hmn
/
orca
5
Fork 2
Code Issues 21 Pull Requests 1 Packages Projects Releases Wiki Activity

[wip, win32, canvas] dispatch path setup, segment setup and backprop in batches of GL_MAX_COMPUTE_WORK_GROUP_COUNT elements

This commit is contained in:
martinfouilleul 2023-07-03 16:16:39 +02:00
parent e24a872fad
commit 311cef65e8
1 changed files with 38 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

42
src/gl_canvas.c
View File

@ -76,6 +76,7 @@ typedef struct mg_gl_path_queue
{ {
vec4 area; vec4 area;
int tileQueues; int tileQueues;
u8 pad[12];
} mg_gl_path_queue; } mg_gl_path_queue;
typedef struct mg_gl_tile_op typedef struct mg_gl_tile_op
@ -895,24 +896,36 @@ void mg_gl_render_batch(mg_gl_canvas_backend* backend,
glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT); glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
//NOTE: path setup pass //NOTE: path setup pass
int maxWorkGroupCount = 0;
glGetIntegeri_v(GL_MAX_COMPUTE_WORK_GROUP_COUNT, 0, &maxWorkGroupCount);
//NOTE: glDispatchCompute errors if work group count is greater _or equal_ to GL_MAX_COMPUTE_WORK_GROUP_COUNT
// so the maximum _allowed_ group count is one less.
maxWorkGroupCount--;
glUseProgram(backend->pathSetup); glUseProgram(backend->pathSetup);
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, 0, backend->pathBuffer, backend->pathBufferOffset, pathCount*sizeof(mg_gl_path));
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->pathQueueBuffer);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->tileQueueCountBuffer); glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->tileQueueCountBuffer);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->tileQueueBuffer); glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->tileQueueBuffer);
glUniform1i(0, tileSize); glUniform1i(0, tileSize);
glUniform1f(1, scale); glUniform1f(1, scale);
glDispatchCompute(pathCount, 1, 1); for(int i=0; i<pathCount; i += maxWorkGroupCount)
{
int pathOffset = backend->pathBufferOffset + i*sizeof(mg_gl_path);
int pathQueueOffset = i*sizeof(mg_gl_path_queue);
int count = minimum(maxWorkGroupCount, pathCount-i);
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, 0, backend->pathBuffer, pathOffset, count*sizeof(mg_gl_path));
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, 1, backend->pathQueueBuffer, pathQueueOffset, count*sizeof(mg_gl_path_queue));
glDispatchCompute(count, 1, 1);
glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT); glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
}
//NOTE: segment setup pass //NOTE: segment setup pass
glUseProgram(backend->segmentSetup); glUseProgram(backend->segmentSetup);
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, 0, backend->elementBuffer, backend->elementBufferOffset, eltCount*sizeof(mg_gl_path_elt));
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->segmentCountBuffer); glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->segmentCountBuffer);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->segmentBuffer); glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->segmentBuffer);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->pathQueueBuffer); glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->pathQueueBuffer);
@ -923,17 +936,32 @@ void mg_gl_render_batch(mg_gl_canvas_backend* backend,
glUniform1f(0, scale); glUniform1f(0, scale);
glUniform1ui(1, tileSize); glUniform1ui(1, tileSize);
glDispatchCompute(eltCount, 1, 1); for(int i=0; i<eltCount; i += maxWorkGroupCount)
{
int offset = backend->elementBufferOffset + i*sizeof(mg_gl_path_elt);
int count = minimum(maxWorkGroupCount, eltCount-i);
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, 0, backend->elementBuffer, offset, count*sizeof(mg_gl_path_elt));
glDispatchCompute(count, 1, 1);
glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT); glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
}
//NOTE: backprop pass //NOTE: backprop pass
glUseProgram(backend->backprop); glUseProgram(backend->backprop);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, backend->pathQueueBuffer);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->tileQueueBuffer); glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->tileQueueBuffer);
glDispatchCompute(pathCount, 1, 1); for(int i=0; i<pathCount; i += maxWorkGroupCount)
{
int offset = i*sizeof(mg_gl_path_queue);
int count = minimum(maxWorkGroupCount, pathCount-i);
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, 0, backend->pathQueueBuffer, offset, count*sizeof(mg_gl_path_queue));
glDispatchCompute(count, 1, 1);
glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT); glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
}
//NOTE: merge pass //NOTE: merge pass
glUseProgram(backend->merge); glUseProgram(backend->merge);