Merge pull request #84 from tribal-tec/master

Fix #73: Merge data- and textureDataCache, rename commandline options

doc/Changelog.md

@@ -1,30 +1,40 @@
 Changelog {#Changelog}
 =========
 
-## Release 0.4 (git master) {Release040}
+# Release 0.4 (git master) {#Release040}
 
+* [#84](https://github.com/BlueBrain/Livre/pull/84):
+  Cache statistics are printed when showing statistics (press 's' key)
+* [#84](https://github.com/BlueBrain/Livre/pull/84):
+  Fix [#73](https://github.com/BlueBrain/Livre/issues/73): Merge data- and
+  textureDataCache, commandline options for changing the cache sizes are now
+  --gpu-cache-mem (instead of --texture-cache-mem) and --cpu-cache-mem (instead
+  of --data-cache-mem and --texturedata-cache-mem)
 * [#78](https://github.com/BlueBrain/Livre/pull/78):
   New command line parameter to load a transfer function from file (".1dt" file
   extension, based on the format used by ImageVis3D)
 * [#75](https://github.com/BlueBrain/Livre/pull/75):
   Separate ZeroEQ communication to class zeq::Communicator
 
-## Release 0.3 (2015-07-07) {Release030}
+# Release 0.3 (2015-07-07) {#Release030}
 
 * [#37](https://github.com/BlueBrain/Livre/pull/37):
   New command line option to enable synchronous mode (wait until every block has
   been processed before rendering them)
 * [#33](https://github.com/BlueBrain/Livre/pull/33):
   New command line option to enable ZeroEQ camera synchronization
 * [#44](https://github.com/BlueBrain/Livre/pull/44):
-  New command line option to specify a range of frames to render from the command line
+  New command line option to specify a range of frames to render from the
+  command line
 * [#57](https://github.com/BlueBrain/Livre/pull/57):
-  New command line parameter to set the camera lookAt vector to define its orientation
+  New command line parameter to set the camera lookAt vector to define its
+  orientation
 * New command line parameter to set camera position
 * [#17](https://github.com/BlueBrain/Livre/pull/17):
   New command line parameters for RESTBridge (hostname, port, ZeroEQ schema)
 * [#53](https://github.com/BlueBrain/Livre/pull/53):
-  New command line parameters to specify the minimum and maximum levels of detail to render
+  New command line parameters to specify the minimum and maximum levels of
+  detail to render
 * New heartbeat messaging mechanism
 * [#43](https://github.com/BlueBrain/Livre/pull/43):
   New livre_batch.py script for offline image series rendering
@@ -44,7 +54,8 @@ Changelog {#Changelog}
 * Fix broken memory mapped reading in UVF
 * Fix GLEW_MX linking
 * [#45](https://github.com/BlueBrain/Livre/pull/45):
-  Fix the "Empty servus implementation" exception that was raised when zeroconf wasn not available
+  Fix the "Empty servus implementation" exception that was raised when zeroconf
+  was not available
 * [#58](https://github.com/BlueBrain/Livre/pull/58):
   Fix the LOD computation based on the screen height
 * [#33](https://github.com/BlueBrain/Livre/pull/33):
@@ -53,14 +64,16 @@ Changelog {#Changelog}
   Graceful exit when ZeroEQ EXIT event is received
 * Improve glslshader/raycastrenderer error handling and API
 * [#28](https://github.com/BlueBrain/Livre/pull/28):
-  Improve performance and memory usage through a new implicit, flat octree implementation
+  Improve performance and memory usage through a new implicit, flat octree
+  implementation
 * [#48](https://github.com/BlueBrain/Livre/pull/48):
   Improve rendering performance through frustum culling
 * Optimization: use lunchbox::Buffer instead of std::vector in MemoryUnit and
   MemoryDataSource (+35% speed in remote datasource test)
 * [#34](https://github.com/BlueBrain/Livre/pull/34):
-  Refactor the code to setup the data and texture UploadProcessors, removing race
-  conditions and fixing crashes (at exit and with multiple local GPUs) in the process
+  Refactor the code to setup the data and texture UploadProcessors, removing
+  race conditions and fixing crashes (at exit and with multiple local GPUs) in
+  the process
 * [#35](https://github.com/BlueBrain/Livre/pull/35)
   [#36](https://github.com/BlueBrain/Livre/pull/36):
   Several fixes in the tree creation and the LOD-selection algorithm
@@ -70,9 +83,9 @@ Changelog {#Changelog}
   Use one GPU by default
 * Use ZeroEQ for the transfer function editor GUI
 * [#70](https://github.com/BlueBrain/Livre/pull/70):
-  Window title has been improved to show the application name and its version number
+  Window title has been improved to show the application name and its version
+  number
 
-
-## Release 0.2 (2014-10-15) {Release020}
+# Release 0.2 (2014-10-15) {#Release020}
 
 * BBP internal release

doc/user-guide.dox

@@ -18,16 +18,17 @@
 
 @subsection Overview Overview
 
-Livre is a out-of-core volume rendering engine depending Equalizer and OpenGL for rendering, and
-Tuvok library for data management.
+Livre is a out-of-core volume rendering engine depending Equalizer and OpenGL
+for rendering, and Tuvok library for data management.
 
 @section INSTALLATION Installation
 
 @subsection Setup Setup
 
-Livre, currently doesn't have Ubuntu or Redhat packages. It should be build from the source code at
-github.(The compiler should support the C++11 - minimum g++ 4.6 - to compile the project because of
-the Tuvok library dependency). The basic building procedure is as below:
+Livre, currently doesn't have Ubuntu or Redhat packages. It should be build from
+the source code at github.(The compiler should support the C++11 - minimum g++
+4.6 - to compile the project because of the Tuvok library dependency). The basic
+building procedure is as below:
 
 \code
 $ git clone https://github.com/BlueBrain/Livre.git
@@ -39,10 +40,11 @@ $ make
 $ make install
 \endcode
 
-The above commands will fetch the necessary packages and build the dependencies, and install the
-executables and libraries.
+The above commands will fetch the necessary packages and build the dependencies,
+and install the executables and libraries.
 
-After a successful build and installation, the following executables will be available:
+After a successful build and installation, the following executables will be
+available:
 
 - livre (The command line render application)
 - livre_batch.py (Python script to launch livre in batch mode using sbatch to
@@ -55,21 +57,23 @@ After a successful build and installation, the following executables will be ava
 
 @subsection Pre-processing_Tools Volume pre-processing tools
 
-The volumetric data should be pre-processed in order to visualize in Livre. More information about
-the pipeline of preprocessing can be found at @ref Tuvok_Tool tools page.
+The volumetric data should be pre-processed in order to visualize in Livre. More
+information about the pipeline of preprocessing can be found at
+@ref Tuvok_Tool tools page.
 
 @subsection Livre Livre
 
-The volume renderer is executed through the command line. To test if the volume renderer is working
-or not, you can simply execute :
+The volume renderer is executed through the command line. To test if the volume
+renderer is working, you can simply execute :
 
 \code
-livre [--volume mem://1024,1024,512,32]
+livre --volume mem://1024,1024,512,32
 \endcode
 
-which will render an in memory test volume data structure. Here 1024, 1024 and 512 are the width,
-height and length of the volume in voxels, respectively. 32 is the block size and "mem://" is the
-scheme to generate in memory volume. You should be getting an image like below:
+which will render an in memory test volume data structure. Here 1024, 1024 and
+512 are the width, height and length of the volume in voxels, respectively. 32
+is the block size and "mem://" is the scheme to generate in memory volume. You
+should be getting an image like below:
 
 @image html livre.png "Livre"
 
@@ -82,60 +86,61 @@ Application Parameters
                                         delta for animation speed, use
                                         --animation=-<int> for reverse
                                         animation)
-  --frames arg (=[0 4294967295])
+  --frames arg (=[          0 4294967295 ])
                                         Frames to render 'start end'
   --num-frames arg (=4294967295)        Maximum nuber of frames to render
-  --camera-position arg (=[0 0 2])
+  --camera-position arg (=[          0          0          2 ])
                                         Camera position
-  --camera-lookat arg (=[0 0 0])
+  --camera-lookat arg (=[          0          0          0 ])
                                         Camera orientation
   --sync-camera                         Synchronize camera with other
                                         applications
 
 REST Parameters
-  --rest-host arg (=localhost)          RESTBridge's http hostname
-  --rest-port arg (=4020)               RESTBridge's http port
-  --zeq-schema arg (=rest)              RESTBridge's internal zeq schema
-  --use-rest                            Use RESTBridge
+  --rest-host arg (=localhost) RESTBridge's http hostname
+  --rest-port arg (=4020)      RESTBridge's http port
+  --zeq-schema arg (=rest)     RESTBridge's internal zeq schema
+  --use-rest                   Use RESTBridge
 
 Volume Renderer Parameters
-  --data-cache-mem arg (=1024)          Maximum data cache memory (MB) - caches
-                                        the raw data read from I/O in system
-                                        memory
-  --texture-cache-mem arg (=3072)       Maximum texture cache memory (MB) -
-                                        caches the texture data on GPU memory
-  --texturedata-cache-mem arg (=8192)   Maximum texture data cache memory (MB) -
-                                        caches the data that has been converted
-                                        into internal texture format in system
-                                        memory
-  --sse arg (=4)                        Screen space error
-  --synchronous                         Enable synchronous mode
-  --min-lod arg (=0)                    Minimum level of detail
-  --max-lod arg (=9)                    Maximum level of detail
+  --gpu-cache-mem arg (=3072)  Maximum GPU cache memory (MB) - caches the
+                               texture data in GPU memory
+  --cpu-cache-mem arg (=8192)  Maximum CPU cache memory (MB) - caches the
+                               volume data in CPU memory
+  --sse arg (=4)               Screen space error
+  --synchronous                Enable synchronous mode
+  --min-lod arg (=0)           Minimum level of detail
+  --max-lod arg (=9)           Maximum level of detail
+  --samples-per-ray arg (=512) Number of samples per ray
+  --transfer-function arg      Path to transfer function file
 \endcode
 
 The important parameters are :
 
 - "--volume" URL of the volume
-- "--sse" Quality of the rendering, the lower the value, the higher the quality. If value is 1,
-1 pixel is 1 voxel, for lower quality values should be larger than 1 (4 by default)
+- "--sse" Quality of the rendering, the lower the value, the higher the quality.
+  If value is 1, 1 pixel is 1 voxel, for lower quality values should be bigger
+  than 1 (4 by default).
 
 @subsection LivreGUI Transfer function editor
 
-Transfer function editor helps the users to attach different colors and opacities to voxel values.
-The "livreGUI" is the main application for editing the transfer function. It can connect the "livre
-"application remotely and can modify the colors. Each color channel can be modified separately.
-The starting dialog is as below:
+Transfer function editor helps the users to attach different colors and
+opacities to voxel values. The "livreGUI" is the main application for editing
+the transfer function. It can connect the "livre "application remotely and can
+modify the colors. Each color channel can be modified separately. The starting
+dialog is as below:
 
 @image html MainWindow.png "Main Window"
 
-For connection, hit the connect button and enter the hostname where "livre" application is running.
+For connection, hit the connect button and enter the hostname where "livre"
+application is running.
 
 @image html Connection.png "Connection Dialog"
 
-After successfull connection, you can open the transfer function editor. In the plot area x axis
-represents the values of voxel and the y axis represents the opacity values. Each color channel can
-be modified seperately by selecting the channel from the color channels.
+After successfull connection, you can open the transfer function editor. In the
+plot area x axis represents the values of voxel and the y axis represents the
+opacity values. Each color channel can be modified seperately by selecting the
+channel from the color channels.
 
 @image html TransferFunctionEditor.png "Transfer Function Editor"
 

livre/Eq/Channel.cpp

@@ -2,6 +2,7 @@
 /* Copyright (c) 2006-2015, Stefan Eilemann <eile@equalizergraphics.com>
  *                          Maxim Makhinya <maxmah@gmail.com>
  *                          Ahmet Bilgili <ahmet.bilgili@epfl.ch>
+ *                          Daniel Nachbaur <daniel.nachbaur@epfl.ch>
  *
  * This file is part of Livre <https://github.com/BlueBrain/Livre>
  *
@@ -19,36 +20,37 @@
  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  */
 
-#include <livre/Eq/FrameData.h>
-#include <livre/Eq/Settings/FrameSettings.h>
 #include <livre/Eq/Channel.h>
-#include <livre/Eq/Pipe.h>
-#include <livre/Eq/Node.h>
-#include <livre/Eq/Window.h>
-#include <livre/Eq/Error.h>
 #include <livre/Eq/Config.h>
-#include <livre/Eq/Settings/CameraSettings.h>
-#include <livre/Eq/Settings/RenderSettings.h>
+#include <livre/Eq/Error.h>
+#include <livre/Eq/FrameData.h>
+#include <livre/Eq/FrameGrabber.h>
+#include <livre/Eq/Node.h>
+#include <livre/Eq/Pipe.h>
 #include <livre/Eq/Render/EqContext.h>
 #include <livre/Eq/Render/RayCastRenderer.h>
-#include <livre/Eq/FrameGrabber.h>
+#include <livre/Eq/Settings/CameraSettings.h>
+#include <livre/Eq/Settings/FrameSettings.h>
+#include <livre/Eq/Settings/RenderSettings.h>
+#include <livre/Eq/Window.h>
 
-#include <livre/core/DashPipeline/DashProcessorOutput.h>
+#include <livre/core/Dash/DashRenderStatus.h>
+#include <livre/core/Dash/DashTree.h>
 #include <livre/core/DashPipeline/DashProcessorInput.h>
+#include <livre/core/DashPipeline/DashProcessorOutput.h>
+#include <livre/core/Data/VolumeDataSource.h>
+#include <livre/core/Render/Frustum.h>
+#include <livre/core/Render/GLWidget.h>
+#include <livre/core/Render/RenderBrick.h>
 #include <livre/core/Visitor/RenderNodeVisitor.h>
+
+#include <livre/Lib/Cache/TextureCache.h>
+#include <livre/Lib/Cache/TextureDataCache.h>
 #include <livre/Lib/Cache/TextureObject.h>
 #include <livre/Lib/Render/AvailableSetGenerator.h>
 #include <livre/Lib/Render/RenderView.h>
 #include <livre/Lib/Render/ScreenSpaceLODEvaluator.h>
 
-
-#include <livre/core/Render/Frustum.h>
-#include <livre/core/Render/GLWidget.h>
-#include <livre/core/Render/RenderBrick.h>
-#include <livre/core/Dash/DashRenderStatus.h>
-#include <livre/core/Dash/DashTree.h>
-#include <livre/core/Data/VolumeDataSource.h>
-
 #include <eq/eq.h>
 #include <eq/gl.h>
 
@@ -449,7 +451,58 @@ class Channel
 
         FrameSettingsPtr frameSettingsPtr = getFrameData()->getFrameSettings();
         if( frameSettingsPtr->getStatistics( ))
+        {
             _channel->drawStatistics();
+            drawCacheStatistics();
+        }
+    }
+
+    void drawCacheStatistics()
+    {
+        glLogicOp( GL_XOR );
+        glEnable( GL_COLOR_LOGIC_OP );
+        glDisable( GL_LIGHTING );
+        glDisable( GL_DEPTH_TEST );
+
+        glColor3f( 1.f, 1.f, 1.f );
+
+        const eq::PixelViewport& pvp = _channel->getPixelViewport();
+        const eq::Viewport& vp = _channel->getViewport();
+        const float height = pvp.h / vp.h;
+
+        glMatrixMode( GL_PROJECTION );
+        glLoadIdentity();
+        _channel->applyScreenFrustum();
+        glMatrixMode( GL_MODELVIEW );
+
+        livre::Node* node = static_cast< livre::Node* >( _channel->getNode( ));
+        std::ostringstream dataStr;
+        dataStr << node->getTextureDataCache().getStatistics();
+        float y = height/2;
+        _drawText( dataStr.str(), y );
+
+        Window* window = static_cast< Window* >( _channel->getWindow( ));
+        std::ostringstream textureStr;
+        textureStr << window->getTextureCache().getStatistics();
+        y = height/4;
+        _drawText( textureStr.str(), y );
+    }
+
+    void _drawText( std::string text, float y )
+    {
+        const eq::util::BitmapFont* font = _channel->getWindow()->getSmallFont();
+        for( size_t pos = text.find( '\n' ); pos != std::string::npos;
+             pos = text.find( '\n' ))
+        {
+            glRasterPos3f( 10.f, y, 0.99f );
+
+            font->draw( text.substr( 0, pos ));
+            text = text.substr( pos + 1 );
+            y -= 16.f;
+        }
+        // last line
+        glRasterPos3f( 10.f, y, 0.99f );
+        font->draw( text );
     }
 
     void frameFinish()
@@ -582,7 +635,9 @@ void Channel::frameViewStart( const uint128_t& frameId )
 
 void Channel::frameViewFinish( const eq::uint128_t &frameID )
 {
+    setupAssemblyState();
     _impl->frameViewFinish();
+    resetAssemblyState();
     eq::Channel::frameViewFinish( frameID );
     _impl->removeImageListener();
 }