Build test fails when trying to use /usr/local/cuda/bin/nvcc

[GoogleCodeExporter]

What steps will reproduce the problem?
1. sudo ./build.py --test_lists parboil --test_level=basic
2.
3.

What is the expected output? What do you see instead?
Build of the test

Building rules for Ocelot CUDA 4.1 SDK....
 Imported environment from higher level SCons script.
Building rules for Parboil Benchmark suite....
 Imported environment from higher level SCons script.
scons: building associated VariantDir targets: .release_build/ocelot
config_h_build([".release_build/ocelot/configure.h"], ["ocelot/configure.h.in"]
/usr/local/cuda/bin/nvcc -arch=sm_20 ocelot/ocelot/cuda/test/globals/global.cu 
-c -o .release_build/ocelot/ocelot/cuda/test/globals/global.o
sh: 1: /usr/local/cuda/bin/nvcc: not found

What version of the product are you using? On what operating system?
svn checkout http://gpuocelot.googlecode.com/svn/trunk/ gpuocelot-­read-­only 
        gpuocelot - Revision 2216
Ubuntu 13.04

Please provide any additional information below.
There are nothing installed in /usr/local/cuda/bin/nvcc directory, i am 
actually a little confused if ocelot should install something in this folder.
All required compilers and tools are installed. LLVM is installed and Boost. 
LibGlew, LibGlut work fine. And the build "sudo build.py ­­--install" was 
compiles without errors although "sudo python build.py ­­--install" was 
needed. /usr/local/lib/libocelot.so is present. 

Original issue reported on code.google.com by kasper.S...@gmail.com on 29 Apr 2013 at 10:52