QSGBatchRenderer: Fix a performance regression noted by qmlbench

0a88774a0f11bf96a87012ac4a83e31ced19460b switched from using client side
pointers to VBOs, ostensibly for core profile support. Unfortunately,
some GPUs/drivers seem to suffer quite a bit with these, so only use
them if we must.

Results for changing_over_isolated_with_clip_rotated.qml:

eskil_linux_tx1:
    Before: Average: 65.6  frames; using  samples; MedianAll=65; StdDev=0.894427, CoV=0.0136346
    After:  Average: 120  frames; using  samples; MedianAll=120; StdDev=0, CoV=0

eskil_linux_focault (nvidia) has a similar regression in all branches
where the original commit is present, too:
    5.6 & v5.8.0 tag: 600
    5.8, 5.9 & dev branches: 399

Assuming this is the sole regression in that case and we end up at the
original 600 again, that would be a 50% gain there, and an 81% increase
on TX1.

Change-Id: I44af9e67698356200f9587e77c9409fdb756519d
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Reviewed-by: Gunnar Sletta <gunnar@crimson.no>

1 files changed, 28 insertions, 11 deletions

diff --git a/src/quick/scenegraph/coreapi/qsgbatchrenderer.cpp b/src/quick/scenegraph/coreapi/qsgbatchrenderer.cpp
index 14f8514289..bf54319db6 100644
--- a/src/quick/scenegraph/coreapi/qsgbatchrenderer.cpp
+++ b/src/quick/scenegraph/coreapi/qsgbatchrenderer.cpp
@@ -2029,6 +2029,15 @@ Renderer::ClipType Renderer::updateStencilClip(const QSGClipNode *clip)
     GLuint vbo = 0;
     int vboSize = 0;
 
+    bool useVBO = false;
+    QOpenGLContext *ctx = QOpenGLContext::currentContext();
+    QSurfaceFormat::OpenGLContextProfile profile = ctx->format().profile();
+
+    if (!ctx->isOpenGLES() && profile == QSurfaceFormat::CoreProfile) {
+        // VBO are more expensive, so only use them if we must.
+        useVBO = true;
+    }
+
     glDisable(GL_SCISSOR_TEST);
 
     m_currentStencilValue = 0;
@@ -2113,20 +2122,27 @@ Renderer::ClipType Renderer::updateStencilClip(const QSGClipNode *clip)
             Q_ASSERT(g->attributeCount() > 0);
             const QSGGeometry::Attribute *a = g->attributes();
 
-            if (!vbo)
-                glGenBuffers(1, &vbo);
+            const GLvoid *pointer;
+            if (!useVBO) {
+                pointer = g->vertexData();
+            } else {
+                if (!vbo)
+                    glGenBuffers(1, &vbo);
 
-            glBindBuffer(GL_ARRAY_BUFFER, vbo);
+                glBindBuffer(GL_ARRAY_BUFFER, vbo);
 
-            const int vertexByteSize = g->sizeOfVertex() * g->vertexCount();
-            if (vboSize < vertexByteSize) {
-                vboSize = vertexByteSize;
-                glBufferData(GL_ARRAY_BUFFER, vertexByteSize, g->vertexData(), GL_STATIC_DRAW);
-            } else {
-                glBufferSubData(GL_ARRAY_BUFFER, 0, vertexByteSize, g->vertexData());
+                const int vertexByteSize = g->sizeOfVertex() * g->vertexCount();
+                if (vboSize < vertexByteSize) {
+                    vboSize = vertexByteSize;
+                    glBufferData(GL_ARRAY_BUFFER, vertexByteSize, g->vertexData(), GL_STATIC_DRAW);
+                } else {
+                    glBufferSubData(GL_ARRAY_BUFFER, 0, vertexByteSize, g->vertexData());
+                }
+
+                pointer = 0;
             }
 
-            glVertexAttribPointer(0, a->tupleSize, a->type, GL_FALSE, g->sizeOfVertex(), 0);
+            glVertexAttribPointer(0, a->tupleSize, a->type, GL_FALSE, g->sizeOfVertex(), pointer);
 
             m_clipProgram.setUniformValue(m_clipMatrixId, m);
             if (g->indexCount()) {
@@ -2135,7 +2151,8 @@ Renderer::ClipType Renderer::updateStencilClip(const QSGClipNode *clip)
                 glDrawArrays(g->drawingMode(), 0, g->vertexCount());
             }
 
-            glBindBuffer(GL_ARRAY_BUFFER, 0);
+            if (useVBO)
+                glBindBuffer(GL_ARRAY_BUFFER, 0);
 
             ++m_currentStencilValue;
         }