#include <stdio.h>
#include "common.h"
+#include "../../gpulib/gpu_timing.h"
#ifndef command_lengths
const u8 command_lengths[256] =
#define SET_Ex(r, v)
#endif
-u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size, u32 *last_command)
+u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
+ s32 *cpu_cycles_sum_out, s32 *cpu_cycles_last, u32 *last_command)
{
vertex_struct vertexes[4] __attribute__((aligned(16))) = {};
u32 current_command = 0, command_length;
+ u32 cpu_cycles_sum = 0, cpu_cycles = *cpu_cycles_last;
u32 *list_start = list;
u32 *list_end = list + (size / 4);
for(; list < list_end; list += 1 + command_length)
{
- s16 *list_s16 = (void *)list;
- current_command = *list >> 24;
- command_length = command_lengths[current_command];
- if (list + 1 + command_length > list_end) {
- current_command = (u32)-1;
- break;
- }
-
- switch(current_command)
- {
- case 0x00:
- break;
-
- case 0x02:
+ s16 *list_s16 = (void *)list;
+ current_command = *list >> 24;
+ command_length = command_lengths[current_command];
+ if (list + 1 + command_length > list_end) {
+ current_command = (u32)-1;
+ break;
+ }
+
+ switch(current_command)
+ {
+ case 0x00:
+ break;
+
+ case 0x02:
{
u32 x = list_s16[2] & 0x3FF;
u32 y = list_s16[3] & 0x1FF;
u32 color = list[0] & 0xFFFFFF;
do_fill(psx_gpu, x, y, width, height, color);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_fill(width, height));
+ break;
}
-
- case 0x20 ... 0x23:
+
+ case 0x20 ... 0x23:
{
set_triangle_color(psx_gpu, list[0] & 0xFFFFFF);
get_vertex_data_xy(2, 6);
render_triangle(psx_gpu, vertexes, current_command);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base());
+ break;
}
- case 0x24 ... 0x27:
+ case 0x24 ... 0x27:
{
set_clut(psx_gpu, list_s16[5]);
set_texture(psx_gpu, list_s16[9]);
get_vertex_data_xy_uv(2, 10);
render_triangle(psx_gpu, vertexes, current_command);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_t());
+ break;
}
- case 0x28 ... 0x2B:
+ case 0x28 ... 0x2B:
{
set_triangle_color(psx_gpu, list[0] & 0xFFFFFF);
render_triangle(psx_gpu, vertexes, current_command);
render_triangle(psx_gpu, &(vertexes[1]), current_command);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base());
+ break;
}
- case 0x2C ... 0x2F:
+ case 0x2C ... 0x2F:
{
set_clut(psx_gpu, list_s16[5]);
set_texture(psx_gpu, list_s16[9]);
render_triangle(psx_gpu, vertexes, current_command);
render_triangle(psx_gpu, &(vertexes[1]), current_command);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_t());
+ break;
}
- case 0x30 ... 0x33:
+ case 0x30 ... 0x33:
{
get_vertex_data_xy_rgb(0, 0);
get_vertex_data_xy_rgb(1, 4);
get_vertex_data_xy_rgb(2, 8);
render_triangle(psx_gpu, vertexes, current_command);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_g());
+ break;
}
- case 0x34:
- case 0x35:
- case 0x36:
- case 0x37:
+ case 0x34 ... 0x37:
{
set_clut(psx_gpu, list_s16[5]);
set_texture(psx_gpu, list_s16[11]);
get_vertex_data_xy_uv_rgb(2, 12);
render_triangle(psx_gpu, vertexes, current_command);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_gt());
+ break;
}
- case 0x38:
- case 0x39:
- case 0x3A:
- case 0x3B:
+ case 0x38 ... 0x3B:
{
get_vertex_data_xy_rgb(0, 0);
get_vertex_data_xy_rgb(1, 4);
render_triangle(psx_gpu, vertexes, current_command);
render_triangle(psx_gpu, &(vertexes[1]), current_command);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_g());
+ break;
}
- case 0x3C:
- case 0x3D:
- case 0x3E:
- case 0x3F:
+ case 0x3C ... 0x3F:
{
set_clut(psx_gpu, list_s16[5]);
set_texture(psx_gpu, list_s16[11]);
render_triangle(psx_gpu, vertexes, current_command);
render_triangle(psx_gpu, &(vertexes[1]), current_command);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_gt());
+ break;
}
- case 0x40 ... 0x47:
+ case 0x40 ... 0x47:
{
vertexes[0].x = list_s16[2] + psx_gpu->offset_x;
vertexes[0].y = list_s16[3] + psx_gpu->offset_y;
vertexes[1].y = list_s16[5] + psx_gpu->offset_y;
render_line(psx_gpu, vertexes, current_command, list[0], 0);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
+ break;
}
- case 0x48 ... 0x4F:
+ case 0x48 ... 0x4F:
{
u32 num_vertexes = 1;
u32 *list_position = &(list[2]);
vertexes[1].y = (xy >> 16) + psx_gpu->offset_y;
render_line(psx_gpu, vertexes, current_command, list[0], 0);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
list_position++;
num_vertexes++;
break;
}
- case 0x50 ... 0x57:
+ case 0x50 ... 0x57:
{
vertexes[0].r = list[0] & 0xFF;
vertexes[0].g = (list[0] >> 8) & 0xFF;
vertexes[1].y = list_s16[7] + psx_gpu->offset_y;
render_line(psx_gpu, vertexes, current_command, 0, 0);
- break;
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
+ break;
}
case 0x58 ... 0x5F:
vertexes[1].y = (xy >> 16) + psx_gpu->offset_y;
render_line(psx_gpu, vertexes, current_command, 0, 0);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
list_position += 2;
num_vertexes++;
break;
}
- case 0x60 ... 0x63:
+ case 0x60 ... 0x63:
{
u32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
u32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
- u32 width = list_s16[4] & 0x3FF;
- u32 height = list_s16[5] & 0x1FF;
+ s32 width = list_s16[4] & 0x3FF;
+ s32 height = list_s16[5] & 0x1FF;
- render_sprite(psx_gpu, x, y, 0, 0, width, height, current_command, list[0]);
- break;
+ render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
+ current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
+ break;
}
- case 0x64 ... 0x67:
+ case 0x64 ... 0x67:
{
u32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
u32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
u32 uv = list_s16[4];
- u32 width = list_s16[6] & 0x3FF;
- u32 height = list_s16[7] & 0x1FF;
+ s32 width = list_s16[6] & 0x3FF;
+ s32 height = list_s16[7] & 0x1FF;
set_clut(psx_gpu, list_s16[5]);
- render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF, width, height,
- current_command, list[0]);
- break;
+ render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF,
+ &width, &height, current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
+ break;
}
- case 0x68:
- case 0x69:
- case 0x6A:
- case 0x6B:
+ case 0x68 ... 0x6B:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
+ s32 width = 1, height = 1;
- render_sprite(psx_gpu, x, y, 0, 0, 1, 1, current_command, list[0]);
- break;
+ render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
+ current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(1, 1));
+ break;
}
- case 0x70:
- case 0x71:
- case 0x72:
- case 0x73:
+ case 0x70 ... 0x73:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
+ s32 width = 8, height = 8;
- render_sprite(psx_gpu, x, y, 0, 0, 8, 8, current_command, list[0]);
- break;
+ render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
+ current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
+ break;
}
- case 0x74:
- case 0x75:
- case 0x76:
- case 0x77:
+ case 0x74 ... 0x77:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
u32 uv = list_s16[4];
+ s32 width = 8, height = 8;
set_clut(psx_gpu, list_s16[5]);
- render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF, 8, 8,
- current_command, list[0]);
- break;
+ render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF,
+ &width, &height, current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
+ break;
}
- case 0x78:
- case 0x79:
- case 0x7A:
- case 0x7B:
+ case 0x78 ... 0x7B:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
+ s32 width = 16, height = 16;
- render_sprite(psx_gpu, x, y, 0, 0, 16, 16, current_command, list[0]);
- break;
+ render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
+ current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
+ break;
}
- case 0x7C:
- case 0x7D:
- case 0x7E:
- case 0x7F:
+ case 0x7C ... 0x7F:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
u32 uv = list_s16[4];
+ s32 width = 16, height = 16;
set_clut(psx_gpu, list_s16[5]);
- render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF, 16, 16,
- current_command, list[0]);
- break;
+ render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF,
+ &width, &height, current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
+ break;
}
- case 0x80: // vid -> vid
+#ifdef PCSX
+ case 0x1F: // irq?
+ case 0x80 ... 0x9F: // vid -> vid
+ case 0xA0 ... 0xBF: // sys -> vid
+ case 0xC0 ... 0xDF: // vid -> sys
+ goto breakloop;
+#else
+ case 0x80 ... 0x9F: // vid -> vid
{
u32 sx = list_s16[2] & 0x3FF;
u32 sy = list_s16[3] & 0x1FF;
break;
}
-#ifdef PCSX
- case 0xA0: // sys -> vid
- case 0xC0: // vid -> sys
- goto breakloop;
-#else
- case 0xA0: // sys -> vid
+ case 0xA0 ... 0xBF: // sys -> vid
{
u32 load_x = list_s16[2] & 0x3FF;
u32 load_y = list_s16[3] & 0x1FF;
render_block_copy(psx_gpu, (u16 *)&(list_s16[6]), load_x, load_y,
load_width, load_height, load_width);
- break;
+ break;
}
- case 0xC0: // vid -> sys
- break;
+ case 0xC0 ... 0xDF: // vid -> sys
+ break;
#endif
- case 0xE1:
+ case 0xE1:
set_texture(psx_gpu, list[0]);
if(list[0] & (1 << 9))
psx_gpu->render_state_base &= ~RENDER_STATE_DITHER;
psx_gpu->display_area_draw_enable = (list[0] >> 10) & 0x1;
- SET_Ex(1, list[0]);
- break;
+ SET_Ex(1, list[0]);
+ break;
- case 0xE2:
+ case 0xE2:
{
// TODO: Clean
u32 texture_window_settings = list[0];
psx_gpu->offset_x = offset_x >> 21;
psx_gpu->offset_y = offset_y >> 21;
- SET_Ex(5, list[0]);
- break;
- }
+ SET_Ex(5, list[0]);
+ break;
+ }
- case 0xE6:
+ case 0xE6:
{
u32 mask_settings = list[0];
u16 mask_msb = mask_settings << 15;
psx_gpu->mask_msb = mask_msb;
}
- SET_Ex(6, list[0]);
- break;
+ SET_Ex(6, list[0]);
+ break;
}
- default:
- break;
- }
+ default:
+ break;
+ }
}
breakloop:
- if (last_command != NULL)
- *last_command = current_command;
+ *cpu_cycles_sum_out += cpu_cycles_sum;
+ *cpu_cycles_last = cpu_cycles;
+ *last_command = current_command;
return list - list_start;
}
#ifdef PCSX
-#define ENH_BUF_TABLE_STEP (1024 / sizeof(psx_gpu->enhancement_buf_by_x16))
-
-static int is_new_scanout(psx_gpu_struct *psx_gpu, int x)
+// this thing has become such a PITA, should just handle the 2048 width really
+static void update_enhancement_buf_scanouts(psx_gpu_struct *psx_gpu,
+ int x, int y, int w, int h)
{
- int i, scanout_x;
- for (i = 0; i < ARRAY_SIZE(psx_gpu->enhancement_scanout_x); i++)
- {
- scanout_x = psx_gpu->enhancement_scanout_x[i];
- if (x <= scanout_x && scanout_x < x + ENH_BUF_TABLE_STEP)
- {
- if (x != scanout_x)
- log_anomaly("unaligned scanout x: %d,%d\n", scanout_x, x);
- return 1;
- }
+ int max_bufs = ARRAY_SIZE(psx_gpu->enhancement_scanouts);
+ struct psx_gpu_scanout *s;
+ int i, sel, right, bottom;
+ u32 tol_x = 48, tol_y = 16;
+ u32 intersection;
+
+ //w = (w + 15) & ~15;
+ psx_gpu->saved_hres = w;
+ assert(!(max_bufs & (max_bufs - 1)));
+ for (i = 0; i < max_bufs; i++) {
+ s = &psx_gpu->enhancement_scanouts[i];
+ if (s->x == x && s->y == y && w - s->w <= tol_x && h - s->h <= tol_y)
+ return;
}
- return 0;
-}
-
-static void update_enhancement_buf_table_from_hres(psx_gpu_struct *psx_gpu)
-{
- u32 b, x;
- b = 0;
- psx_gpu->enhancement_buf_by_x16[0] = b;
- psx_gpu->enhancement_buf_start[0] = 0;
- for (x = 1; x < sizeof(psx_gpu->enhancement_buf_by_x16); x++)
- {
- if (b < 3 && is_new_scanout(psx_gpu, x * ENH_BUF_TABLE_STEP)) {
- b++;
- psx_gpu->enhancement_buf_start[b] = x * ENH_BUF_TABLE_STEP;
+ // evict any scanout that intersects
+ right = x + w;
+ bottom = y + h;
+ for (i = 0, sel = -1; i < max_bufs; i++) {
+ s = &psx_gpu->enhancement_scanouts[i];
+ if (s->x >= right) continue;
+ if (s->x + s->w <= x) continue;
+ if (s->y >= bottom) continue;
+ if (s->y + s->h <= y) continue;
+ // ... but allow upto 16 pixels intersection that some games do
+ if ((intersection = s->x + s->w - x) - 1u <= tol_x) {
+ s->w -= intersection;
+ continue;
}
-
- psx_gpu->enhancement_buf_by_x16[x] = b;
+ if ((intersection = s->y + s->h - y) - 1u <= tol_y) {
+ s->h -= intersection;
+ continue;
+ }
+ //printf("%4d%4d%4dx%d evicted\n", s->x, s->y, s->w, s->h);
+ s->w = 0;
+ sel = i;
+ break;
}
-#if 0
- printf("buf_by_x16:\n");
- for (b = 0; b < 3; b++) {
- int first = -1, count = 0;
- for (x = 0; x < sizeof(psx_gpu->enhancement_buf_by_x16); x++) {
- if (psx_gpu->enhancement_buf_by_x16[x] == b) {
- if (first < 0) first = x;
- count++;
+ if (sel >= 0) {
+ // 2nd intersection check
+ for (i = 0; i < max_bufs; i++) {
+ s = &psx_gpu->enhancement_scanouts[i];
+ if (!s->w)
+ continue;
+ if ((intersection = right - s->x) - 1u <= tol_x) {
+ w -= intersection;
+ break;
+ }
+ if ((intersection = bottom - s->y) - 1u <= tol_y) {
+ h -= intersection;
+ break;
}
}
- if (count) {
- assert(first * ENH_BUF_TABLE_STEP == psx_gpu->enhancement_buf_start[b]);
- printf("%d: %3zd-%zd\n", b, first * ENH_BUF_TABLE_STEP,
- (first + count) * ENH_BUF_TABLE_STEP);
- }
+ }
+ else
+ sel = psx_gpu->enhancement_scanout_eselect++;
+ psx_gpu->enhancement_scanout_eselect &= max_bufs - 1;
+ s = &psx_gpu->enhancement_scanouts[sel];
+ s->x = x;
+ s->y = y;
+ s->w = w;
+ s->h = h;
+
+ sync_enhancement_buffers(x, y, w, h);
+#if 0
+ printf("scanouts:\n");
+ for (i = 0; i < ARRAY_SIZE(psx_gpu->enhancement_scanouts); i++) {
+ s = &psx_gpu->enhancement_scanouts[i];
+ if (s->w)
+ printf("%4d%4d%4dx%d\n", s->x, s->y, s->w, s->h);
}
#endif
}
-static void update_enhancement_buf_table_from_x(psx_gpu_struct *psx_gpu,
- u32 x0, u32 len)
+static int select_enhancement_buf_index(psx_gpu_struct *psx_gpu, s32 x, s32 y)
{
-#if 0
- u32 x, b;
+ int i;
+ for (i = 0; i < ARRAY_SIZE(psx_gpu->enhancement_scanouts); i++) {
+ const struct psx_gpu_scanout *s = &psx_gpu->enhancement_scanouts[i];
+ if (s->x <= x && x < s->x + s->w &&
+ s->y <= y && y < s->y + s->h)
+ return i;
+ }
+ return -1;
+}
- for (x = x0, b = 0; x >= len; b++)
- x -= len;
- if (b > 3)
- b = 3;
+#define select_enhancement_buf_by_index(psx_gpu_, i_) \
+ ((psx_gpu_)->enhancement_buf_ptr + ((i_) << 20))
- memset(psx_gpu->enhancement_buf_by_x16 + x0 / ENH_BUF_TABLE_STEP,
- b, (len + ENH_BUF_TABLE_STEP - 1) / ENH_BUF_TABLE_STEP);
-#endif
+static void *select_enhancement_buf_ptr(psx_gpu_struct *psx_gpu, s32 x, s32 y)
+{
+ int i = select_enhancement_buf_index(psx_gpu, x, y);
+ return i >= 0 ? select_enhancement_buf_by_index(psx_gpu, i) : NULL;
}
-#define select_enhancement_buf(psx_gpu) \
- psx_gpu->enhancement_current_buf_ptr = \
- select_enhancement_buf_ptr(psx_gpu, psx_gpu->saved_viewport_start_x)
+static void select_enhancement_buf(psx_gpu_struct *psx_gpu)
+{
+ s32 x = psx_gpu->saved_viewport_start_x + 16;
+ s32 y = psx_gpu->saved_viewport_start_y + 16;
+ psx_gpu->enhancement_current_buf_ptr = select_enhancement_buf_ptr(psx_gpu, x, y);
+}
#define enhancement_disable() { \
psx_gpu->vram_out_ptr = psx_gpu->vram_ptr; \
psx_gpu->uvrgb_phase = 0x8000; \
}
-#define enhancement_enable() { \
- psx_gpu->vram_out_ptr = psx_gpu->enhancement_current_buf_ptr; \
- psx_gpu->viewport_start_x = psx_gpu->saved_viewport_start_x * 2; \
- psx_gpu->viewport_start_y = psx_gpu->saved_viewport_start_y * 2; \
- psx_gpu->viewport_end_x = psx_gpu->saved_viewport_end_x * 2 + 1; \
- psx_gpu->viewport_end_y = psx_gpu->saved_viewport_end_y * 2 + 1; \
- psx_gpu->uvrgb_phase = 0x7fff; \
+static int enhancement_enable(psx_gpu_struct *psx_gpu)
+{
+ if (!psx_gpu->enhancement_current_buf_ptr)
+ return 0;
+ psx_gpu->vram_out_ptr = psx_gpu->enhancement_current_buf_ptr;
+ psx_gpu->viewport_start_x = psx_gpu->saved_viewport_start_x * 2;
+ psx_gpu->viewport_start_y = psx_gpu->saved_viewport_start_y * 2;
+ psx_gpu->viewport_end_x = psx_gpu->saved_viewport_end_x * 2 + 1;
+ psx_gpu->viewport_end_y = psx_gpu->saved_viewport_end_y * 2 + 1;
+ if (psx_gpu->viewport_end_x - psx_gpu->viewport_start_x + 1 > 1024)
+ psx_gpu->viewport_end_x = psx_gpu->viewport_start_x + 1023;
+ psx_gpu->uvrgb_phase = 0x7fff;
+ return 1;
}
#define shift_vertices3(v) { \
}
#endif
-static int disable_main_render;
-
-static int check_enhanced_range(psx_gpu_struct *psx_gpu, int x, int x_end)
+// simple check for a case where no clipping is used
+// - now handled by adjusting the viewport
+static int check_enhanced_range(psx_gpu_struct *psx_gpu, int x, int y)
{
- // simple reject to avoid oveflowing the 1024 width
- // (assume some offscreen render-to-texture thing)
- if (x >= (int)(psx_gpu->saved_viewport_start_x + 512))
- return 0;
-
return 1;
}
vertex_ptrs[i].v = new;
}
+// this sometimes does more harm than good, like in PE2
static void uv_hack(vertex_struct *vertex_ptrs, int vertex_count)
{
int i, u[4], v[4];
if (!prepare_triangle(psx_gpu, vertexes, vertex_ptrs))
return;
- if (!disable_main_render)
+ if (!psx_gpu->hack_disable_main)
render_triangle_p(psx_gpu, vertex_ptrs, current_command);
if (!check_enhanced_range(psx_gpu, vertex_ptrs[0]->x, vertex_ptrs[2]->x))
return;
- enhancement_enable();
+ if (!enhancement_enable(psx_gpu))
+ return;
+
shift_vertices3(vertex_ptrs);
shift_triangle_area();
render_triangle_p(psx_gpu, vertex_ptrs, current_command);
#endif
u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
- u32 *last_command)
+ s32 *cpu_cycles_sum_out, s32 *cpu_cycles_last, u32 *last_command)
{
vertex_struct vertexes[4] __attribute__((aligned(16))) = {};
u32 current_command = 0, command_length;
+ u32 cpu_cycles_sum = 0, cpu_cycles = *cpu_cycles_last;
u32 *list_start = list;
u32 *list_end = list + (size / 4);
u32 width = list_s16[4] & 0x3FF;
u32 height = list_s16[5] & 0x1FF;
u32 color = list[0] & 0xFFFFFF;
- u32 i1, i2;
+ s32 i1, i2;
x &= ~0xF;
width = ((width + 0xF) & ~0xF);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_fill(width, height));
if (width == 0 || height == 0)
break;
do_fill(psx_gpu, x, y, width, height, color);
- i1 = select_enhancement_buf_index(psx_gpu, x);
- i2 = select_enhancement_buf_index(psx_gpu, x + width - 1);
- if (i1 != i2) {
+ i1 = select_enhancement_buf_index(psx_gpu, x, y);
+ i2 = select_enhancement_buf_index(psx_gpu, x + width - 1, y + height - 1);
+ if (i1 < 0 || i1 != i2) {
sync_enhancement_buffers(x, y, width, height);
break;
}
- if (x >= psx_gpu->enhancement_buf_start[i1] + psx_gpu->saved_hres)
- break;
- psx_gpu->vram_out_ptr = select_enhancement_buf_ptr(psx_gpu, x);
+ psx_gpu->vram_out_ptr = select_enhancement_buf_by_index(psx_gpu, i1);
x *= 2;
y *= 2;
width *= 2;
get_vertex_data_xy(2, 6);
do_triangle_enhanced(psx_gpu, vertexes, current_command);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base());
break;
}
get_vertex_data_xy_uv(2, 10);
do_triangle_enhanced(psx_gpu, vertexes, current_command);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_t());
break;
}
get_vertex_data_xy(3, 8);
do_quad_enhanced(psx_gpu, vertexes, current_command);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base());
break;
}
get_vertex_data_xy_uv(2, 10);
get_vertex_data_xy_uv(3, 14);
- uv_hack(vertexes, 4);
+ if (psx_gpu->hack_texture_adj)
+ uv_hack(vertexes, 4);
do_quad_enhanced(psx_gpu, vertexes, current_command);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_t());
break;
}
get_vertex_data_xy_rgb(2, 8);
do_triangle_enhanced(psx_gpu, vertexes, current_command);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_g());
break;
}
- case 0x34:
- case 0x35:
- case 0x36:
- case 0x37:
+ case 0x34 ... 0x37:
{
set_clut(psx_gpu, list_s16[5]);
set_texture(psx_gpu, list_s16[11]);
get_vertex_data_xy_uv_rgb(2, 12);
do_triangle_enhanced(psx_gpu, vertexes, current_command);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_gt());
break;
}
- case 0x38:
- case 0x39:
- case 0x3A:
- case 0x3B:
+ case 0x38 ... 0x3B:
{
get_vertex_data_xy_rgb(0, 0);
get_vertex_data_xy_rgb(1, 4);
get_vertex_data_xy_rgb(3, 12);
do_quad_enhanced(psx_gpu, vertexes, current_command);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_g());
break;
}
- case 0x3C:
- case 0x3D:
- case 0x3E:
- case 0x3F:
+ case 0x3C ... 0x3F:
{
set_clut(psx_gpu, list_s16[5]);
set_texture(psx_gpu, list_s16[11]);
get_vertex_data_xy_uv_rgb(2, 12);
get_vertex_data_xy_uv_rgb(3, 18);
- uv_hack(vertexes, 4);
+ if (psx_gpu->hack_texture_adj)
+ uv_hack(vertexes, 4);
do_quad_enhanced(psx_gpu, vertexes, current_command);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_gt());
break;
}
vertexes[1].y = list_s16[5] + psx_gpu->offset_y;
render_line(psx_gpu, vertexes, current_command, list[0], 0);
- enhancement_enable();
- render_line(psx_gpu, vertexes, current_command, list[0], 1);
+ if (enhancement_enable(psx_gpu))
+ render_line(psx_gpu, vertexes, current_command, list[0], 1);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
break;
}
enhancement_disable();
render_line(psx_gpu, vertexes, current_command, list[0], 0);
- enhancement_enable();
- render_line(psx_gpu, vertexes, current_command, list[0], 1);
+ if (enhancement_enable(psx_gpu))
+ render_line(psx_gpu, vertexes, current_command, list[0], 1);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
list_position++;
num_vertexes++;
vertexes[1].y = list_s16[7] + psx_gpu->offset_y;
render_line(psx_gpu, vertexes, current_command, 0, 0);
- enhancement_enable();
- render_line(psx_gpu, vertexes, current_command, 0, 1);
+ if (enhancement_enable(psx_gpu))
+ render_line(psx_gpu, vertexes, current_command, 0, 1);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
break;
}
enhancement_disable();
render_line(psx_gpu, vertexes, current_command, 0, 0);
- enhancement_enable();
- render_line(psx_gpu, vertexes, current_command, 0, 1);
+ if (enhancement_enable(psx_gpu))
+ render_line(psx_gpu, vertexes, current_command, 0, 1);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
list_position += 2;
num_vertexes++;
{
u32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
u32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
- u32 width = list_s16[4] & 0x3FF;
- u32 height = list_s16[5] & 0x1FF;
+ s32 width = list_s16[4] & 0x3FF;
+ s32 height = list_s16[5] & 0x1FF;
- render_sprite(psx_gpu, x, y, 0, 0, width, height, current_command, list[0]);
+ render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
+ current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
- if (check_enhanced_range(psx_gpu, x, x + width))
+ if (check_enhanced_range(psx_gpu, x, x + width)) {
+ width = list_s16[4] & 0x3FF;
+ height = list_s16[5] & 0x1FF;
do_sprite_enhanced(psx_gpu, x, y, 0, 0, width, height, list[0]);
+ }
break;
}
u32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
u8 u = list_s16[4];
u8 v = list_s16[4] >> 8;
- u32 width = list_s16[6] & 0x3FF;
- u32 height = list_s16[7] & 0x1FF;
+ s32 width = list_s16[6] & 0x3FF;
+ s32 height = list_s16[7] & 0x1FF;
set_clut(psx_gpu, list_s16[5]);
- render_sprite(psx_gpu, x, y, u, v, width, height,
- current_command, list[0]);
+ render_sprite(psx_gpu, x, y, u, v,
+ &width, &height, current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
- if (check_enhanced_range(psx_gpu, x, x + width))
+ if (check_enhanced_range(psx_gpu, x, x + width)) {
+ width = list_s16[6] & 0x3FF;
+ height = list_s16[7] & 0x1FF;
do_sprite_enhanced(psx_gpu, x, y, u, v, width, height, list[0]);
+ }
break;
}
- case 0x68:
- case 0x69:
- case 0x6A:
- case 0x6B:
+ case 0x68 ... 0x6B:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
+ s32 width = 1, height = 1;
- render_sprite(psx_gpu, x, y, 0, 0, 1, 1, current_command, list[0]);
+ render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
+ current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(1, 1));
if (check_enhanced_range(psx_gpu, x, x + 1))
do_sprite_enhanced(psx_gpu, x, y, 0, 0, 1, 1, list[0]);
break;
}
- case 0x70:
- case 0x71:
- case 0x72:
- case 0x73:
+ case 0x70 ... 0x73:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
+ s32 width = 8, height = 8;
- render_sprite(psx_gpu, x, y, 0, 0, 8, 8, current_command, list[0]);
+ render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
+ current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
if (check_enhanced_range(psx_gpu, x, x + 8))
do_sprite_enhanced(psx_gpu, x, y, 0, 0, 8, 8, list[0]);
break;
}
- case 0x74:
- case 0x75:
- case 0x76:
- case 0x77:
+ case 0x74 ... 0x77:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
u8 u = list_s16[4];
u8 v = list_s16[4] >> 8;
+ s32 width = 8, height = 8;
set_clut(psx_gpu, list_s16[5]);
- render_sprite(psx_gpu, x, y, u, v, 8, 8,
- current_command, list[0]);
+ render_sprite(psx_gpu, x, y, u, v,
+ &width, &height, current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
if (check_enhanced_range(psx_gpu, x, x + 8))
do_sprite_enhanced(psx_gpu, x, y, u, v, 8, 8, list[0]);
break;
}
- case 0x78:
- case 0x79:
- case 0x7A:
- case 0x7B:
+ case 0x78 ... 0x7B:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
+ s32 width = 16, height = 16;
- render_sprite(psx_gpu, x, y, 0, 0, 16, 16, current_command, list[0]);
+ render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
+ current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
if (check_enhanced_range(psx_gpu, x, x + 16))
do_sprite_enhanced(psx_gpu, x, y, 0, 0, 16, 16, list[0]);
break;
}
- case 0x7C:
- case 0x7D:
- case 0x7E:
- case 0x7F:
+ case 0x7C ... 0x7F:
{
s32 x = sign_extend_11bit(list_s16[2] + psx_gpu->offset_x);
s32 y = sign_extend_11bit(list_s16[3] + psx_gpu->offset_y);
u8 u = list_s16[4];
u8 v = list_s16[4] >> 8;
+ s32 width = 16, height = 16;
set_clut(psx_gpu, list_s16[5]);
- render_sprite(psx_gpu, x, y, u, v, 16, 16, current_command, list[0]);
+ render_sprite(psx_gpu, x, y, u, v,
+ &width, &height, current_command, list[0]);
+ gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
if (check_enhanced_range(psx_gpu, x, x + 16))
do_sprite_enhanced(psx_gpu, x, y, u, v, 16, 16, list[0]);
break;
}
-
- case 0x80: // vid -> vid
- {
- u32 sx = list_s16[2] & 0x3FF;
- u32 sy = list_s16[3] & 0x1FF;
- u32 dx = list_s16[4] & 0x3FF;
- u32 dy = list_s16[5] & 0x1FF;
- u32 w = ((list_s16[6] - 1) & 0x3FF) + 1;
- u32 h = ((list_s16[7] - 1) & 0x1FF) + 1;
-
- if (sx == dx && sy == dy && psx_gpu->mask_msb == 0)
- break;
- render_block_move(psx_gpu, sx, sy, dx, dy, w, h);
- sync_enhancement_buffers(dx, dy, w, h);
- break;
- }
-
- case 0xA0: // sys -> vid
- case 0xC0: // vid -> sys
+ case 0x80 ... 0x9F: // vid -> vid
+ case 0xA0 ... 0xBF: // sys -> vid
+ case 0xC0 ... 0xDF: // vid -> sys
goto breakloop;
case 0xE1:
{
s16 viewport_start_x = list[0] & 0x3FF;
s16 viewport_start_y = (list[0] >> 10) & 0x1FF;
- u32 w;
- s32 d;
if(viewport_start_x == psx_gpu->viewport_start_x &&
viewport_start_y == psx_gpu->viewport_start_y)
psx_gpu->saved_viewport_start_x = viewport_start_x;
psx_gpu->saved_viewport_start_y = viewport_start_y;
- w = (u32)psx_gpu->viewport_end_x - (u32)viewport_start_x + 1;
- d = psx_gpu->saved_hres - w;
- if(-16 <= d && d <= 16)
- {
- update_enhancement_buf_table_from_x(psx_gpu,
- viewport_start_x, w);
- }
select_enhancement_buf(psx_gpu);
#ifdef TEXTURE_CACHE_4BPP
{
s16 viewport_end_x = list[0] & 0x3FF;
s16 viewport_end_y = (list[0] >> 10) & 0x1FF;
- u32 w;
- s32 d;
if(viewport_end_x == psx_gpu->viewport_end_x &&
viewport_end_y == psx_gpu->viewport_end_y)
psx_gpu->saved_viewport_end_x = viewport_end_x;
psx_gpu->saved_viewport_end_y = viewport_end_y;
- w = (u32)viewport_end_x - (u32)psx_gpu->viewport_start_x + 1;
- d = psx_gpu->saved_hres - w;
- if(-16 <= d && d <= 16)
- {
- update_enhancement_buf_table_from_x(psx_gpu,
- psx_gpu->viewport_start_x, w);
- }
select_enhancement_buf(psx_gpu);
-
+#if 0
+ if (!psx_gpu->enhancement_current_buf_ptr)
+ log_anomaly("vp %3d,%3d %3d,%d - no buf\n",
+ psx_gpu->viewport_start_x, psx_gpu->viewport_start_y,
+ viewport_end_x, viewport_end_y);
+#endif
#ifdef TEXTURE_CACHE_4BPP
psx_gpu->viewport_mask =
texture_region_mask(psx_gpu->viewport_start_x,
enhancement_disable();
breakloop:
- if (last_command != NULL)
- *last_command = current_command;
+ *cpu_cycles_sum_out += cpu_cycles_sum;
+ *cpu_cycles_last = cpu_cycles;
+ *last_command = current_command;
return list - list_start;
}
notaz.gp2x.de / pcsx_rearmed.git / blobdiff