feat(video): bound per-row sprite preparation work

Apply nominal 1210/954-cycle allowances, affine setup costs, OAM-order allocation, and partial sprite prefixes.

src/video/sprites.rs

@@ -1,6 +1,7 @@
-//! Regular and affine OBJ sampling.
+//! Regular and affine OBJ sampling with a nominal per-row work allowance.
 //! Individual fetch timing and OAM contention are not modeled.
 
+mod budget;
 pub(crate) mod pipeline;
 
 use super::{halfword, Pixel, VideoError, WIDTH};
@@ -53,6 +54,7 @@ pub(crate) fn prepare(
         return Ok(output);
     }
     let samples = &mut output.samples;
+    let mut budgets = vec![budget::Budget::new(control); rows.len()];
     let mosaic_height = usize::from(mosaic >> 12) + 1;
     let one_dimensional = control & 0x40 != 0;
     for index in 0..128 {
@@ -60,7 +62,10 @@ pub(crate) fn prepare(
         let b = usize::from(halfword(oam, index * 8 + 2));
         let c = usize::from(halfword(oam, index * 8 + 4));
         if a & 0x300 == 0x200 {
-            continue; // Disabled regular OBJ.
+            for budget in &mut budgets {
+                budget.skip();
+            }
+            continue; // Disabled regular OBJ still consumes nominal OAM inspection.
         }
         let reason = if a & 0xc00 == 0xc00 {
             Some("prohibited OBJ mode")
@@ -89,17 +94,27 @@ pub(crate) fn prepare(
         };
         let eight_bit = a & 0x2000 != 0;
         let mut tile = c & 0x3ff;
-        if control & 7 >= 3 && tile < 512 {
+        // Restricted bitmap tiles produce no pixels, but active sprite work
-            continue; // Bitmap modes reserve the first half of OBJ tile memory.
+        // still uses the allowance. They are not disabled OAM entries.
-        }
+        let tile_visible = !(control & 7 >= 3 && tile < 512);
         if eight_bit && !one_dimensional {
             tile &= !1; // 2D 8bpp tiles are aligned to pairs of 32-byte slots.
         }
         let slot_width = if eight_bit { 2 } else { 1 };
         let priority = ((c >> 10) & 3) as u8;
         for y in rows.clone() {
+            let budget = &mut budgets[y - rows.start];
             let sy = (y + 256 - (a & 255)) & 255;
-            if sy >= canvas_height {
+            let left = b & 511;
+            let intersects_screen = left < WIDTH || left + canvas_width > 512;
+            if sy >= canvas_height || !intersects_screen {
+                budget.skip();
+                continue;
+            }
+            // Charge before texel lookup or priority/window masking. Transparent,
+            // hidden, and out-of-texture samples still require preparation work.
+            let columns = budget.draw(canvas_width, affine);
+            if !tile_visible {
                 continue;
             }
             // Preparation latches the ahead-of-display OBJ counter; snapshots
@@ -111,7 +126,7 @@ pub(crate) fn prepare(
             } else {
                 sy
             };
-            for sx in 0..canvas_width {
+            for sx in 0..columns {
                 let x = ((b & 511) + sx) & 511;
                 if x >= WIDTH {
                     continue;

src/video/sprites/budget.rs

@@ -0,0 +1,124 @@
+//! Nominal per-row OBJ work allowance, not a timed OAM/VRAM fetch scheduler.
+//! GBATEK gives 1210/954 cycles and costs of width or 10+2*width.
+//! Inactive entries cost two inspection cycles, following emulator references.
+//! Active costs include that inspection; clipped active canvases cost full width.
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+pub(super) struct Budget {
+    remaining: usize,
+}
+
+impl Budget {
+    pub fn new(control: u16) -> Self {
+        Self {
+            remaining: if control & 0x20 != 0 { 954 } else { 1210 },
+        }
+    }
+
+    pub fn skip(&mut self) {
+        self.remaining = self.remaining.saturating_sub(2);
+    }
+
+    /// Return the prepared prefix of the drawing canvas. Unfinished work consumes
+    /// the remaining allowance; it cannot be reused by a later, cheaper object.
+    pub fn draw(&mut self, width: usize, affine: bool) -> usize {
+        if self.remaining < 2 {
+            self.remaining = 0;
+            return 0;
+        }
+        let (setup, per_pixel) = if affine { (10, 2) } else { (0, 1) };
+        let pixels = (self.remaining.saturating_sub(setup) / per_pixel).min(width);
+        self.remaining = self.remaining.saturating_sub(setup + per_pixel * width);
+        pixels
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn every_allowance_and_width_matches_independent_cycle_consumption() {
+        for initial in 0usize..=1210 {
+            for width in [8, 16, 32, 64, 128] {
+                for affine in [false, true] {
+                    let mut expected = initial;
+                    let mut pixels = 0;
+                    if expected < 2 {
+                        expected = 0;
+                    } else {
+                        if affine {
+                            for _ in 0..10 {
+                                expected = expected.saturating_sub(1);
+                            }
+                        }
+                        for _ in 0..width {
+                            let needed = if affine { 2 } else { 1 };
+                            if expected < needed {
+                                expected = 0;
+                                break;
+                            }
+                            expected -= needed;
+                            pixels += 1;
+                        }
+                    }
+                    let mut budget = Budget { remaining: initial };
+                    assert_eq!(
+                        budget.draw(width, affine),
+                        pixels,
+                        "initial={initial} width={width} affine={affine}"
+                    );
+                    assert_eq!(budget.remaining, expected);
+                }
+            }
+        }
+    }
+
+    #[test]
+    fn inactive_scan_and_active_costs_do_not_double_charge_inspection() {
+        let mut normal = Budget::new(0);
+        assert_eq!(normal.draw(64, false), 64);
+        assert_eq!(normal.remaining, 1210 - 64);
+        normal.skip();
+        assert_eq!(normal.remaining, 1210 - 66);
+        assert_eq!(normal.draw(128, true), 128);
+        assert_eq!(normal.remaining, 1210 - 66 - 266);
+        let mut short = Budget::new(0x20);
+        for _ in 0..119 {
+            assert_eq!(short.draw(8, false), 8);
+        }
+        assert_eq!(short.draw(8, false), 2);
+        assert_eq!(short.draw(8, false), 0);
+    }
+
+    #[test]
+    fn random_mixed_work_matches_an_independent_remaining_time_model() {
+        let mut seed = 0x8321_602bu32;
+        for control in [0, 0x20] {
+            for _ in 0..1000 {
+                let mut budget = Budget::new(control);
+                let mut expected: i32 = if control == 0 { 1210 } else { 954 };
+                for _ in 0..128 {
+                    seed = seed.wrapping_mul(1664525).wrapping_add(1013904223);
+                    let width = 8 << ((seed >> 4) % 5);
+                    let affine = seed & 8 != 0;
+                    if seed & 3 == 0 {
+                        budget.skip();
+                        expected = (expected - 2).max(0);
+                    } else {
+                        let columns = if expected < 2 {
+                            0
+                        } else if affine {
+                            ((expected - 10).max(0) / 2).min(width)
+                        } else {
+                            expected.min(width)
+                        };
+                        assert_eq!(budget.draw(width as usize, affine), columns as usize);
+                        expected = (expected - if affine { 10 + 2 * width } else { width }).max(0);
+                    }
+                    assert_eq!(budget.remaining, expected as usize);
+                }
+            }
+        }
+    }
+}

tests/sprite_budget.rs

@@ -0,0 +1,306 @@
+//! Regression tests for the nominal allowance, not hardware fetch cutoffs.
+use gba_rust::{
+    cpu::Cpu,
+    display::{CYCLES_PER_LINE as LINE, VBLANK_START as VB},
+    dma::DMA_BASE,
+    io::*,
+    machine::Machine,
+    memory::{Memory, OAM_START as OAM, PALETTE_START as PAL, ROM_START, VRAM_START as VRAM},
+    video::{rgb555_to_rgb888 as rgb, Framebuffer, VideoError, WIDTH},
+};
+
+fn scene(free: bool) -> Memory {
+    let mut m = Memory::new(vec![]).unwrap();
+    m.set_scanline_rendering(true);
+    m.write16(DISPCNT, if free { 0x1060 } else { 0x1040 })
+        .unwrap();
+    m.write16(PAL, 0x7c00).unwrap();
+    m.write16(PAL + 0x202, 31).unwrap();
+    m.write16(PAL + 0x204, 0x3e0).unwrap();
+    for i in 0..128 {
+        obj(&mut m, i, 0x200, 0, 0);
+    }
+    // Tile zero is transparent. Separate solid 64x64 textures use indices 1/2.
+    for offset in (0..2048).step_by(2) {
+        m.write16(VRAM + 0x11000 + offset, 0x1111).unwrap();
+        m.write16(VRAM + 0x12000 + offset, 0x2222).unwrap();
+        m.write16(VRAM + 0x14000 + offset, 0x2222).unwrap();
+    }
+    m
+}
+
+fn obj(m: &mut Memory, index: u32, a: u16, b: u16, c: u16) {
+    m.write16(OAM + index * 8, a).unwrap();
+    m.write16(OAM + index * 8 + 2, b).unwrap();
+    m.write16(OAM + index * 8 + 4, c).unwrap();
+}
+
+fn fillers(m: &mut Memory, count: u32) {
+    for i in 0..count {
+        obj(m, i, 0, 0xc000, 0);
+    }
+}
+
+fn snapshot(m: &Memory) -> Framebuffer {
+    let before = (m.cycles(), m.display_position());
+    let mut f = Framebuffer::default();
+    m.render_frame(&mut f).unwrap();
+    assert_eq!((m.cycles(), m.display_position()), before);
+    f
+}
+
+fn finish(m: &mut Memory) -> Framebuffer {
+    let p = m.display_position();
+    m.advance_cycles(VB - u32::from(p.scanline) * LINE - u32::from(p.line_cycle));
+    let mut f = Framebuffer::default();
+    assert!(m.present_frame(&mut f).unwrap());
+    f
+}
+
+fn at(f: &Framebuffer, x: usize, y: usize) -> u32 {
+    f.pixels()[y * WIDTH + x]
+}
+
+fn prefix(f: &Framebuffer, y: usize, left: usize, width: usize, visible: usize) {
+    for x in left..left + width {
+        assert_eq!(
+            at(f, x, y),
+            rgb(if x < left + visible { 0x3e0 } else { 0x7c00 }),
+            "pixel={x},{y}, prefix={visible}"
+        );
+    }
+}
+
+#[test]
+fn regular_prefix_exhausts_allowance_and_omits_later_entries() {
+    for free in [false, true] {
+        let mut m = scene(free);
+        let count = if free { 14 } else { 18 };
+        fillers(&mut m, count);
+        obj(&mut m, count, 0, 0xc000 | 80, 256 | 0xc00);
+        obj(&mut m, count + 1, 0, 0xc000 | 160, 256);
+        let f = snapshot(&m);
+        prefix(&f, 0, 80, 64, 58);
+        prefix(&f, 0, 160, 64, 0);
+        assert_eq!(finish(&mut m).pixels(), f.pixels());
+    }
+}
+
+#[test]
+fn small_sprites_have_no_fixed_thirty_two_object_limit() {
+    for (free, count, visible) in [(false, 127, 8), (true, 119, 2), (true, 120, 0)] {
+        let mut m = scene(free);
+        for i in 0..count {
+            obj(&mut m, i, 0, 0, 0);
+        }
+        obj(&mut m, count, 0, 80, 256);
+        prefix(&snapshot(&m), 0, 80, 8, visible);
+    }
+}
+
+#[test]
+fn affine_setup_and_double_size_use_drawing_canvas_width() {
+    for free in [false, true] {
+        for doubled in [false, true] {
+            let mut m = scene(free);
+            let count = if free { 14 } else { 18 };
+            fillers(&mut m, count); // 58 cycles remain.
+                                    // Zero matrices sample the opaque center for the entire canvas.
+            obj(&mut m, count, 0x100, 80, 256); // 10+2*8 = 26.
+            obj(
+                &mut m,
+                count + 1,
+                if doubled { 0x300 } else { 0x100 },
+                if doubled { 112 } else { 0x4000 | 112 },
+                256,
+            ); // 32 cycles remain: setup plus 11 canvas pixels.
+            let f = snapshot(&m);
+            prefix(&f, 0, 80, 8, 8);
+            prefix(&f, 0, 112, 16, 11);
+        }
+    }
+}
+
+#[test]
+fn unfinished_affine_setup_does_not_refund_work_to_regular_objects() {
+    for free in [false, true] {
+        let mut m = scene(free);
+        let count = if free { 14 } else { 18 };
+        fillers(&mut m, count);
+        obj(&mut m, count, 0, 0x8000, 0); // 32 cycles.
+        obj(&mut m, count + 1, 0, 0x4000, 0); // 16 cycles.
+        obj(&mut m, count + 2, 0x100, 80, 256); // Only setup fits.
+        obj(&mut m, count + 3, 0, 96, 256);
+        let f = snapshot(&m);
+        prefix(&f, 0, 80, 8, 0);
+        prefix(&f, 0, 96, 8, 0);
+    }
+}
+
+#[test]
+fn disabled_off_row_and_fully_offscreen_entries_cost_inspection_only() {
+    for free in [false, true] {
+        for (a, b) in [(0x200, 0), (80, 0xc000), (0, 0xc000 | 240)] {
+            let mut m = scene(free);
+            for i in 0..100 {
+                obj(&mut m, i, a, b, 0);
+            }
+            let count = if free { 11 } else { 15 };
+            for i in 100..100 + count {
+                obj(&mut m, i, 0, 0xc000, 0);
+            }
+            obj(&mut m, 100 + count, 0, 0xc000 | 80, 256);
+            prefix(&snapshot(&m), 0, 80, 64, 50);
+        }
+    }
+}
+
+#[test]
+fn partially_clipped_objects_charge_full_canvas_in_this_nominal_model() {
+    let mut m = scene(false);
+    fillers(&mut m, 18);
+    obj(&mut m, 18, 0, 0xc000 | 508, 256); // Left four pixels are clipped.
+    prefix(&snapshot(&m), 0, 0, 60, 54);
+
+    let mut m = scene(false);
+    fillers(&mut m, 17);
+    obj(&mut m, 17, 0, 0xc000 | 238, 256); // Only two pixels are on screen.
+    obj(&mut m, 18, 0, 0xc000 | 80, 256);
+    let f = snapshot(&m);
+    prefix(&f, 0, 238, 2, 2);
+    prefix(&f, 0, 80, 64, 58);
+}
+
+#[test]
+fn transparent_semitransparent_and_disabled_obj_window_pixels_still_cost_work() {
+    for a in [0, 0x400, 0x800] {
+        let mut m = scene(false);
+        for i in 0..18 {
+            obj(&mut m, i, a, 0xc000, if a == 0x800 { 128 } else { 0 });
+        }
+        obj(&mut m, 18, 0, 0xc000 | 80, 256);
+        prefix(&snapshot(&m), 0, 80, 64, 58);
+    }
+}
+
+#[test]
+fn oam_order_not_display_priority_allocates_work_even_for_occluded_objects() {
+    let mut m = scene(false);
+    for i in 0..18 {
+        obj(&mut m, i, 0, 0xc000 | 80, 128 | 0xc00);
+    }
+    obj(&mut m, 18, 0, 0xc000 | 80, 256);
+    let f = snapshot(&m);
+    for x in 80..144 {
+        assert_eq!(at(&f, x, 0), rgb(if x < 138 { 0x3e0 } else { 31 }));
+    }
+}
+
+#[test]
+fn large_affine_and_out_of_texture_samples_use_full_canvas_cost() {
+    for doubled in [false, true] {
+        let mut m = scene(false);
+        let count = if doubled { 4 } else { 8 };
+        for i in 0..count {
+            obj(&mut m, i, if doubled { 0x300 } else { 0x100 }, 0xc000, 0);
+        }
+        // Most samples are outside the texture. This does not refund work.
+        m.write16(OAM + 6, 0x7fff).unwrap();
+        m.write16(OAM + 30, 0x7fff).unwrap();
+        obj(&mut m, count, 0, 0xc000 | 16, 256);
+        obj(&mut m, count + 1, 0, 0xc000 | 80, 256);
+        obj(&mut m, count + 2, 0, 0xc000 | 144, 256);
+        let f = snapshot(&m);
+        prefix(&f, 0, 16, 64, 64);
+        prefix(&f, 0, 80, 64, if doubled { 64 } else { 42 });
+        prefix(&f, 0, 144, 64, if doubled { 18 } else { 0 });
+    }
+}
+
+#[test]
+fn bitmap_restricted_tiles_do_not_refund_sprite_work() {
+    let mut m = scene(false);
+    m.write16(DISPCNT, 0x1043).unwrap();
+    fillers(&mut m, 18); // Tile zero is unavailable to OBJ in bitmap modes.
+    obj(&mut m, 18, 0, 0xc000 | 80, 512);
+    prefix(&snapshot(&m), 0, 80, 64, 58);
+}
+
+#[test]
+fn each_row_gets_an_independent_allowance() {
+    let mut m = scene(false);
+    for i in 0..18 {
+        obj(&mut m, i, 1, 0xc000, 0);
+    }
+    obj(&mut m, 18, 0, 0xc000 | 80, 256);
+    let f = snapshot(&m);
+    prefix(&f, 0, 80, 64, 64);
+    prefix(&f, 1, 80, 64, 58);
+    assert_eq!(finish(&mut m).pixels(), f.pixels());
+}
+
+#[test]
+fn hblank_free_is_sampled_at_preparation_not_composition() {
+    for elapsed in [39, 40, 41] {
+        let mut m = scene(false);
+        fillers(&mut m, 18);
+        obj(&mut m, 18, 0, 0xc000 | 80, 256);
+        m.advance_cycles(elapsed);
+        m.write16(DISPCNT, 0x1060).unwrap();
+        prefix(&snapshot(&m), 0, 80, 64, 0);
+        let f = finish(&mut m);
+        prefix(&f, 0, 80, 64, 58);
+        prefix(&f, 1, 80, 64, if elapsed < 40 { 0 } else { 58 });
+        prefix(&f, 2, 80, 64, 0);
+    }
+}
+
+#[test]
+fn hblank_dma_updates_allowance_for_the_following_preparation() {
+    let mut m = scene(false);
+    fillers(&mut m, 18);
+    obj(&mut m, 18, 0, 0xc000 | 80, 256);
+    for row in 0..160 {
+        m.write16(
+            0x0200_0000 + row * 2,
+            if row % 2 == 0 { 0x1060 } else { 0x1040 },
+        )
+        .unwrap();
+    }
+    m.write32(DMA_BASE, 0x0200_0000).unwrap();
+    m.write32(DMA_BASE + 4, DISPCNT).unwrap();
+    m.write32(DMA_BASE + 8, 0xa240_0001).unwrap();
+    m.write8(HALTCNT, 0).unwrap();
+    let mut machine = Machine::new(Cpu::new(ROM_START), m);
+    machine.run_until_vblank(1000).unwrap();
+    let mut f = Framebuffer::default();
+    assert!(machine.memory().present_frame(&mut f).unwrap());
+    for y in 0..64 {
+        prefix(&f, y, 80, 64, if y < 2 || y % 2 == 1 { 58 } else { 0 });
+    }
+    assert!(machine.halted());
+}
+
+#[test]
+fn exhausted_work_preserves_development_diagnostics_and_output_atomicity() {
+    let mut m = scene(false);
+    fillers(&mut m, 19);
+    obj(&mut m, 127, 0xc00, 0, 0);
+    let mut f = Framebuffer::default();
+    f.clear(0x3e0);
+    assert!(matches!(
+        m.render_frame(&mut f),
+        Err(VideoError::UnsupportedObject { index: 127, .. })
+    ));
+    assert!(f.pixels().iter().all(|&p| p == rgb(0x3e0)));
+    m.advance_cycles(VB);
+    assert!(matches!(
+        m.present_frame(&mut f),
+        Err(VideoError::UnsupportedObject { index: 127, .. })
+    ));
+    assert!(f.pixels().iter().all(|&p| p == rgb(0x3e0)));
+    for control in [0x40, 0x10c0] {
+        m.write16(DISPCNT, control).unwrap();
+        m.render_frame(&mut f).unwrap();
+    }
+}