swupiox wrote
I finished chapter 7 and all tests pass, but it is bugging me that I had to use a temp register for pop to store the address of base+offset. Is that expected or is there some clever way to only use A and D that I have missed? In the VME the temp registers don't seem to be affected by pop.
One thing I considered was writing A=A+1 each time for the number of offsets from base, but although it would work I don't think it is a good way to do it. I am not asking for the solution (if there is one) but just a hint. Thanks.
My generic "pop pointer N" uses a temp register. It's 12 instructions long. My code generator writes shorter code for N=0 and 1, but I didn't bother to write specific code for N>1 cases that would be shorted than 12 instructions. A quick search through the OS .vm files showed that optimizing N=0,1 would have a good return.
Here are the counts for my game with my OS.
pop.argument.0 9
pop.argument.1 8
pop.argument.2 1
pop.argument.3 1
pop.local.0 57
pop.local.1 37
pop.local.10 2
pop.local.2 26
pop.local.3 20
pop.local.4 12
pop.local.5 5
pop.local.6 3
pop.local.7 5
pop.local.8 5
pop.local.9 2
pop.temp.0 283
pop.that.0 45
pop.this.0 9
pop.this.1 9
pop.this.2 14
pop.this.3 7
pop.this.4 1
pop.this.5 2
Looks like it might be worth looking at the A=A+1 optimization for N>1 cases...
--Mark
Locked
