speed of compilation

[slenkar]

it does seem to compile faster good one

[LCD]

I noticed a speedup of circa 100%. Excellent!!!
1500 lines of program compile now in 9 seconds with s888, same program was compiled in 16 seconds with s882.

[britlion]

I was reading this over, and decided to give -O3 another try. I generally don't use it, because even though it should produce a binary that does the same thing, I've never actually got it to work. It either fails to compile, or produces a binary that doesn't behave the same way..

With the latest version of Berksman I have, and the latest version of the compiler:

Without O3 - compiles perfectly. (And I notice a score bug I have to fix!)

With O3: Fails to compile.

Traceback (most recent call last):
File "zxb.py", line 317, in <module>
File "zxb.py", line 272, in main
File "optimizer.pyc", line 2278, in optimize
File "optimizer.pyc", line 1456, in update_goes_and_comes
IndexError: list index out of range
Build Failed!


(To be fair, I did expect the fact I have an Im2 routine to be a deal breaker. No idea how to persuade it to keep that)

Edit2: Actually, the ISR doesn't call any named functions. It just jumps to a piece of code @address. That ought to work..

Latest source attached