Project4 Submission

.ycm_extra_conf.py

@@ -0,0 +1,139 @@
+# This file is NOT licensed under the GPLv3, which is the license for the rest
+# of YouCompleteMe.
+#
+# Here's the license text for this file:
+#
+# This is free and unencumbered software released into the public domain.
+#
+# Anyone is free to copy, modify, publish, use, compile, sell, or
+# distribute this software, either in source code form or as a compiled
+# binary, for any purpose, commercial or non-commercial, and by any
+# means.
+#
+# In jurisdictions that recognize copyright laws, the author or authors
+# of this software dedicate any and all copyright interest in the
+# software to the public domain. We make this dedication for the benefit
+# of the public at large and to the detriment of our heirs and
+# successors. We intend this dedication to be an overt act of
+# relinquishment in perpetuity of all present and future rights to this
+# software under copyright law.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+# IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.
+#
+# For more information, please refer to <http://unlicense.org/>
+
+import os
+import ycm_core
+
+# These are the compilation flags that will be used in case there's no
+# compilation database set (by default, one is not set).
+# CHANGE THIS LIST OF FLAGS. YES, THIS IS THE DROID YOU HAVE BEEN LOOKING FOR.
+flags = [
+'-lglut',
+'-lGL',
+'-lGLEW',
+'-Wall',
+'-Wextra',
+'-Werror',
+'-Wc++98-compat',
+'-Wno-long-long',
+'-Wno-variadic-macros',
+'-fexceptions',
+'-DNDEBUG',
+'-DUSE_CLANG_COMPLETER',
+# THIS IS IMPORTANT! Without a "-std=<something>" flag, clang won't know which
+# language to use when compiling headers. So it will guess. Badly. So C++
+# headers will be compiled as C headers. You don't want that so ALWAYS specify
+# a "-std=<something>".
+# For a C project, you would set this to something like 'c99' instead of
+# 'c++11'.
+'-std=c++11',
+# ...and the same thing goes for the magic -x option which specifies the
+# language that the files to be compiled are written in. This is mostly
+# relevant for c++ headers.
+# For a C project, you would set this to 'c' instead of 'c++'.
+'-x',
+'c++',
+'-I/usr/local/cuda/samples/common/inc',
+'-I/usr/local/cuda/include'
+]
+
+# Set this to the absolute path to the folder (NOT the file!) containing the
+# compile_commands.json file to use that instead of 'flags'. See here for
+# more details: http://clang.llvm.org/docs/JSONCompilationDatabase.html
+#
+# Most projects will NOT need to set this to anything; you can just change the
+# 'flags' list of compilation flags. Notice that YCM itself uses that approach.
+compilation_database_folder = ''
+
+if compilation_database_folder:
+  database = ycm_core.CompilationDatabase( compilation_database_folder )
+else:
+  database = None
+
+
+def DirectoryOfThisScript():
+  return os.path.dirname( os.path.abspath( __file__ ) )
+
+
+def MakeRelativePathsInFlagsAbsolute( flags, working_directory ):
+  if not working_directory:
+    return list( flags )
+  new_flags = []
+  make_next_absolute = False
+  path_flags = [ '-isystem', '-I', '-iquote', '--sysroot=' ]
+  for flag in flags:
+    new_flag = flag
+
+    if make_next_absolute:
+      make_next_absolute = False
+      if not flag.startswith( '/' ):
+        new_flag = os.path.join( working_directory, flag )
+
+    for path_flag in path_flags:
+      if flag == path_flag:
+        make_next_absolute = True
+        break
+
+      if flag.startswith( path_flag ):
+        path = flag[ len( path_flag ): ]
+        new_flag = path_flag + os.path.join( working_directory, path )
+        break
+
+    if new_flag:
+      new_flags.append( new_flag )
+  return new_flags
+
+
+def FlagsForFile( filename ):
+  if database:
+    # Bear in mind that compilation_info.compiler_flags_ does NOT return a
+    # python list, but a "list-like" StringVec object
+    compilation_info = database.GetCompilationInfoForFile( filename )
+    final_flags = MakeRelativePathsInFlagsAbsolute(
+      compilation_info.compiler_flags_,
+      compilation_info.compiler_working_dir_ )
+
+    # NOTE: This is just for YouCompleteMe; it's highly likely that your project
+    # does NOT need to remove the stdlib flag. DO NOT USE THIS IN YOUR
+    # ycm_extra_conf IF YOU'RE NOT 100% YOU NEED IT.
+    '''
+    try:
+      final_flags.remove( '-stdlib=libc++' )
+    except ValueError:
+      pass
+    '''
+  else:
+    relative_to = DirectoryOfThisScript()
+    final_flags = MakeRelativePathsInFlagsAbsolute( flags, relative_to )
+
+  return {
+    'flags': final_flags,
+    'do_cache': True
+  }

PROJ4_NIX/makefile

@@ -1,6 +1,6 @@
 #All cuda paths are for v5.0+
 
-NVCC     = nvcc -m64
+NVCC     = nvcc -m64 -arch=sm_13
 CC       = gcc -m64
 CPPC     = g++ -m64
 

README.md

@@ -1,3 +1,130 @@
+![cow_shade_fill](renders/cow_shade_fill.png)
+
+-------------------------------------------------------------------------------
+CIS565: Project 4: CUDA Rasterizer
+-------------------------------------------------------------------------------
+Fall 2013
+
+For this project I've implemented the following features:
+* Vertex Shading
+* Primitive Assembly with support for triangle VBOs/IBOs
+* Perspective Transformation
+* Rasterization through either a scanline or a tiled approach
+* Fragment Shading
+* A depth buffer for storing and depth testing fragments
+* Fragment to framebuffer writing
+* Lambert shading in the fragment shader
+
+
+And the following additional features:
+* Correct color interpolation ( I don't expect this feature to add much computational effort since I'm already performing barycentric coordinate based computations for normals, depth, etc ... )
+* Mouse based interactive camera
+* Back-face culling
+
+Color interpolation ( shaded and non-shaded ):
+![cow_interp](renders/cow_interp.png)
+![cow_interp_fill](renders/cow_shade_interp.png)
+
+I also implemented some simple debugging modes:
+
+Normals:
+
+![cow_normals](renders/cow_smooth_normals.png)
+
+Featureless fill:
+
+![cow_fill](renders/cow_fill.png)
+
+
+-------------------------------------------------------------------------------
+Video:
+-------------------------------------------------------------------------------
+Here is a video of me showing off the interactive camera and viewing modes
+
+http://youtu.be/CzJuF8zyuok
+
+The camera is a bit awkward to use, but its a nice start. 
+* left-click: rotate-x/y
+* left-click-shift: pan-x/y
+* left-click-ctrl: zoom / rotate-z
+
+
+-------------------------------------------------------------------------------
+Performance:
+-------------------------------------------------------------------------------
+
+TileSize vs Framerate for cow_smooth2.obj:
+* 1   | 12
+* 2   | 14
+* 4   | 15
+* 8   | 16
+* 16  | 17
+* 32  | 18
+* 64  | 19
+* 128 | 18
+* 256 | 18
+* 512 | 17
+
+A tilesize of 64 is the sweeet spot leading to maximum framerate. It seems that
+the tilesize doesn't have a huge effect on framerate except towards the smaller
+end. In my implementation memory is allocated/freed every frame so I feel that
+a performance improvement can be gained there. 
+
+The greatest performance hit occurs as individual triangles take up larger 
+amounts of the screen size since my implementation has a single thread 
+per triangle bounding box. A performance improvement could be gained there
+by dividing that bounding box into smaller tiles and allocated a thread per 
+tile. 
+
+Finally, I just use flags to indicate if a primitive should be processed or not. 
+I think performance could be improved by not wasting threads by performing
+stream compaction on the primitives. 
+
+#### Color interpolation
+Removing all barycentric operations leads to the same 19fps for a tilesize of 64.
+When it comes down to it the barycentric coordinate computations cosist of a
+few multiplies.
+
+#### Back-face culling
+Without back-face culling the framerate drops to 17fps for a tilesize of 64. 
+This is expected, and awesome! 
+
+#### Interactive camera
+The primary losses with regards to the interactive camera come from the user
+being able to get too-close to the model triangles leading to large performance 
+losses as mentioned above
+
+
+-------------------------------------------------------------------------------
+Thanks:
+-------------------------------------------------------------------------------
+Thanks to Nathan Marshak for providing the cow_smoothed2.obj cow.obj with smoothed normals
+
+
+-------------------------------------------------------------------------------
+First Steps
+-------------------------------------------------------------------------------
+![first_steps](renders/first_steps.png)
+
+I've implemented the first steps of the rasterizer and the above image shows
+the progress I've made so far
+- Vertex Shader with identity model-view-projection matrix and perspective division.
+  I'm not sure if this should be done here but it works
+- Primitive Assembly
+- Rasterization, using the triangle AABB and barycentric coordinates to check
+  if the pixel is inside the triangle. I also added linear vertex color 
+  interpolation in window space using the barycentric coordinates because its 
+  pretty 
+
+
+Theres still a lot to do:
+- Actual model-view-project transformations, hopefully using interactive camera :)
+- Fragment Shading
+- Depth Buffer
+- Lighting
+
+I still haven't decided what additional features would be cool.
+
 -------------------------------------------------------------------------------
 CIS565: Project 4: CUDA Rasterizer
 -------------------------------------------------------------------------------