... fixed intervals of about 350 T states. Not a small feat by ...

... ;[7]
[/code]
19 T states vs. 24.

... sides too) where contention isn't an issue.[/quote]
I ... fluctuations up to about 12K T states, as well as maybe help ...

... .. that's interesting. I hadn't though of doing that. Does ... (give or take a few t-states) when I sync.

... .. that's interesting. I hadn't though of doing that. Does ... (give or take a few t-states) when I sync.

Good, but could be optimized a tiny bit to shave off 20 T states and 3 bytes:

[code]
ld (quit_sp+1),sp

...
...

quit_sp ld sp,0

[/code]

... bit to shave off 20 T states and 3 bytes:

... or may not correspond to T states at which these bytes are ...

... or may not correspond to T states at which these bytes are ...

... or may not correspond to T states at which these bytes are ... normal though). If it doesn't, well, too bad!

... at the left aren't addresses, they're 'T states', basically timing information ... divided into a number of T states, the exact value depending on ... it has read, along with T state numbers at which they ... something. The reason I don't think it's necessary is ...

... border or the four #FF T states between each two bitmap/attribute ...

... +2A/+3 is displayed at T state 14365 and the contention ... border and the blank (#FF) T states. Odd how it repeats the ... on the actual hardware. Don’t know whom to believe anymore ...

... and the blank (#FF) T states. Odd how it repeats the ... the actual hardware. Don’t know whom to believe ... other, like a few T states maybe, but that's ... perhaps they just aren't referring to what I ... port reading and whatnot shouldn't be affected by contention ...

... .[/quote]
Ah, no. Can’t do that, because I’m ... bitmap bytes. At all. Counting T states seems to be an exercise ... (LD A,(0) [i]doesn’t[/i] work).

This ...

... rock solid. Sadly, I don't think this sort of magic ... ]

Different timing (25 T states vs. 21) yet it’s ... bitmap pattern), but it doesn’t. No contention, mind. I’ll ...

... have empty gaps of 100 T states (corresponding to the border at ...

... very different anyway. I don't see the need for this ... seen what is returned at T states 'before the screen' in Woody ... , but not my test (the T state listing part). If it ...

... into side border time (100 T states). You have a LD A ... time is reduced to 100 T states per line again, in turn ...

... border or during the four T states in between reading the two ...

... border or during the four T states in between reading the two ...

... the game (yes, those 28 t-states per screen update will really ...

I've managed to lose another 10880 t-states (a whopping 0.003 seconds!) from a complete build cycle by moving a few bytes around, but not managed to save any more bytes....not YET

Oh hang on...-1 byte, so now [b]725 bytes![/b]

[quote="Arkannoyed;926363"]I've managed to lose another 10880 t-states[/quote]

!!!!!

... little too by removing 7 t-states per line search cycle. Doesn ... ’t seem much, but that area ...

... . The new version is 13 t-states faster in the same size ... />
Oh and another 11 t-states per line saved now too ...

... all a saving of 768 t-states.

[code]
nxt_block ...

... a huge saving of 2112 t-states.

Amazing to think ...

... to exploit, but I haven't tried it myself. I ... read / write to 2 T-states out of every 8. But ... written over 16 T-states (one byte every 8 T-states).

... gaps of 7 and 9 T-states (plus any multiple of ... a 48K or 128K wouldn't have a problem with ...

... , and will save you 5 T states per big_lp iteration (and even ... 1 T state per an entire routine ...