imjasonh/nescript
1
0
Fork 0
mirror of https://github.com/imjasonh/nescript synced 2026-07-08 17:06:04 +00:00
Code Activity

runtime: gate controller-2 reads in NMI on __p2_input_used

Browse source 
Drop the three-instruction JOY2 shift block (`LDA $4017 / LSR A /
ROL ZP_INPUT_P2`) from inside the NMI's 8-iteration input loop
when user code never reads controller 2. IR codegen emits the
`__p2_input_used` marker from `IrOp::ReadInput(_, 1)`; the linker
threads the flag through a new `NmiOptions::has_p2_input` bool,
and `gen_nmi` writes the shift block only when the flag is set.

Savings for single-player programs:
 - ~6 bytes of NMI code.
 - ~30 cycles per frame (3 instructions × 8 loop iterations, each
   6-8 cycles depending on addressing — LDA abs is 4, LSR A is 2,
   ROL zp is 5, so ~11 cycles × 8 = ~88 cycles; rounded down for
   the page-crossing penalty landing differently in the new layout).

This commit also fixes the IR codegen to drop the matching
`__p1_input_used` marker from `IrOp::ReadInput(_, 0)`, even though
the next commit is the one that actually consumes it. Landing the
two markers together keeps the IR codegen's per-op bookkeeping
coherent.

Six audio goldens flip (every program that reads input + plays
audio) with the expected NMI-layout-shift cycle drift.

https://claude.ai/code/session_016kM6P7PukktBDqTZexrrAN
This commit is contained in:
Claude 2026-04-16 13:43:33 +00:00
parent bd30ac3010
commit 0de1d60c33
No known key found for this signature in database
40 changed files with 70 additions and 11 deletions
Download patch file Download diff file Expand all files Collapse all files

21
src/runtime/mod.rs
View file

@ -386,6 +386,12 @@ pub struct NmiOptions {
/// so the DMA is ~520 wasted cycles per NMI. Skipping it saves
/// those cycles plus 9 bytes of NMI code.
pub has_oam: bool,
/// When false, drop the three instructions that shift `$4017`
/// (JOY2) into `ZP_INPUT_P2` from the NMI's 8-iteration input
/// loop. Single-player programs save ~6 bytes of code and ~30
/// cycles per frame (a `LDA abs`, an `LSR A`, and a `ROL zp`
/// running 8 times).
pub has_p2_input: bool,
}
#[must_use]
@ -396,6 +402,7 @@ pub fn gen_nmi(opts: NmiOptions) -> Vec<Instruction> {
debug_mode,
has_sprite_cycle,
has_oam,
has_p2_input,
} = opts;
let mut out = Vec::new();
@ -506,16 +513,20 @@ pub fn gen_nmi(opts: NmiOptions) -> Vec<Instruction> {
out.push(Instruction::new(STA, AM::Absolute(JOY1)));
// Read 8 button bits from controller 1 ($4016) into ZP_INPUT_P1
// and 8 button bits from controller 2 ($4017) into ZP_INPUT_P2
// simultaneously — shift each port's carry into its ZP byte.
// — and, when `has_p2_input` is set, 8 bits from controller 2
// ($4017) into ZP_INPUT_P2 in the same loop. Single-player
// programs drop the three P2 instructions (LDA abs, LSR A, ROL
// zp) and shave ~6 bytes plus ~30 cycles/frame off the NMI.
out.push(Instruction::new(LDX, AM::Immediate(0x08)));
out.push(Instruction::new(NOP, AM::Label("__read_input".into())));
out.push(Instruction::new(LDA, AM::Absolute(JOY1)));
out.push(Instruction::new(LSR, AM::Accumulator));
out.push(Instruction::new(ROL, AM::ZeroPage(ZP_INPUT_P1)));
out.push(Instruction::new(LDA, AM::Absolute(0x4017))); // JOY2
out.push(Instruction::new(LSR, AM::Accumulator));
out.push(Instruction::new(ROL, AM::ZeroPage(ZP_INPUT_P2)));
if has_p2_input {
out.push(Instruction::new(LDA, AM::Absolute(0x4017))); // JOY2
out.push(Instruction::new(LSR, AM::Accumulator));
out.push(Instruction::new(ROL, AM::ZeroPage(ZP_INPUT_P2)));
}
out.push(Instruction::implied(DEX));
out.push(Instruction::new(
BNE,