I also don't mind reading documentation/instructions if
anyone can point me in the right direction...

Thanks,
David Carr