New Due: Friday, March. 19, 6:00 pm
Hyesoon Kim, Instructor
Introduction
This is two-student group assignment.
In this assignment, you will improve a ray tracing algorithm
implementation of CUDA.
Installation Guide
You download Raytracingtutorial. Untar
the tarfile under NVIDIA_GPU_COMPUTING_SKD/C
You also have to install AntTweakBar library.
tar -zxvf AntTweakBar.tar.gz
cd AntTweakBar
cd src
make
After you install the library, you need to modify Makefile to change the include path and (GPU_INCLUDE_DIR) and (GPU_LIB_DIR). The provided Makefile assumes that AntTweakBar library is installed under "/home/comparch/Download/AntTweakBar".
If you install AntTweakBar in a different directory, you change ANTTWEA parameter in Makefile.
If you use a 32-bit machine instead of 64-bit machine, replace -lGLEW_x64_64 with -lGLEW in GPU_LIB in the makefile.
You also need to modify your .cshrch or .bashrc script to include the library path such as "LD_LIB_PATH". For example,
LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64:home/comparch/Download/AntTweakBar/lib
Error messages
error:AntTweakBar.h: No such file or directory
Please check your ANTTWEA is the right director.
libAntTweakBar.so: cannot open shared object file: No such file or directory
Please check your shell script and change your LD_LIBRARY_PATH that includes your AntTweakBar/lib location. If you have changed the shell script, make it sure you open a new shell window (or execute the shell script).
What to do?
The raytracing tutorial is based on the link . Your assignment is implementing two antialising algorithms on top of this ray tracing algorithm and compare them.
Suggesting algorithms are Adaptive Sampling, Stochasitc Sampling.
More suggestions or directions of antialiasing Algorithm will be posted by Monday (3/1/10).
Grading
The grading is based on supporting features and the report.
Basic functionality (you need to demo your program on 3/19 (F) or before). 50%
Performance oriented optimizations (e.g. instead of using for-loops, use multiple threads, use adaptive sampling etc.) 30%
Report (different sampling rate and performance, functionality issues) 20%