multithreaded download and files that probably should have been in the last
commit For testing the new iteration, the texture cache disk download was written into the main thread instead of into the separate threads. Now they're back in separate threads again. Also I think some of these files probably should have been in the previous commit.
This commit is contained in:
@@ -40,7 +40,7 @@ bool OpenGLBackend::InitInternal()
|
||||
|
||||
// Connect to it
|
||||
connect(processor, SIGNAL(RequestSibling(NodeDependency)), this, SLOT(ThreadRequestedSibling(NodeDependency)));
|
||||
connect(processor, SIGNAL(CompletedFrame(NodeDependency)), this, SLOT(CompletedFrame(NodeDependency)));
|
||||
connect(processor, SIGNAL(CompletedFrame(NodeDependency)), this, SLOT(ThreadCompletedFrame(NodeDependency)));
|
||||
|
||||
// Finally, we can move it to its own thread
|
||||
processor->moveToThread(thread);
|
||||
@@ -66,16 +66,21 @@ bool OpenGLBackend::InitInternal()
|
||||
|
||||
void OpenGLBackend::GenerateFrame(const rational &time)
|
||||
{
|
||||
qDebug() << "Compiled state:" << compiled_;
|
||||
if (!compiled_) {
|
||||
Compile();
|
||||
}
|
||||
|
||||
NodeDependency dep = NodeDependency(viewer_node()->texture_input()->get_connected_output(), time, time);
|
||||
|
||||
QMetaObject::invokeMethod(processors_.first(),
|
||||
"Render",
|
||||
Qt::QueuedConnection,
|
||||
Q_ARG(NodeDependency, dep));
|
||||
foreach (OpenGLWorker* worker, processors_) {
|
||||
if (worker->IsAvailable() || worker == processors_.last()) {
|
||||
QMetaObject::invokeMethod(worker,
|
||||
"Render",
|
||||
Qt::QueuedConnection,
|
||||
Q_ARG(NodeDependency, dep));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
void OpenGLBackend::CloseInternal()
|
||||
@@ -196,66 +201,28 @@ bool OpenGLBackend::TraverseCompiling(Node *n)
|
||||
return true;
|
||||
}
|
||||
|
||||
#include <QOpenGLExtraFunctions>
|
||||
#include <OpenImageIO/imageio.h>
|
||||
#include "common/define.h"
|
||||
#include "render/pixelservice.h"
|
||||
void OpenGLBackend::CompletedFrame(NodeDependency path)
|
||||
void OpenGLBackend::ThreadCompletedFrame(NodeDependency path)
|
||||
{
|
||||
caching_ = false;
|
||||
|
||||
OpenGLTexturePtr texture = path.node()->get_cached_value(path.range()).value<OpenGLTexturePtr>();
|
||||
qDebug() << "Retrieved texture for time" << path.in();
|
||||
|
||||
qDebug() << "Texture is" << texture.get();
|
||||
|
||||
if (texture == nullptr) {
|
||||
QOpenGLFunctions* f = QOpenGLContext::currentContext()->functions();
|
||||
QOpenGLExtraFunctions* xf = QOpenGLContext::currentContext()->extraFunctions();
|
||||
|
||||
PixelFormatInfo format_info = PixelService::GetPixelFormatInfo(params().format());
|
||||
QVector<char> data_buffer(PixelService::GetBufferSize(params().format(), params().width(), params().height()));
|
||||
qDebug() << "Created buffer of size" << data_buffer.size();
|
||||
|
||||
// Set up OIIO::ImageSpec for compressing cached images on disk
|
||||
OIIO::ImageSpec spec(params().width(), params().height(), kRGBAChannels, format_info.oiio_desc);
|
||||
spec.attribute("compression", "dwaa:200");
|
||||
|
||||
f->glBindFramebuffer(GL_READ_FRAMEBUFFER, copy_buffer_.buffer());
|
||||
|
||||
xf->glFramebufferTexture2D(GL_READ_FRAMEBUFFER,
|
||||
GL_COLOR_ATTACHMENT0,
|
||||
GL_TEXTURE_2D,
|
||||
texture->texture(),
|
||||
0);
|
||||
|
||||
f->glReadPixels(0,
|
||||
0,
|
||||
texture->width(),
|
||||
texture->height(),
|
||||
format_info.pixel_format,
|
||||
format_info.gl_pixel_type,
|
||||
data_buffer.data());
|
||||
|
||||
xf->glFramebufferTexture2D(GL_READ_FRAMEBUFFER,
|
||||
GL_COLOR_ATTACHMENT0,
|
||||
GL_TEXTURE_2D,
|
||||
0,
|
||||
0);
|
||||
|
||||
f->glBindFramebuffer(GL_READ_FRAMEBUFFER, 0);
|
||||
|
||||
if (texture != nullptr) {
|
||||
QString cache_fn = CachePathName(QStringLiteral("%1-%2").arg(QString::number(path.in().numerator()), QString::number(path.in().denominator())).toLatin1());
|
||||
std::string working_fn_std = cache_fn.toStdString();
|
||||
|
||||
std::unique_ptr<OIIO::ImageOutput> out = OIIO::ImageOutput::create(working_fn_std);
|
||||
|
||||
if (out) {
|
||||
out->open(working_fn_std, spec);
|
||||
out->write_image(format_info.oiio_desc, data_buffer.data());
|
||||
out->close();
|
||||
} else {
|
||||
qWarning() << "Failed to open output file:" << cache_fn;
|
||||
// Find an available worker to download this texture
|
||||
foreach (OpenGLWorker* worker, processors_) {
|
||||
// Check if one is available, but worst case if none of them are available, just queue it on the last worker since
|
||||
// it's the least likely to get work
|
||||
if (worker->IsAvailable() || worker == processors_.last()) {
|
||||
QMetaObject::invokeMethod(worker,
|
||||
"Download",
|
||||
Q_ARG(NodeDependency, path),
|
||||
Q_ARG(QVariant, QVariant::fromValue(texture)),
|
||||
Q_ARG(QString, cache_fn));
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -273,7 +240,6 @@ void OpenGLBackend::ThreadCallback(OpenGLTexturePtr texture, const rational& tim
|
||||
// We received a texture, time to start downloading it
|
||||
QString fn = CachePathName(hash);
|
||||
|
||||
qDebug() << "INSERT DOWNLOAD CODE!";
|
||||
/*download_threads_[last_download_thread_%download_threads_.size()]->Queue(texture,
|
||||
fn,
|
||||
hash);*/
|
||||
@@ -317,12 +283,10 @@ void OpenGLBackend::ThreadCallback(OpenGLTexturePtr texture, const rational& tim
|
||||
|
||||
void OpenGLBackend::ThreadRequestedSibling(NodeDependency dep)
|
||||
{
|
||||
Q_UNUSED(dep)
|
||||
|
||||
// Try to queue another thread to run this dep in advance
|
||||
for (int i=1;i<processors_.size();i++) {
|
||||
if (processors_.at(i)->IsAvailable()) {
|
||||
QMetaObject::invokeMethod(processors_.at(i),
|
||||
foreach (OpenGLWorker* worker, processors_) {
|
||||
if (worker->IsAvailable()) {
|
||||
QMetaObject::invokeMethod(worker,
|
||||
"RenderAsSibling",
|
||||
Qt::QueuedConnection,
|
||||
Q_ARG(NodeDependency, dep));
|
||||
|
||||
Reference in New Issue
Block a user