static void *psp_init(const video_info_t *video,
const input_driver_t **input, void **input_data)
{
// to-do : add ASSERT() checks or use main RAM if VRAM is too low for desired video->input_scale
void *pspinput;
int pixel_format, lut_pixel_format, lut_block_count;
unsigned int red_shift, color_mask;
void *displayBuffer, *LUT_r, *LUT_b;
psp1_video_t *psp = (psp1_video_t*)calloc(1, sizeof(psp1_video_t));
if (!psp)
return NULL;
sceGuInit();
psp->main_dList = memalign(16, 256); // make sure to allocate more space if bigger display lists are needed.
psp->frame_dList = memalign(16, 256);
psp->rgui.dList = memalign(16, 256);
psp->rgui.frame = memalign(16, 2 * 480 * 272);
psp->frame_coords = memalign(64, 1 * sizeof(psp1_sprite_t));
psp->rgui.frame_coords = memalign(64, 16 * sizeof(psp1_sprite_t));
memset(psp->frame_coords , 0, 1 * sizeof(psp1_sprite_t));
memset(psp->rgui.frame_coords , 0, 16 * sizeof(psp1_sprite_t));
sceKernelDcacheWritebackInvalidateAll();
psp->frame_coords = TO_UNCACHED_PTR(psp->frame_coords);
psp->rgui.frame_coords = TO_UNCACHED_PTR(psp->rgui.frame_coords);;
psp->frame_coords->v0.x = 60;
psp->frame_coords->v0.y = 0;
psp->frame_coords->v0.u = 0;
psp->frame_coords->v0.v = 0;
psp->frame_coords->v1.x = 420;
psp->frame_coords->v1.y = SCEGU_SCR_HEIGHT;
psp->frame_coords->v1.u = 256;
psp->frame_coords->v1.v = 240;
psp->vsync = video->vsync;
psp->rgb32 = video->rgb32;
if(psp->rgb32)
{
uint32_t* LUT_r_local = (uint32_t*)(SCEGU_VRAM_BP32_2);
uint32_t* LUT_b_local = (uint32_t*)(SCEGU_VRAM_BP32_2) + (1 << 8);
red_shift = 8 + 8;
color_mask = 0xFF;
lut_block_count = (1 << 8) / 8;
psp->texture = (void*)(LUT_b_local + (1 << 8));
psp->draw_buffer = SCEGU_VRAM_BP32_0;
psp->bpp_log2 = 2;
pixel_format = GU_PSM_8888;
lut_pixel_format = GU_PSM_T32;
displayBuffer = SCEGU_VRAM_BP32_1;
for (u32 i=0; i < (1 << 8); i++) {
LUT_r_local[i]= i;
LUT_b_local[i]= i << (8 + 8);
}
LUT_r = (void*)LUT_r_local;
LUT_b = (void*)LUT_b_local;
}
else
{
uint16_t* LUT_r_local = (uint16_t*)(SCEGU_VRAM_BP_2);
uint16_t* LUT_b_local = (uint16_t*)(SCEGU_VRAM_BP_2) + (1 << 5);
red_shift = 6 + 5;
color_mask = 0x1F;
lut_block_count = (1 << 5) / 8;
psp->texture = (void*)(LUT_b_local + (1 << 5));
psp->draw_buffer = SCEGU_VRAM_BP_0;
psp->bpp_log2 = 1;
pixel_format = GU_PSM_5650;
lut_pixel_format = GU_PSM_T16;
displayBuffer = SCEGU_VRAM_BP_1;
for (u16 i = 0; i < (1 << 5); i++) {
LUT_r_local[i]= i;
LUT_b_local[i]= i << (5 + 6);
}
LUT_r = (void*)LUT_r_local;
LUT_b = (void*)LUT_b_local;
}
sceDisplayWaitVblankStart(); // TODO : check if necessary
sceGuDisplay(GU_FALSE);
sceGuStart(GU_DIRECT, psp->main_dList);
sceGuDrawBuffer(pixel_format, TO_GU_POINTER(psp->draw_buffer), SCEGU_VRAM_WIDTH);
sceGuDispBuffer(SCEGU_SCR_WIDTH, SCEGU_SCR_HEIGHT, TO_GU_POINTER(displayBuffer), SCEGU_VRAM_WIDTH);
sceGuClearColor(0);
sceGuScissor(0, 0, SCEGU_SCR_WIDTH, SCEGU_SCR_HEIGHT);
sceGuEnable(GU_SCISSOR_TEST);
sceGuTexFilter(GU_LINEAR, GU_LINEAR); // TODO , move this to display list
sceGuTexWrap (GU_CLAMP, GU_CLAMP);
sceGuEnable(GU_TEXTURE_2D);
sceGuDisable(GU_DEPTH_TEST);
sceGuCallMode(GU_FALSE);
sceGuFinish();
sceGuSync(0, 0);
sceDisplayWaitVblankStart(); // TODO : check if necessary
sceGuDisplay(GU_TRUE);
pspDebugScreenSetColorMode(pixel_format);
pspDebugScreenSetBase(psp->draw_buffer);
// fill frame_dList :
sceGuStart(GU_CALL, psp->frame_dList);
sceGuTexMode(pixel_format, 0, 0, GU_FALSE);
sceGuTexFunc(GU_TFX_REPLACE, GU_TCC_RGB);
sceGuEnable(GU_BLEND);
sceGuBlendFunc(GU_ADD, GU_FIX, GU_FIX, 0x0000FF00, 0xFFFFFFFF); // green only
sceGuDrawArray(GU_SPRITES, GU_TEXTURE_16BIT | GU_COLOR_4444 | GU_VERTEX_16BIT | GU_TRANSFORM_2D, 2, NULL, (void*)(psp->frame_coords));
sceGuBlendFunc(GU_ADD, GU_FIX, GU_FIX, 0xFFFFFFFF, 0xFFFFFFFF); // restore
sceGuTexMode(lut_pixel_format, 0, 0, GU_FALSE);
sceGuClutMode(pixel_format, red_shift, color_mask, 0);
sceGuClutLoad(lut_block_count, LUT_r);
sceGuDrawArray(GU_SPRITES, GU_TEXTURE_16BIT | GU_COLOR_4444 | GU_VERTEX_16BIT | GU_TRANSFORM_2D, 2, NULL, (void*)(psp->frame_coords));
sceGuClutMode(pixel_format, 0, color_mask, 0);
sceGuClutLoad(lut_block_count, LUT_b);
sceGuDrawArray(GU_SPRITES, GU_TEXTURE_16BIT | GU_COLOR_4444 | GU_VERTEX_16BIT | GU_TRANSFORM_2D, 2, NULL, (void*)(psp->frame_coords));
sceGuFinish();
if (input && input_data)
{
pspinput = input_psp.init();
*input = pspinput ? &input_psp : NULL;
*input_data = pspinput;
}
return psp;
error:
RARCH_ERR("PSP1 video could not be initialized.\n");
return (void*)-1;
}
static void *psp_init(const video_info_t *video,
const input_driver_t **input, void **input_data)
{
/* to-do : add ASSERT() checks or use main RAM if
* VRAM is too low for desired video->input_scale. */
void *pspinput;
int pixel_format, lut_pixel_format, lut_block_count;
unsigned int red_shift, color_mask;
void *displayBuffer, *LUT_r, *LUT_b;
psp1_video_t *psp = (psp1_video_t*)calloc(1, sizeof(psp1_video_t));
if (!psp)
return NULL;
sceGuInit();
psp->vp.x = 0;
psp->vp.y = 0;
psp->vp.width = SCEGU_SCR_WIDTH;
psp->vp.height = SCEGU_SCR_HEIGHT;
psp->vp.full_width = SCEGU_SCR_WIDTH;
psp->vp.full_height = SCEGU_SCR_HEIGHT;
/* Make sure anything using uncached pointers reserves
* whole cachelines (memory address and size need to be a multiple of 64)
* so it isn't overwritten by an unlucky cache writeback.
*
* This includes display lists since the Gu library uses
* uncached pointers to write to them. */
/* Allocate more space if bigger display lists are needed. */
psp->main_dList = memalign(64, 256);
psp->frame_dList = memalign(64, 256);
psp->menu.dList = memalign(64, 256);
psp->menu.frame = memalign(16, 2 * 480 * 272);
psp->frame_coords = memalign(64,
(((PSP_FRAME_SLICE_COUNT * sizeof(psp1_sprite_t)) + 63) & ~63));
psp->menu.frame_coords = memalign(64,
(((PSP_FRAME_SLICE_COUNT * sizeof(psp1_sprite_t)) + 63) & ~63));
memset(psp->frame_coords, 0,
PSP_FRAME_SLICE_COUNT * sizeof(psp1_sprite_t));
memset(psp->menu.frame_coords, 0,
PSP_FRAME_SLICE_COUNT * sizeof(psp1_sprite_t));
sceKernelDcacheWritebackInvalidateAll();
psp->frame_coords = TO_UNCACHED_PTR(psp->frame_coords);
psp->menu.frame_coords = TO_UNCACHED_PTR(psp->menu.frame_coords);
psp->frame_coords->v0.x = 60;
psp->frame_coords->v0.y = 0;
psp->frame_coords->v0.u = 0;
psp->frame_coords->v0.v = 0;
psp->frame_coords->v1.x = 420;
psp->frame_coords->v1.y = SCEGU_SCR_HEIGHT;
psp->frame_coords->v1.u = 256;
psp->frame_coords->v1.v = 240;
psp->vsync = video->vsync;
psp->rgb32 = video->rgb32;
if(psp->rgb32)
{
uint32_t* LUT_r_local = (uint32_t*)(SCEGU_VRAM_BP32_2);
uint32_t* LUT_b_local = (uint32_t*)(SCEGU_VRAM_BP32_2) + (1 << 8);
red_shift = 8 + 8;
color_mask = 0xFF;
lut_block_count = (1 << 8) / 8;
psp->texture = (void*)(LUT_b_local + (1 << 8));
psp->draw_buffer = SCEGU_VRAM_BP32_0;
psp->bpp_log2 = 2;
pixel_format = GU_PSM_8888;
lut_pixel_format = GU_PSM_T32;
displayBuffer = SCEGU_VRAM_BP32_1;
for (u32 i=0; i < (1 << 8); i++){
LUT_r_local[i]= i;
LUT_b_local[i]= i << (8 + 8);
}
LUT_r = (void*)LUT_r_local;
LUT_b = (void*)LUT_b_local;
}
else
{
uint16_t* LUT_r_local = (uint16_t*)(SCEGU_VRAM_BP_2);
uint16_t* LUT_b_local = (uint16_t*)(SCEGU_VRAM_BP_2) + (1 << 5);
red_shift = 6 + 5;
color_mask = 0x1F;
lut_block_count = (1 << 5) / 8;
psp->texture = (void*)(LUT_b_local + (1 << 5));
psp->draw_buffer = SCEGU_VRAM_BP_0;
psp->bpp_log2 = 1;
pixel_format =
(g_extern.system.pix_fmt == RETRO_PIXEL_FORMAT_0RGB1555)
? GU_PSM_5551 : GU_PSM_5650 ;
lut_pixel_format = GU_PSM_T16;
displayBuffer = SCEGU_VRAM_BP_1;
for (u16 i = 0; i < (1 << 5); i++)
{
LUT_r_local[i]= i;
LUT_b_local[i]= i << (5 + 6);
}
LUT_r = (void*)LUT_r_local;
LUT_b = (void*)LUT_b_local;
}
psp->tex_filter = video->smooth? GU_LINEAR : GU_NEAREST;
/* TODO: check if necessary. */
sceDisplayWaitVblankStart();
sceGuDisplay(GU_FALSE);
sceGuStart(GU_DIRECT, psp->main_dList);
sceGuDrawBuffer(pixel_format, TO_GU_POINTER(psp->draw_buffer),
SCEGU_VRAM_WIDTH);
sceGuDispBuffer(SCEGU_SCR_WIDTH, SCEGU_SCR_HEIGHT,
TO_GU_POINTER(displayBuffer), SCEGU_VRAM_WIDTH);
sceGuClearColor(0);
sceGuScissor(0, 0, SCEGU_SCR_WIDTH, SCEGU_SCR_HEIGHT);
sceGuEnable(GU_SCISSOR_TEST);
sceGuTexFilter(psp->tex_filter, psp->tex_filter);
sceGuTexWrap (GU_CLAMP, GU_CLAMP);
sceGuEnable(GU_TEXTURE_2D);
sceGuDisable(GU_DEPTH_TEST);
sceGuCallMode(GU_FALSE);
sceGuFinish();
sceGuSync(0, 0);
/* TODO : check if necessary */
sceDisplayWaitVblankStart();
sceGuDisplay(GU_TRUE);
pspDebugScreenSetColorMode(pixel_format);
pspDebugScreenSetBase(psp->draw_buffer);
/* fill frame_dList : */
sceGuStart(GU_CALL, psp->frame_dList);
sceGuTexMode(pixel_format, 0, 0, GU_FALSE);
sceGuTexFunc(GU_TFX_REPLACE, GU_TCC_RGB);
sceGuEnable(GU_BLEND);
/* green only */
sceGuBlendFunc(GU_ADD, GU_FIX, GU_FIX, 0x0000FF00, 0xFFFFFFFF);
sceGuDrawArray(GU_SPRITES, GU_TEXTURE_32BITF | GU_VERTEX_32BITF |
GU_TRANSFORM_2D, PSP_FRAME_VERTEX_COUNT, NULL,
(void*)(psp->frame_coords));
/* restore */
sceGuBlendFunc(GU_ADD, GU_FIX, GU_FIX, 0xFFFFFFFF, 0xFFFFFFFF);
sceGuTexMode(lut_pixel_format, 0, 0, GU_FALSE);
sceGuClutMode(pixel_format, red_shift, color_mask, 0);
sceGuClutLoad(lut_block_count, LUT_r);
sceGuDrawArray(GU_SPRITES, GU_TEXTURE_32BITF | GU_VERTEX_32BITF |
GU_TRANSFORM_2D, PSP_FRAME_VERTEX_COUNT, NULL,
(void*)(psp->frame_coords));
sceGuClutMode(pixel_format, 0, color_mask, 0);
sceGuClutLoad(lut_block_count, LUT_b);
sceGuDrawArray(GU_SPRITES, GU_TEXTURE_32BITF | GU_VERTEX_32BITF |
GU_TRANSFORM_2D, PSP_FRAME_VERTEX_COUNT, NULL,
(void*)(psp->frame_coords));
sceGuFinish();
if (input && input_data)
{
pspinput = input_psp.init();
*input = pspinput ? &input_psp : NULL;
*input_data = pspinput;
}
psp->vblank_not_reached = true;
sceKernelRegisterSubIntrHandler(PSP_VBLANK_INT, 0, psp_on_vblank, psp);
sceKernelEnableSubIntr(PSP_VBLANK_INT, 0);
psp->keep_aspect = true;
psp->should_resize = true;
psp->hw_render = false;
return psp;
error:
RARCH_ERR("PSP1 video could not be initialized.\n");
return (void*)-1;
}
int main(int argc, char* argv[])
{
unsigned int i,j;
pspDebugScreenInit();
SetupCallbacks();
#ifdef ENABLE_PROFILER
// Enable profiling
pspDebugProfilerClear();
pspDebugProfilerEnable();
#endif
// initialize global context
g_context.iterationCount = NUM_VERTEX_BUFFERS * NUM_ITERATIONS;
g_context.t = 0;
g_context.sint = 0;
// initialize torus
for (i = 0; i < NUM_SLICES; ++i)
{
for (j = 0; j < NUM_ROWS; ++j)
{
float s = i + 0.5f, t = j;
float x,y,z;
x = (RING_SIZE + RING_RADIUS * cosf(s * ((GU_PI*2)/NUM_SLICES))) * cosf(t * ((GU_PI*2)/NUM_ROWS));
y = (RING_SIZE + RING_RADIUS * cosf(s * ((GU_PI*2)/NUM_SLICES))) * sinf(t * ((GU_PI*2)/NUM_ROWS));
z = RING_RADIUS * sinf(s * ((GU_PI*2)/NUM_SLICES));
torus_vertices[j + i * NUM_ROWS].x = x;
torus_vertices[j + i * NUM_ROWS].y = y;
torus_vertices[j + i * NUM_ROWS].z = z;
}
}
// initialize torus modifiers
for (j = 0; j < NUM_ROWS; ++j)
{
float t = j;
torus_modifiers[j].x = 0;
torus_modifiers[j].y = 0;
torus_modifiers[j].z = 0.3*cosf( t * 8.0f *((GU_PI*2)/NUM_ROWS) );
}
// init GU and set callbacks
sceGuInit();
// 0x01 - user callback
// 0x04 - 'rendering finished' callback
sceGuSetCallback(1, &mySignalHandler);
sceGuSetCallback(4, &myFinishHandler);
// setup GU
sceGuStart(GU_DIRECT,list);
sceGuDrawBuffer(GU_PSM_8888,(void*)0,BUF_WIDTH);
sceGuDispBuffer(SCR_WIDTH,SCR_HEIGHT,(void*)0x88000,BUF_WIDTH);
sceGuDepthBuffer((void*)0x110000,BUF_WIDTH);
sceGuOffset(2048 - (SCR_WIDTH/2),2048 - (SCR_HEIGHT/2));
sceGuViewport(2048,2048,SCR_WIDTH,SCR_HEIGHT);
sceGuDepthRange(0xc350,0x2710);
sceGuScissor(0,0,SCR_WIDTH,SCR_HEIGHT);
sceGuEnable(GU_SCISSOR_TEST);
sceGuAlphaFunc(GU_GREATER,0,0xff);
sceGuEnable(GU_ALPHA_TEST);
sceGuDepthFunc(GU_GEQUAL);
sceGuEnable(GU_DEPTH_TEST);
sceGuFrontFace(GU_CW);
sceGuShadeModel(GU_SMOOTH);
sceGuEnable(GU_CULL_FACE);
sceGuEnable(GU_TEXTURE_2D);
sceGuFinish();
sceGuSync(0,0);
sceDisplayWaitVblankStart();
sceGuDisplay(GU_TRUE);
// run sample
#ifdef USING_SIGNALS
sceGuCallMode(1);
#endif
// generate callable command-list with texture setup
{
sceGuStart(GU_CALL, smallList1);
// setup texture
sceGuTexMode(GU_PSM_5551,0,0,0);
sceGuTexImage(0,32,32,32,ball_start); // width, height, buffer width, tbp
sceGuTexFunc(GU_TFX_MODULATE,GU_TCC_RGBA); // NOTE: this enables reads of the alpha-component from the texture, otherwise blend/test won't work
sceGuTexFilter(GU_NEAREST,GU_NEAREST);
sceGuTexWrap(GU_CLAMP,GU_CLAMP);
sceGuTexScale(1,1);
sceGuTexOffset(0,0);
sceGuAmbientColor(0xffffffff);
sceGuFinish();
sceGuSync(0,0);
}
// generate callable command-list for cube rendering
{
sceGuStart(GU_CALL, smallList2);
// draw cube
sceGuDrawArray(GU_TRIANGLES,GU_TEXTURE_32BITF|GU_COLOR_8888|GU_VERTEX_32BITF|GU_TRANSFORM_3D,12*3,0,cubeVertices);
sceGuFinish();
sceGuSync(0,0);
}
for(;;)
{
sceGuStart(GU_DIRECT,list);
unsigned int i = 0;
for( ; i < NUM_VERTEX_BUFFERS; i++ )
g_context.vbuffer[i] = sceGuGetMemory((NUM_SLICES/g_context.iterationCount) * 2 * NUM_ROWS * sizeof(Vertex));
g_context.vertsRendered = 0;
// clear screen
sceGuClearColor(0x00334455);
sceGuClearDepth(0);
sceGuClear(GU_COLOR_BUFFER_BIT|GU_DEPTH_BUFFER_BIT);
// setup matrices
sceGumMatrixMode(GU_PROJECTION);
sceGumLoadIdentity();
sceGumPerspective(75.0f,16.0f/9.0f,0.5f,1000.0f);
sceGumMatrixMode(GU_VIEW);
sceGumLoadIdentity();
sceGumMatrixMode(GU_MODEL);
{
ScePspFVector3 pos = {0.0f,0.0f,-3.5f};
ScePspFVector3 rot = {g_context.t * 0.3f * (GU_PI/180.0f), g_context.t * 0.7f * (GU_PI/180.0f), g_context.t * 1.3f * (GU_PI/180.0f)};
sceGumLoadIdentity();
sceGumTranslate(&pos);
sceGumRotateXYZ(&rot);
}
sceGumStoreMatrix(&g_context.world);
// call pregenerated command-list to setup texture
sceGuCallList(smallList1);
// start billboard rendering
render_billboards(0);
// call pregenerated command-list to render cube
{
ScePspFVector3 scale = {0.3f, 0.3f, 0.3f};
sceGumScale(&scale);
}
sceGumUpdateMatrix();
sceGuCallList(smallList2);
#ifndef USING_SIGNALS
// HACK: sceGuFinish() is called inside the signal interupt handler when all rendering job is done
// this is done in order to stall GPU if it is ahead of CPU
sceGuFinish();
#endif
sceGuSync(0,0);
#ifndef ENABLE_FRAMERATE
// wait for next frame
sceDisplayWaitVblankStart();
#endif
sceGuSwapBuffers();
pspDebugScreenSetXY(0,0);
#ifdef ENABLE_PROFILER
// Print profile information to the screen
pspDebugProfilerPrint();
#endif
#ifdef ENABLE_FRAMERATE
// simple frame rate counter
static float curr_ms = 1.0f;
static struct timeval time_slices[16];
static int t = 0;
float curr_fps = 1.0f / curr_ms;
t++;
float vertsPerSec = g_context.vertsRendered*curr_fps;
float kbPerSec = vertsPerSec * sizeof(Vertex) / 1024.0f;
gettimeofday(&time_slices[t & 15],0);
pspDebugScreenPrintf("fps: %d.%03d ms: %d vert/s: %dK MB/s: %d.%03d",(int)curr_fps, ((int)(curr_fps*1000.0f)%1000), (int)(curr_ms*1000.0f),
(int)(vertsPerSec/1000.0f),
(int)(kbPerSec/1024.0f), (int)((1000.0f/1024.0f)*((int)kbPerSec%1024)) );
if (!(t & 15))
{
struct timeval last_time = time_slices[0];
unsigned int i;
curr_ms = 0;
for (i = 1; i < 16; ++i)
{
struct timeval curr_time = time_slices[i];
int curr_time_usec = curr_time.tv_usec + curr_time.tv_sec * 1000000;
int last_time_usec = last_time.tv_usec + last_time.tv_sec * 1000000;
if( last_time_usec < curr_time_usec )
curr_ms += (( curr_time_usec - last_time_usec ) * (1.0f/1000000.0f));
last_time = time_slices[i];
}
curr_ms /= 15.0f;
}
#endif
}
sceGuTerm();
sceKernelExitGame();
return 0;
}
