mirror/RetroArch
1
0
Fork 0
mirror of https://github.com/libretro/RetroArch synced 2025-02-26 15:39:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

(RGL) Optimize for 1 faced, 1 level images

Browse Source
This commit is contained in:
twinaphex 2013-03-26 20:17:17 +01:00
parent da91b64b45
commit 443c18d7cc
2 changed files with 73 additions and 155 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
ps3/rgl/include/RGL/private.h
View File

@ -186,7 +186,6 @@ GLuint rglValidateAttributesSlow (void *data, GLboolean *isMain);
//----------------------------------------
// Raster/.../PlatformTexture.c
//----------------------------------------
extern int rglPlatformTextureMaxUnits (void);
extern void rglPlatformCreateTexture (void *data);
extern void rglPlatformDestroyTexture (void *data);
void rglPlatformValidateVertexTextures (void);

227
ps3/rgl/src/ps3/rgl_ps3_raster.cpp
View File

@ -1486,45 +1486,6 @@ GLenum rglPlatformFramebufferCheckStatus (void *data)
return GL_FRAMEBUFFER_COMPLETE_OES;
}
// Calculate byte offset for given texture layout, number of faces, and levels
static GLuint rglGetGcmImageOffset (void *data, GLuint face, GLuint level)
{
rglGcmTextureLayout *layout = (rglGcmTextureLayout*)data;
GLuint bytes = 0;
GLuint faceAlign = layout->pitch ? 1 : 128;
GLuint minWidth = 1;
GLuint minHeight = 1;
for (GLuint f = 0; f < layout->faces; ++f)
{
GLuint width = layout->baseWidth;
GLuint height = layout->baseHeight;
GLuint depth = layout->baseDepth;
for ( GLuint i = 0;i < layout->levels; ++i )
{
if ((level == i) && (face == f))
return bytes;
width = MAX( minWidth, width );
height = MAX( minHeight, height );
depth = MAX( 1U, depth );
if ( !layout->pitch )
bytes += layout->pixelBits * width * height * depth / 8;
else
bytes += height * depth * layout->pitch;
width >>= 1;
height >>= 1;
depth >>= 1;
}
bytes = rglPad( bytes, faceAlign );
}
return 0;
}
static void rglPlatformValidateTextureResources (void *data);
void rglPlatformFramebuffer::validate (void *data)
@ -1581,7 +1542,7 @@ void rglPlatformFramebuffer::validate (void *data)
// set the render target
rt.colorId[i] = nvTexture->gpuAddressId;
rt.colorIdOffset[i] = nvTexture->gpuAddressIdOffset + rglGetGcmImageOffset( &nvTexture->gpuLayout, face, 0 );
rt.colorIdOffset[i] = nvTexture->gpuAddressIdOffset;
rt.colorPitch[i] = nvTexture->gpuLayout.pitch ? nvTexture->gpuLayout.pitch : nvTexture->gpuLayout.pixelBits * nvTexture->gpuLayout.baseWidth / 8;
width = MIN( width, nvTexture->gpuLayout.baseWidth );
@ -2141,47 +2102,23 @@ static GLuint rglGetGcmTextureSize (void *data)
GLuint bytesNeeded = 0;
GLuint faceAlign = layout->pitch ? 1 : 128;
GLuint minWidth = 1;
GLuint minHeight = 1;
for ( GLuint face = 0;face < layout->faces;++face )
{
GLuint width = layout->baseWidth;
GLuint height = layout->baseHeight;
GLuint depth = layout->baseDepth;
for ( GLuint i = 0;i < layout->levels; ++i )
{
width = MAX( minWidth, width );
height = MAX( minHeight, height );
depth = MAX( 1U, depth );
GLuint width = layout->baseWidth;
GLuint height = layout->baseHeight;
if ( !layout->pitch )
bytesNeeded += layout->pixelBits * width * height * depth / 8;
else
bytesNeeded += height * depth * layout->pitch;
width = MAX(1U, width );
height = MAX(1U, height );
width >>= 1;
height >>= 1;
depth >>= 1;
}
if ( !layout->pitch )
bytesNeeded += layout->pixelBits * width * height / 8;
else
bytesNeeded += height * layout->pitch;
bytesNeeded = rglPad( bytesNeeded, faceAlign );
}
return bytesNeeded;
return rglPad( bytesNeeded, faceAlign );
}
// Calculate pitch for a texture
static GLuint _getTexturePitch (const void *data)
{
const rglTexture *texture = (const rglTexture*)data;
return rglPad( rglGetStorageSize( texture->image->format, texture->image->type, texture->image->width, 1, 1 ), 64 ); // TransferVid2Vid needs 64byte pitch alignment
}
// Return maximum number of texture image units
int rglPlatformTextureMaxUnits (void)
{
return RGLGCM_MAX_TEXIMAGE_COUNT;
}
// TransferVid2Vid needs 64byte pitch alignment
#define GET_TEXTURE_PITCH(texture) (rglPad( rglGetStorageSize( texture->image->format, texture->image->type, texture->image->width, 1, 1 ), 64 ))
// Create a gcm texture by initializing memory to 0
void rglPlatformCreateTexture (void *data)
@ -2286,26 +2223,6 @@ static inline GLenum unFilter( GLenum filter )
return newFilter;
}
// Choose a texture layout and store it to newLayout, based on texture's filtering mode, swizzling, and size
void rglPlatformChooseGPUFormatAndLayout(
const void *data, GLboolean forceLinear,
GLuint pitch, rglGcmTextureLayout* newLayout )
{
const rglTexture *texture = (const rglTexture*)data;
rglImage *image = texture->image + texture->baseLevel;
GLuint levels = 1;
newLayout->levels = levels;
newLayout->faces = texture->faceCount;
newLayout->baseWidth = image->width;
newLayout->baseHeight = image->height;
newLayout->baseDepth = image->depth;
newLayout->internalFormat = ( rglGcmEnum )image->internalFormat;
newLayout->pixelBits = rglPlatformGetBitsPerPixel( newLayout->internalFormat );
newLayout->pitch = pitch ? pitch : _getTexturePitch(texture);
}
// texture strategy actions
// A texture strategy is a sequence of actions represented by these tokens.
// RGL_TEXTURE_STRATEGY_END must be the last token in any strategy.
@ -2408,47 +2325,59 @@ void rglPlatformReallocateGcmTexture (void *data)
forceLinear = GL_TRUE;
break;
case RGL_TEXTURE_STRATEGY_UNTILED_ALLOC:
// get layout and size compatible with this pool
rglPlatformChooseGPUFormatAndLayout( texture, forceLinear, 0, &newLayout );
size = rglGetGcmTextureSize( &newLayout );
// determine if current allocation already works
// If the current allocation has the right size and pool, we
// don't have to do anything. If not, we only drop from the
// target pool because we may reuse the allocation from a
// different pool in a later step.
if ( gcmTexture->pool == RGLGCM_SURFACE_POOL_LINEAR )
{
if ( currentSize >= size && newLayout.pitch == currentLayout.pitch )
{
gcmTexture->gpuLayout = newLayout;
done = GL_TRUE;
}
else
rglPlatformDropTexture( texture );
}
// get layout and size compatible with this pool
rglImage *image = texture->image + texture->baseLevel;
if ( !done )
{
// allocate in the specified pool
id = gmmAlloc((CellGcmContextData*)&rglGcmState_i.fifo,
CELL_GCM_LOCATION_LOCAL,
0,
size);
if ( id != GMM_ERROR )
newLayout.levels = 1;
newLayout.faces = texture->faceCount;
newLayout.baseWidth = image->width;
newLayout.baseHeight = image->height;
newLayout.baseDepth = image->depth;
newLayout.internalFormat = ( rglGcmEnum )image->internalFormat;
newLayout.pixelBits = rglPlatformGetBitsPerPixel( newLayout.internalFormat );
newLayout.pitch = GET_TEXTURE_PITCH(texture);
size = rglGetGcmTextureSize( &newLayout );
// determine if current allocation already works
// If the current allocation has the right size and pool, we
// don't have to do anything. If not, we only drop from the
// target pool because we may reuse the allocation from a
// different pool in a later step.
if ( gcmTexture->pool == RGLGCM_SURFACE_POOL_LINEAR )
{
// drop old allocation
if ( gcmTexture->pool != RGLGCM_SURFACE_POOL_NONE )
if ( currentSize >= size && newLayout.pitch == currentLayout.pitch )
{
gcmTexture->gpuLayout = newLayout;
done = GL_TRUE;
}
else
rglPlatformDropTexture( texture );
}
// set new
gcmTexture->pool = RGLGCM_SURFACE_POOL_LINEAR;
gcmTexture->gpuAddressId = id;
gcmTexture->gpuAddressIdOffset = 0;
gcmTexture->gpuSize = size;
gcmTexture->gpuLayout = newLayout;
if ( !done )
{
// allocate in the specified pool
id = gmmAlloc((CellGcmContextData*)&rglGcmState_i.fifo,
CELL_GCM_LOCATION_LOCAL,
0,
size);
if ( id != GMM_ERROR )
{
// drop old allocation
if ( gcmTexture->pool != RGLGCM_SURFACE_POOL_NONE )
rglPlatformDropTexture( texture );
done = GL_TRUE;
// set new
gcmTexture->pool = RGLGCM_SURFACE_POOL_LINEAR;
gcmTexture->gpuAddressId = id;
gcmTexture->gpuAddressIdOffset = 0;
gcmTexture->gpuSize = size;
gcmTexture->gpuLayout = newLayout;
done = GL_TRUE;
}
}
}
break;
@ -2554,12 +2483,6 @@ source: RGLGCM_SURFACE_SOURCE_TEXTURE,
if ( image->dataState == RGL_IMAGE_DATASTATE_HOST )
{
// determine image offset from base address
// TODO: compute all offsets at once for efficiency
// This is the offset in bytes for this face/image from the
// texture base address.
const GLuint dataOffset = rglGetGcmImageOffset( layout, 0, 0 );
// set source pixel buffer
src.ppuData = image->data;
@ -2572,10 +2495,10 @@ source: RGLGCM_SURFACE_SOURCE_TEXTURE,
{
// copy image to bounce buffer
src.dataId = bounceBufferId;
src.dataIdOffset = dataOffset;
src.dataIdOffset = 0;
// NPOT DXT
__builtin_memcpy( gmmIdToAddress( src.dataId ) + dataOffset,
__builtin_memcpy( gmmIdToAddress( src.dataId ),
image->data, image->storageSize );
}
@ -2587,24 +2510,11 @@ source: RGLGCM_SURFACE_SOURCE_TEXTURE,
dst.width = src.width;
dst.height = image->height;
dst.dataId = gcmTexture->gpuAddressId;
dst.dataIdOffset = gcmTexture->gpuAddressIdOffset + dataOffset;
GLuint offsetHeight = 0;
if(dst.pitch)
{
// linear (not swizzled)
// The tiled linear format requires that render
// targets be aligned to 8*pitch from the start of
// the tiled region.
offsetHeight = ( dataOffset / dst.pitch ) % 8;
dst.height += offsetHeight;
dst.dataIdOffset -= offsetHeight * dst.pitch;
}
dst.dataIdOffset = gcmTexture->gpuAddressIdOffset;
rglGcmCopySurface(
&src, 0, 0,
&dst, 0, offsetHeight,
&dst, 0, 0,
src.width, src.height,
GL_TRUE ); // don't bypass GPU pipeline
@ -2882,7 +2792,16 @@ GLAPI void APIENTRY glTextureReferenceSCE( GLenum target, GLuint levels,
// XXX check pitch restrictions ?
rglGcmTextureLayout newLayout;
rglPlatformChooseGPUFormatAndLayout( texture, true, pitch, &newLayout );
rglImage *image = texture->image + texture->baseLevel;
newLayout.levels = 1;
newLayout.faces = texture->faceCount;
newLayout.baseWidth = image->width;
newLayout.baseHeight = image->height;
newLayout.baseDepth = image->depth;
newLayout.internalFormat = ( rglGcmEnum )image->internalFormat;
newLayout.pixelBits = rglPlatformGetBitsPerPixel( newLayout.internalFormat );
newLayout.pitch = pitch ? pitch : GET_TEXTURE_PITCH(texture);
GLboolean isRenderTarget = GL_FALSE;
GLboolean vertexEnable = GL_FALSE;