#include <stdio.h>
#include <stdlib.h>
+#include <stdint.h>
#include <string.h>
+#include <assert.h>
#include "common.h"
#ifndef NEON_BUILD
#endif
#include "psx_gpu_simd.h"
+#if 0
+void dump_r_d(const char *name, void *dump);
+void dump_r_q(const char *name, void *dump);
+#define dumprd(n) dump_r_d(#n, n.e)
+#define dumprq(n) dump_r_q(#n, n.e)
+#endif
+
u32 span_pixels = 0;
u32 span_pixel_blocks = 0;
u32 spans = 0;
printf("mismatch on %s %s: %x vs %x\n", #_a, #_b, _a, _b) \
-#ifndef NDEBUG
-#define setup_spans_debug_check(span_edge_data_element) \
-{ \
- u32 _num_spans = &span_edge_data_element - psx_gpu->span_edge_data; \
- if (_num_spans > MAX_SPANS) \
- *(int *)0 = 1; \
- if (_num_spans < psx_gpu->num_spans) \
- { \
- if(span_edge_data_element.num_blocks > MAX_BLOCKS_PER_ROW) \
- *(int *)0 = 1; \
- if(span_edge_data_element.y > 2048) \
- *(int *)0 = 1; \
- } \
-} \
-
+#if !defined(NEON_BUILD) && !defined(NDEBUG)
+static void setup_spans_debug_check(psx_gpu_struct *psx_gpu,
+ edge_data_struct *span_edge_data_element)
+{
+ u32 _num_spans = span_edge_data_element - psx_gpu->span_edge_data;
+ if (_num_spans > MAX_SPANS)
+ *(volatile int *)0 = 1;
+ if (_num_spans < psx_gpu->num_spans)
+ {
+ if(span_edge_data_element->num_blocks > MAX_BLOCKS_PER_ROW)
+ *(volatile int *)0 = 2;
+ if(span_edge_data_element->y >= 2048)
+ *(volatile int *)0 = 3;
+ }
+}
#else
-#define setup_spans_debug_check(span_edge_data_element) \
-
+#define setup_spans_debug_check(psx_gpu, span_edge_data_element)
#endif
#define setup_spans_prologue_alternate_yes() \
vec_2x64s alternate_x; \
vec_2x64s alternate_dx_dy; \
vec_4x32s alternate_x_32; \
- vec_2x32s alternate_x_16; \
+ vec_4x16u alternate_x_16; \
\
vec_4x16u alternate_select; \
vec_4x16s y_mid_point; \
span_b_offset = psx_gpu->span_b_offset; \
\
vec_8x16u c_0x0001; \
+ vec_4x16u c_max_blocks_per_row; \
\
dup_8x16b(c_0x0001, 0x0001); \
dup_8x16b(left_edge, psx_gpu->viewport_start_x); \
dup_4x16b(c_0x04, 0x04); \
dup_4x16b(c_0x07, 0x07); \
dup_4x16b(c_0xFFFE, 0xFFFE); \
+ dup_4x16b(c_max_blocks_per_row, MAX_BLOCKS_PER_ROW); \
#define compute_edge_delta_x2() \
and_4x16b(span_shift, left_right_x_16.high, c_0x07); \
shl_variable_4x16b(span_shift, c_0xFFFE, span_shift); \
shr_4x16b(left_right_x_16.high, left_right_x_16.high, 3); \
+ min_4x16b(left_right_x_16.high, left_right_x_16.high, c_max_blocks_per_row); \
\
u32 i; \
for(i = 0; i < 4; i++) \
span_edge_data[i].num_blocks = left_right_x_16.high.e[i]; \
span_edge_data[i].right_mask = span_shift.e[i]; \
span_edge_data[i].y = y_x4.e[i]; \
- setup_spans_debug_check(span_edge_data[i]); \
+ setup_spans_debug_check(psx_gpu, &span_edge_data[i]); \
} \
\
span_edge_data += 4; \
\
setup_spans_prologue_b(); \
\
- if(height > 0) \
+ if (height > 512) \
+ height = 512; \
+ if (height > 0) \
{ \
y_x4.e[0] = y_a; \
y_x4.e[1] = y_a + 1; \
\
setup_spans_prologue_b(); \
\
- if(height > 0) \
+ if (height > 512) \
+ height = 512; \
+ if (height > 0) \
{ \
y_x4.e[0] = y_a; \
y_x4.e[1] = y_a - 1; \
setup_spans_prologue_b();
- if(height_minor_a > 0)
+ if (height_minor_a > 512)
+ height_minor_a = 512;
+ if (height_minor_a > 0)
{
y_x4.e[0] = y_a;
y_x4.e[1] = y_a - 1;
setup_spans_clip(increment, no);
}
- if(height_minor_b > 0)
+ if (height_minor_b > 512)
+ height_minor_b = 512;
+ if (height_minor_b > 0)
{
y_x4.e[0] = y_a;
y_x4.e[1] = y_a + 1;
}
}
}
+ assert(psx_gpu->span_edge_data[0].y < 1024u);
u32 render_state = flags &
(RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND |
#define setup_sprite_tile_fetch_texel_block_8bpp(offset) \
- texture_block_ptr = psx_gpu->texture_page_ptr + \
+ texture_block_ptr = (u8 *)psx_gpu->texture_page_ptr + \
((texture_offset + offset) & texture_mask); \
\
load_64b(texels, texture_block_ptr) \
#define setup_sprite_tile_half_8bpp(edge) \
{ \
- setup_sprite_tile_add_blocks(sub_tile_height * 2); \
+ setup_sprite_tile_add_blocks(sub_tile_height); \
\
while(sub_tile_height) \
{ \
u32 num_blocks = psx_gpu->num_blocks; \
block_struct *block = psx_gpu->blocks + num_blocks; \
\
- u16 *texture_block_ptr; \
+ u8 *texture_block_ptr; \
vec_8x8u texels; \
\
setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
#ifndef NEON_BUILD
-void setup_sprite_untextured(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u,
+void setup_sprite_untextured_512(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u,
s32 v, s32 width, s32 height, u32 color)
{
- if((psx_gpu->render_state & (RENDER_STATE_MASK_EVALUATE |
- RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND)) == 0 &&
- (psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) == 0)
- {
- setup_sprite_untextured_simple(psx_gpu, x, y, u, v, width, height, color);
- return;
- }
-
u32 right_width = ((width - 1) & 0x7) + 1;
u32 right_mask_bits = (0xFF << right_width);
u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + x;
#endif
-void setup_sprite_untextured_simple(psx_gpu_struct *psx_gpu, s32 x, s32 y,
- s32 u, s32 v, s32 width, s32 height, u32 color)
+static void __attribute__((noinline))
+setup_sprite_untextured_simple(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u,
+ s32 v, s32 width, s32 height, u32 color)
{
u32 r = color & 0xFF;
u32 g = (color >> 8) & 0xFF;
u32 num_width;
- if(psx_gpu->num_blocks > MAX_BLOCKS)
+ if(psx_gpu->num_blocks)
{
flush_render_block_buffer(psx_gpu);
}
num_width = width;
vram_ptr = (void *)vram_ptr16;
- if((long)vram_ptr16 & 2)
+ if((uintptr_t)vram_ptr16 & 2)
{
*vram_ptr16 = color_32bpp;
vram_ptr = (void *)(vram_ptr16 + 1);
}
}
+void setup_sprite_untextured_512(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u,
+ s32 v, s32 width, s32 height, u32 color);
+
+void setup_sprite_untextured(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u,
+ s32 v, s32 width, s32 height, u32 color)
+{
+ if((psx_gpu->render_state & (RENDER_STATE_MASK_EVALUATE |
+ RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND)) == 0 &&
+ (psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) == 0)
+ {
+ setup_sprite_untextured_simple(psx_gpu, x, y, u, v, width, height, color);
+ return;
+ }
+
+ while (width > 0)
+ {
+ s32 w1 = width > 512 ? 512 : width;
+ setup_sprite_untextured_512(psx_gpu, x, y, 0, 0, w1, height, color);
+ x += 512;
+ width -= 512;
+ }
+}
+
#define setup_sprite_blocks_switch_textured(texture_mode) \
setup_sprite_##texture_mode \
psx_gpu->primitive_type = PRIMITIVE_TYPE_UNKNOWN;
- psx_gpu->enhancement_x_threshold = 256;
+ psx_gpu->saved_hres = 256;
}
u64 get_us(void)