gpu_neon: fix sign extension

[pcsx_rearmed.git] / plugins / gpu_neon / psx_gpu / psx_gpu_parse.c

diff --git a/plugins/gpu_neon/psx_gpu/psx_gpu_parse.c b/plugins/gpu_neon/psx_gpu/psx_gpu_parse.c
index af26fa3..53f33e4 100644 (file)
--- a/plugins/gpu_neon/psx_gpu/psx_gpu_parse.c
+++ b/plugins/gpu_neon/psx_gpu/psx_gpu_parse.c
@@ -200,21 +200,14 @@ static void do_fill(psx_gpu_struct *psx_gpu, u32 x, u32 y,
   }
 }
 
-#define sign_extend_12bit(value)                                               \
-  (((s32)((value) << 20)) >> 20)                                               \
-
 #define sign_extend_11bit(value)                                               \
   (((s32)((value) << 21)) >> 21)                                               \
 
-#define sign_extend_10bit(value)                                               \
-  (((s32)((value) << 22)) >> 22)                                               \
-
-
 #define get_vertex_data_xy(vertex_number, offset16)                            \
   vertexes[vertex_number].x =                                                  \
-   sign_extend_12bit(list_s16[offset16]) + psx_gpu->offset_x;                  \
+   sign_extend_11bit(list_s16[offset16]) + psx_gpu->offset_x;                  \
   vertexes[vertex_number].y =                                                  \
-   sign_extend_12bit(list_s16[(offset16) + 1]) + psx_gpu->offset_y;            \
+   sign_extend_11bit(list_s16[(offset16) + 1]) + psx_gpu->offset_y;            \
 
 #define get_vertex_data_uv(vertex_number, offset16)                            \
   vertexes[vertex_number].u = list_s16[offset16] & 0xFF;                       \
@@ -252,10 +245,11 @@ static void do_fill(psx_gpu_struct *psx_gpu, u32 x, u32 y,
 #endif
 
 u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
- s32 *cpu_cycles_out, 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, cpu_cycles = 0;
+  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);
@@ -284,7 +278,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         u32 color = list[0] & 0xFFFFFF;
 
         do_fill(psx_gpu, x, y, width, height, color);
-        cpu_cycles += gput_fill(width, height);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_fill(width, height));
         break;
       }
 
@@ -297,7 +291,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy(2, 6);
           
         render_triangle(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_poly_base();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base());
         break;
       }
   
@@ -312,7 +306,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy_uv(2, 10);
   
         render_triangle(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_poly_base_t();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_t());
         break;
       }
   
@@ -327,7 +321,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
   
         render_triangle(psx_gpu, vertexes, current_command);
         render_triangle(psx_gpu, &(vertexes[1]), current_command);
-        cpu_cycles += gput_quad_base();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base());
         break;
       }
   
@@ -344,7 +338,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
   
         render_triangle(psx_gpu, vertexes, current_command);
         render_triangle(psx_gpu, &(vertexes[1]), current_command);
-        cpu_cycles += gput_quad_base_t();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_t());
         break;
       }
   
@@ -355,7 +349,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy_rgb(2, 8);
   
         render_triangle(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_poly_base_g();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_g());
         break;
       }
   
@@ -369,7 +363,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy_uv_rgb(2, 12);
 
         render_triangle(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_poly_base_gt();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_gt());
         break;
       }
   
@@ -382,7 +376,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
   
         render_triangle(psx_gpu, vertexes, current_command);
         render_triangle(psx_gpu, &(vertexes[1]), current_command);
-        cpu_cycles += gput_quad_base_g();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_g());
         break;
       }
   
@@ -398,7 +392,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
   
         render_triangle(psx_gpu, vertexes, current_command);
         render_triangle(psx_gpu, &(vertexes[1]), current_command);
-        cpu_cycles += gput_quad_base_gt();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_gt());
         break;
       }
   
@@ -410,7 +404,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         vertexes[1].y = list_s16[5] + psx_gpu->offset_y;
 
         render_line(psx_gpu, vertexes, current_command, list[0], 0);
-        cpu_cycles += gput_line(0);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
         break;
       }
   
@@ -432,7 +426,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
           vertexes[1].y = (xy >> 16) + psx_gpu->offset_y;
 
           render_line(psx_gpu, vertexes, current_command, list[0], 0);
-          cpu_cycles += gput_line(0);
+          gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
 
           list_position++;
           num_vertexes++;
@@ -467,7 +461,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         vertexes[1].y = list_s16[7] + psx_gpu->offset_y;
 
         render_line(psx_gpu, vertexes, current_command, 0, 0);
-        cpu_cycles += gput_line(0);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
         break;
       }
  
@@ -498,7 +492,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
           vertexes[1].y = (xy >> 16) + psx_gpu->offset_y;
 
           render_line(psx_gpu, vertexes, current_command, 0, 0);
-          cpu_cycles += gput_line(0);
+          gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
 
           list_position += 2;
           num_vertexes++;
@@ -527,7 +521,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
            current_command, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
         break;
       }
   
@@ -543,7 +537,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF,
            &width, &height, current_command, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
         break;
       }
   
@@ -555,7 +549,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
            current_command, list[0]);
-        cpu_cycles += gput_sprite(1, 1);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(1, 1));
         break;
       }
   
@@ -567,7 +561,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
            current_command, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
         break;
       }
   
@@ -582,7 +576,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF,
            &width, &height, current_command, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
         break;
       }
   
@@ -594,7 +588,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         render_sprite(psx_gpu, x, y, 0, 0, &width, &height,
            current_command, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
         break;
       }
   
@@ -609,7 +603,7 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         render_sprite(psx_gpu, x, y, uv & 0xFF, (uv >> 8) & 0xFF,
            &width, &height, current_command, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(width, height));
         break;
       }
   
@@ -789,7 +783,8 @@ u32 gpu_parse(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
   }
 
 breakloop:
-  *cpu_cycles_out += cpu_cycles;
+  *cpu_cycles_sum_out += cpu_cycles_sum;
+  *cpu_cycles_last = cpu_cycles;
   *last_command = current_command;
   return list - list_start;
 }
@@ -1016,8 +1011,6 @@ void scale2x_tiles8(void *dst, const void *src, int w8, int h)
 }
 #endif
 
-static int disable_main_render;
-
 // 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)
@@ -1063,6 +1056,7 @@ static void patch_v(vertex_struct *vertex_ptrs, int count, int old, int new)
       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];
@@ -1101,7 +1095,7 @@ static void do_triangle_enhanced(psx_gpu_struct *psx_gpu,
   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))
@@ -1202,10 +1196,11 @@ static void do_sprite_enhanced(psx_gpu_struct *psx_gpu, int x, int y,
 #endif
 
 u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
- s32 *cpu_cycles_out, 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, cpu_cycles = 0;
+  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);
@@ -1244,7 +1239,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         x &= ~0xF;
         width = ((width + 0xF) & ~0xF);
-        cpu_cycles += gput_fill(width, height);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_fill(width, height));
         if (width == 0 || height == 0)
           break;
 
@@ -1275,7 +1270,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy(2, 6);
 
         do_triangle_enhanced(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_poly_base();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base());
         break;
       }
   
@@ -1290,7 +1285,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy_uv(2, 10);
   
         do_triangle_enhanced(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_poly_base_t();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_t());
         break;
       }
   
@@ -1304,7 +1299,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy(3, 8);
 
         do_quad_enhanced(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_quad_base();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base());
         break;
       }
   
@@ -1319,9 +1314,10 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         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);
-        cpu_cycles += gput_quad_base_t();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_t());
         break;
       }
   
@@ -1332,7 +1328,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy_rgb(2, 8);
   
         do_triangle_enhanced(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_poly_base_g();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_g());
         break;
       }
   
@@ -1346,7 +1342,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy_uv_rgb(2, 12);
 
         do_triangle_enhanced(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_poly_base_gt();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_gt());
         break;
       }
   
@@ -1358,7 +1354,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         get_vertex_data_xy_rgb(3, 12);
   
         do_quad_enhanced(psx_gpu, vertexes, current_command);
-        cpu_cycles += gput_quad_base_g();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_g());
         break;
       }
   
@@ -1372,9 +1368,10 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         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);
-        cpu_cycles += gput_quad_base_gt();
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_gt());
         break;
       }
   
@@ -1388,7 +1385,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         render_line(psx_gpu, vertexes, current_command, list[0], 0);
         if (enhancement_enable(psx_gpu))
           render_line(psx_gpu, vertexes, current_command, list[0], 1);
-        cpu_cycles += gput_line(0);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
         break;
       }
   
@@ -1413,7 +1410,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
           render_line(psx_gpu, vertexes, current_command, list[0], 0);
           if (enhancement_enable(psx_gpu))
             render_line(psx_gpu, vertexes, current_command, list[0], 1);
-          cpu_cycles += gput_line(0);
+          gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
 
           list_position++;
           num_vertexes++;
@@ -1450,7 +1447,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
         render_line(psx_gpu, vertexes, current_command, 0, 0);
         if (enhancement_enable(psx_gpu))
           render_line(psx_gpu, vertexes, current_command, 0, 1);
-        cpu_cycles += gput_line(0);
+        gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
         break;
       }
  
@@ -1484,7 +1481,7 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
           render_line(psx_gpu, vertexes, current_command, 0, 0);
           if (enhancement_enable(psx_gpu))
             render_line(psx_gpu, vertexes, current_command, 0, 1);
-          cpu_cycles += gput_line(0);
+          gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0));
 
           list_position += 2;
           num_vertexes++;
@@ -1513,10 +1510,13 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         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]);
-        cpu_cycles += gput_sprite(width, height);
+        }
         break;
       }
   
@@ -1533,10 +1533,13 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         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]);
-        cpu_cycles += gput_sprite(width, height);
+        }
         break;
       }
   
@@ -1548,10 +1551,10 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         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, width, height, list[0]);
-        cpu_cycles += gput_sprite(1, 1);
+          do_sprite_enhanced(psx_gpu, x, y, 0, 0, 1, 1, list[0]);
         break;
       }
   
@@ -1563,10 +1566,10 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         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, width, height, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+          do_sprite_enhanced(psx_gpu, x, y, 0, 0, 8, 8, list[0]);
         break;
       }
   
@@ -1582,10 +1585,10 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         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, width, height, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+          do_sprite_enhanced(psx_gpu, x, y, u, v, 8, 8, list[0]);
         break;
       }
   
@@ -1597,10 +1600,10 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         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, width, height, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+          do_sprite_enhanced(psx_gpu, x, y, 0, 0, 16, 16, list[0]);
         break;
       }
   
@@ -1616,10 +1619,10 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
 
         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, width, height, list[0]);
-        cpu_cycles += gput_sprite(width, height);
+          do_sprite_enhanced(psx_gpu, x, y, u, v, 16, 16, list[0]);
         break;
       }
 
@@ -1736,10 +1739,8 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
   
       case 0xE5:
       {
-        s32 offset_x = list[0] << 21;
-        s32 offset_y = list[0] << 10;
-        psx_gpu->offset_x = offset_x >> 21;
-        psx_gpu->offset_y = offset_y >> 21; 
+        psx_gpu->offset_x = sign_extend_11bit(list[0]);
+        psx_gpu->offset_y = sign_extend_11bit(list[0] >> 11);
   
         SET_Ex(5, list[0]);
         break;
@@ -1773,7 +1774,8 @@ u32 gpu_parse_enhanced(psx_gpu_struct *psx_gpu, u32 *list, u32 size,
   enhancement_disable();
 
 breakloop:
-  *cpu_cycles_out += cpu_cycles;
+  *cpu_cycles_sum_out += cpu_cycles_sum;
+  *cpu_cycles_last = cpu_cycles;
   *last_command = current_command;
   return list - list_start;
 }