[Bf-blender-cvs] [09665a5] temp_display_optimization: GLSL drawing redesign:
Antony Riakiotakis
noreply at git.blender.org
Mon Jun 29 17:46:00 CEST 2015
Commit: 09665a5301c11e99e92fe7ce55f1b6d3d4cc232e
Author: Antony Riakiotakis
Date: Mon Jun 29 12:22:56 2015 +0200
Branches: temp_display_optimization
https://developer.blender.org/rB09665a5301c11e99e92fe7ce55f1b6d3d4cc232e
GLSL drawing redesign:
Use new upload scheme. Idea here is that materials share a VBO where
the size of each element is the maximum size of all mesh materials.
While this will waste some size if material sizes differ it is the
simplest scheme to use and allows easy reuse of indices as opposed to
separating the materials in separate vertex buffers. In fact if we do
that, management gets quite complex and code much more error prone.
I may write an extra blog post to explain the choices here
at some point.
===================================================================
M source/blender/blenkernel/intern/cdderivedmesh.c
M source/blender/gpu/GPU_buffers.h
M source/blender/gpu/intern/gpu_buffers.c
===================================================================
diff --git a/source/blender/blenkernel/intern/cdderivedmesh.c b/source/blender/blenkernel/intern/cdderivedmesh.c
index 8744439..a9ae379 100644
--- a/source/blender/blenkernel/intern/cdderivedmesh.c
+++ b/source/blender/blenkernel/intern/cdderivedmesh.c
@@ -559,8 +559,8 @@ static void cdDM_drawFacesTex_common(DerivedMesh *dm,
orig = DM_origindex_mface_mpoly(index_mf_to_mpoly, index_mp_to_orig, actualFace);
if (orig == ORIGINDEX_NONE) {
/* XXX, this is not really correct
- * it will draw the previous faces context for this one when we don't know its settings.
- * but better then skipping it altogether. - campbell */
+ * it will draw the previous faces context for this one when we don't know its settings.
+ * but better then skipping it altogether. - campbell */
draw_option = DM_DRAW_OPTION_NORMAL;
}
else if (drawParamsMapped) {
@@ -577,7 +577,7 @@ static void cdDM_drawFacesTex_common(DerivedMesh *dm,
if (!flush && compareDrawOptions) {
/* also compare draw options and flush buffer if they're different
- * need for face selection highlight in edit mode */
+ * need for face selection highlight in edit mode */
flush |= compareDrawOptions(userData, actualFace, next_actualFace) == 0;
}
@@ -872,11 +872,9 @@ static void cddm_draw_attrib_vertex(DMVertexAttribs *attribs, const MVert *mvert
}
typedef struct {
- GPUBuffer *buffers;
- char *varray;
DMVertexAttribs attribs;
+ int elementsize; /* size of one vertex for the format of this material */
int numdata;
- int elementsize;
GPUAttrib datatypes[GPU_MAX_ATTRIB]; /* TODO, messing up when switching materials many times - [#21056]*/
} GPUMaterialConv;
@@ -1002,10 +1000,13 @@ static void cdDM_drawMappedFacesGLSL(DerivedMesh *dm,
else {
GPUMaterialConv *matconv;
int offset;
- int totloops = 0;
int totface = dm->getNumTessFaces(dm);
int *mat_orig_to_new;
int tot_active_mat;
+ GPUBuffer *buffer = NULL;
+ char *varray;
+ int max_element_size = 0;
+ int tot_loops = 0;
const MFace *mf = mface;
@@ -1030,128 +1031,133 @@ static void cdDM_drawMappedFacesGLSL(DerivedMesh *dm,
do_draw = setMaterial(new_matnr + 1, &gattribs);
if (do_draw) {
- matconv[a].numdata = 0;
+ int numdata = 0;
DM_vertex_attributes_from_gpu(dm, &gattribs, &matconv[a].attribs);
if (matconv[a].attribs.totorco && matconv[a].attribs.orco.array) {
- matconv[a].datatypes[matconv[a].numdata].index = matconv[a].attribs.orco.gl_index;
- matconv[a].datatypes[matconv[a].numdata].size = 3;
- matconv[a].datatypes[matconv[a].numdata].type = GL_FLOAT;
- matconv[a].numdata++;
+ matconv[a].datatypes[numdata].index = matconv[a].attribs.orco.gl_index;
+ matconv[a].datatypes[numdata].size = 3;
+ matconv[a].datatypes[numdata].type = GL_FLOAT;
+ numdata++;
}
for (b = 0; b < matconv[a].attribs.tottface; b++) {
if (matconv[a].attribs.tface[b].array) {
- matconv[a].datatypes[matconv[a].numdata].index = matconv[a].attribs.tface[b].gl_index;
- matconv[a].datatypes[matconv[a].numdata].size = 2;
- matconv[a].datatypes[matconv[a].numdata].type = GL_FLOAT;
- matconv[a].numdata++;
+ matconv[a].datatypes[numdata].index = matconv[a].attribs.tface[b].gl_index;
+ matconv[a].datatypes[numdata].size = 2;
+ matconv[a].datatypes[numdata].type = GL_FLOAT;
+ numdata++;
}
}
for (b = 0; b < matconv[a].attribs.totmcol; b++) {
if (matconv[a].attribs.mcol[b].array) {
- matconv[a].datatypes[matconv[a].numdata].index = matconv[a].attribs.mcol[b].gl_index;
- matconv[a].datatypes[matconv[a].numdata].size = 4;
- matconv[a].datatypes[matconv[a].numdata].type = GL_UNSIGNED_BYTE;
- matconv[a].numdata++;
+ matconv[a].datatypes[numdata].index = matconv[a].attribs.mcol[b].gl_index;
+ matconv[a].datatypes[numdata].size = 4;
+ matconv[a].datatypes[numdata].type = GL_UNSIGNED_BYTE;
+ numdata++;
}
}
if (matconv[a].attribs.tottang && matconv[a].attribs.tang.array) {
- matconv[a].datatypes[matconv[a].numdata].index = matconv[a].attribs.tang.gl_index;
- matconv[a].datatypes[matconv[a].numdata].size = 4;
- matconv[a].datatypes[matconv[a].numdata].type = GL_FLOAT;
- matconv[a].numdata++;
+ matconv[a].datatypes[numdata].index = matconv[a].attribs.tang.gl_index;
+ matconv[a].datatypes[numdata].size = 4;
+ matconv[a].datatypes[numdata].type = GL_FLOAT;
+ numdata++;
}
- if (matconv[a].numdata != 0) {
- matconv[a].elementsize = GPU_attrib_element_size(matconv[a].datatypes, matconv[a].numdata);
-
- matconv[a].buffers = GPU_buffer_alloc(matconv[a].elementsize * dm->drawObject->tot_loop_verts, false);
-
- if (matconv[a].buffers == NULL) {
- matconv[a].buffers = GPU_buffer_alloc(matconv[a].elementsize * dm->drawObject->tot_loop_verts, true);
- }
- matconv[a].varray = GPU_buffer_lock_stream(matconv[a].buffers);
- if (matconv[a].varray == NULL) {
- GPU_buffer_unbind();
- GPU_buffer_free(matconv[a].buffers);
- MEM_freeN(mat_orig_to_new);
- MEM_freeN(matconv);
- fprintf(stderr, "Out of memory, can't draw object\n");
- return;
- }
+ if (numdata != 0) {
+ matconv[a].elementsize = GPU_attrib_element_size(matconv[a].datatypes, numdata);
+ max_element_size = max_ii(matconv[a].elementsize, max_element_size);
}
}
}
- mface = mf;
- /* part two, fill the arrays with the data */
- for (a = 0; a < totface; a++, mface++) {
- int i = mat_orig_to_new[mface->mat_nr];
+ /* part two, generate and fill the arrays with the data */
+ if (max_element_size > 0) {
+ buffer = GPU_buffer_alloc(max_element_size * dm->drawObject->tot_loop_verts, false);
- if (do_draw && matconv[i].numdata != 0) {
- int elementsize = matconv[i].elementsize;
- offset = 0;
- if (matconv[i].attribs.totorco && matconv[i].attribs.orco.array) {
- copy_v3_v3((float *)&matconv[i].varray[elementsize * totloops], (float *)matconv[i].attribs.orco.array[mface->v1]);
- copy_v3_v3((float *)&matconv[i].varray[elementsize * totloops + elementsize], (float *)matconv[i].attribs.orco.array[mface->v2]);
- copy_v3_v3((float *)&matconv[i].varray[elementsize * totloops + elementsize * 2], (float *)matconv[i].attribs.orco.array[mface->v3]);
- if (mface->v4) {
- copy_v3_v3((float *)&matconv[i].varray[elementsize * totloops + elementsize * 3], (float *)matconv[i].attribs.orco.array[mface->v4]);
- }
- offset += sizeof(float) * 3;
- }
- for (b = 0; b < matconv[i].attribs.tottface; b++) {
- if (matconv[i].attribs.tface[b].array) {
- MTFace *tf = &matconv[i].attribs.tface[b].array[a];
- copy_v2_v2((float *)&matconv[i].varray[elementsize * totloops + offset], tf->uv[0]);
- copy_v2_v2((float *)&matconv[i].varray[elementsize * totloops + offset + elementsize], tf->uv[1]);
- copy_v2_v2((float *)&matconv[i].varray[elementsize * totloops + offset + elementsize * 2], tf->uv[2]);
+ if (buffer == NULL) {
+ buffer = GPU_buffer_alloc(max_element_size * dm->drawObject->tot_loop_verts, true);
+ }
+ varray = GPU_buffer_lock_stream(buffer);
+ if (varray == NULL) {
+ GPU_buffer_unbind();
+ GPU_buffer_free(buffer);
+ MEM_freeN(mat_orig_to_new);
+ MEM_freeN(matconv);
+ fprintf(stderr, "Out of memory, can't draw object\n");
+ return;
+ }
+
+ mface = mf;
+
+ for (a = 0; a < totface; a++, mface++) {
+ int i = mat_orig_to_new[mface->mat_nr];
+ offset = tot_loops * max_element_size;
+
+ if (do_draw && matconv[i].elementsize != 0) {
+ int elementsize = max_element_size;
+ if (matconv[i].attribs.totorco && matconv[i].attribs.orco.array) {
+ copy_v3_v3((float *)&varray[offset], (float *)matconv[i].attribs.orco.array[mface->v1]);
+ copy_v3_v3((float *)&varray[offset + elementsize], (float *)matconv[i].attribs.orco.array[mface->v2]);
+ copy_v3_v3((float *)&varray[offset + elementsize * 2], (float *)matconv[i].attribs.orco.array[mface->v3]);
if (mface->v4) {
- copy_v2_v2((float *)&matconv[i].varray[elementsize * totloops + offset + elementsize * 3], tf->uv[3]);
+ copy_v3_v3((float *)&varray[offset + elementsize * 3], (float *)matconv[i].attribs.orco.array[mface->v4]);
}
- offset += sizeof(float) * 2;
+ offset += sizeof(float) * 3;
}
- }
- for (b = 0; b < matconv[i].attribs.totmcol; b++) {
- if (matconv[i].attribs.mcol[b].array) {
- MCol *cp = &matconv[i].attribs.mcol[b].array[a * 4 + 0];
- GLubyte col[4];
- col[0] = cp->b; col[1] = cp->g; col[2] = cp->r; col[3] = cp->a;
- copy_v4_v4_char((char *)&matconv[i].varray[elementsize * totloops + offset], (char *)col);
- cp = &matconv[i].attribs.mcol[b].array[a * 4 + 1];
- col[0] = cp->b; col[1] = cp->g; col[2] = cp->r; col[3] = cp->a;
- copy_v4_v4_char((char *)&matconv[i].varray[elementsize * totloops + offset + elementsize], (char *)col);
- cp = &matconv[i].attribs.mcol[b].array[a * 4 + 2];
- col[0] = cp->b; col[1] = cp->g; col[2] = cp->r; col[3] = cp->a;
- copy_v4_v4_char((char *)&matconv[i].varray[elementsize * totloops + offset + elementsize * 2], (char *)col);
- if (mface->v4) {
- cp = &matconv[i].attribs.mcol[b].array[a * 4 + 3];
+ for (b = 0; b < matconv[i].attribs.tottface; b++) {
+ if (matconv[i].attribs.tface[b].array) {
+ MTFace *tf = &matconv[i].attribs.tface[b].array[a];
+ copy_v2_v2((float *)&varray[offset], tf->uv[0]);
+ copy_v2_v2((float *)&varray[offset + elementsize], tf->uv[1]);
+ copy_v2_v2((float *)&varray[offset + elementsize * 2], tf->uv[2]);
+ if (mface->v4) {
+ copy_v2_v2((float *)&varray[offset + elementsize * 3], tf->uv[3]);
+ }
+ offset += sizeof(float) * 2;
+ }
+ }
+ for (b = 0; b < matconv[i].attribs.totmcol; b++) {
+ if (matconv[i].attribs.mcol[b].array) {
+ MCol *cp = &matconv[i].attribs.mcol[b].array[a * 4 + 0];
+ GLubyte col[4];
+ col[0] = cp->b; col[1] = cp->g; col[2] = cp->r; col[3] = cp->a;
+ copy_v4_v4_char((char *)&varray[offset], (char *
@@ Diff output truncated at 10240 characters. @@
More information about the Bf-blender-cvs
mailing list