test please delete

[SimonMac]

To

[SimonMac]

Try

[SimonMac]

And

[SimonMac]

Pad

[SimonMac]

This

[SimonMac]

Out

[SimonMac]

Just

[NickFitz]

Oh it's even better than that, it won't compile at all if you include all the function calls, so you have to do a conditional compile just to fit the fecker into memory.

It runs out of ram rather than rom.

Ah, those were the days

When I was doing Night Hunter on the Amiga, I configured the assembler to stick the Copper display list at the end of memory, because if I ran out of memory that meant that the first frame overwrote the end of the display list, which caused the display to collapse. If that happened, I knew I had to find a way to shrink things down by enough bytes to make it fit. No memory management in hardware on the 68000

I once spent an entire day identifying a routine I could modify without any complex ramifications (all registers in and out and no memory locations modified) and, after fine-tuning my algorithm, gaining about twelve bytes. The next day, I managed to find the additional four bytes I needed before I could finally test the code I'd (by then) written the day before yesterday

IIRC, the finished product had about 112 bytes free out of 512MB. That was where the code that handled serial port receive interrupts lived. Those allowed me to stop the Amiga in its tracks with a utility I'd written in Forth on the ST, and use a tiny protocol of my own devising to poke around inside the Amiga getting memory and register dumps back from it. It was the only way to get any debugging information when there wasn't enough memory left for a debugger.

By God, we knew how to make the most of computers in those times

[SimonMac]

So

[SimonMac]

I.