2014-10-10 12:14:29 +02:00
|
|
|
avremu: An AVR Emulator written in pure LaTeX
|
2014-09-30 14:48:31 +02:00
|
|
|
======================================================
|
|
|
|
|
2014-10-13 13:48:21 +02:00
|
|
|
![A Mandelbrot Set. The image took 6 hours to be calculated](imgs/mandelbrot-250x250.png)[5]
|
2014-09-30 14:48:31 +02:00
|
|
|
|
|
|
|
LaTeX is known as a typesetting system. But the underlying TeX system
|
|
|
|
is a powerful macro processor. In fact, TeX is a Turing-complete
|
|
|
|
programming language. TeX can compute anything, that is
|
|
|
|
computable. Computeability is a concept from theoretical computer
|
|
|
|
science. After visiting a theoretical CS course, you will know that
|
|
|
|
there are things that cannot be solved by a machine. Never. Look out
|
2014-10-09 18:15:27 +02:00
|
|
|
for the halting problem. So, Turing-complete does not say "It's a
|
2014-09-30 14:48:31 +02:00
|
|
|
proper programming language", but more "someone can write a proper
|
|
|
|
interpreter for any programming language in it, theoretically".
|
|
|
|
|
|
|
|
So back to TeX. To be honest, TeX is a horrible language to code
|
|
|
|
in. After writing my master thesis in LaTex, after crafting every
|
|
|
|
graphic within with TiKz[1] and pgfplots[2], and after writing my
|
|
|
|
first LaTeX package[3], I can honestly say: TeX is the Quasimodo of
|
|
|
|
all widely used programming languages. If you can avoid it, avoid it.
|
|
|
|
But if you are forced to use it, you can abstract from TeX, since it
|
|
|
|
is a "proper" programming language.
|
|
|
|
|
|
|
|
Let's say we want to program our document in C on Top of Tex. Then
|
|
|
|
there would be the need to interpret the C semantic. Basically we
|
|
|
|
would have to write a compiler for C in TeX. Not a very promising
|
|
|
|
future, if you start this. But, we can use a normal C compiler to
|
|
|
|
compile the C code to some simple platform. Like AVR[4]. Since I have
|
|
|
|
other things to do than writing an AVR Emulator in TeX, here it is!
|
|
|
|
|
2014-10-10 12:14:29 +02:00
|
|
|
The newest source can be found at https://gitlab.brokenpipe.de/stettberger/avremu
|
2014-10-09 18:15:27 +02:00
|
|
|
|
2014-09-30 14:48:31 +02:00
|
|
|
## You are writing an CPU emulator in TeX, the TYPESETTING system? ##
|
|
|
|
|
|
|
|
Yep.
|
|
|
|
|
|
|
|
## What Microcontroller is emulated? ##
|
|
|
|
|
|
|
|
I aimed for an ATmega8. The AVR platform is an 8 Bit Processor with 16
|
|
|
|
Bit memory addresses. The hardware version is shipped with 1 KByte of
|
|
|
|
RAM and 8 Kbyte Flash memory, but of course, the emulated processor
|
|
|
|
can handle more memory.
|
|
|
|
|
|
|
|
## Are you insane? ##
|
|
|
|
|
|
|
|
Not that anybody knows of.
|
|
|
|
|
|
|
|
## Are all OP codes implemented? ##
|
|
|
|
|
|
|
|
Unfortunately not. I'm still missing 5 opcodes (MULS, MULSU, FMUL,
|
|
|
|
FMULS, FMULSU), but until now I did not encounter a C file, the
|
|
|
|
AVR-GCC emits these opcodes. The fractional multiply opcodes are also
|
|
|
|
somewhat arcane.
|
|
|
|
|
|
|
|
## Is it fast? ##
|
|
|
|
|
|
|
|
Ok, you are asking me, whether a CPU simulator written in TeX, that
|
|
|
|
simulates an 8-Bit microcontroller is fast? Propably not. On my
|
|
|
|
notebook I could achieve 2.5 KHz, so this is a factor 10000 slower
|
|
|
|
than the actual hardware with the highest supported clock rate.
|
|
|
|
|
|
|
|
-----
|
|
|
|
|
|
|
|
[1] http://www.ctan.org/pkg/pgf
|
|
|
|
|
|
|
|
[2] http://pgfplots.sourceforge.net/
|
|
|
|
|
|
|
|
[3] http://www.ctan.org/pkg/dataref
|
|
|
|
|
|
|
|
[4] http://en.wikipedia.org/wiki/Atmel_AVR
|
|
|
|
|
2014-10-13 13:48:21 +02:00
|
|
|
[5] A Mandelbrot Set. The image took 44 hours to be calculated
|