ScuDesignCenter
Remote access to the Design Center at SCU:
ssh <login-name>@linux.dc.engr.scu.edu
ssh <login-name>@linux60801.dc.engr.scu.edu
# The host name can be any linux608XX, where XX is 01, 02, etc., through 24.
Setup CUDA environment:
setup cuda
Find CUDA toolkit documentation:
ls -1 $CUDA_INSTALL_PATH/doc
CUBLAS_Library.pdf
CUDA_4.0_Readiness_Tech_Brief.pdf
CUDA_C_Best_Practices_Guide.pdf
CUDA_C_Programming_Guide.pdf
CUDA_Developer_Guide_for_Optimus_Platforms.pdf
cuda-gdb.pdf
cuda-memcheck.pdf
CUDA_Toolkit_Reference_Manual.html
CUDA_Toolkit_Reference_Manual.pdf
CUDA_Toolkit_Release_Notes.txt
CUDA_VideoDecoder_Library.pdf
CUFFT_Library.pdf
cuobjdump.pdf
CURAND_Library.pdf
CUSPARSE_Library.pdf
EULA.txt
Fermi_Compatibility_Guide.pdf
Fermi_Tuning_Guide.pdf
html/
NPP_Library.pdf
nvcc.pdf
OpenCL_Best_Practices_Guide.pdf
OpenCL_Implementation_Notes.txt
OpenCL_Jumpstart_Guide.pdf
OpenCL_Programming_Guide.pdf
OpenCL_Programming_Overview.pdf
ptx_isa_1.4.pdf
ptx_isa_2.3.pdf
Thrust_Quick_Start_Guide.pdf
Using_Inline_PTX_Assembly_In_CUDA.pdf
Find CUDA SDK:
ls $CUDA_INSTALL_PATH/../C
bin/ doc/ Makefile Samples.html tools/
common/ lib/ releaseNotesData/ src/
CUDA Profiler Documentation:
ls -1 $CUDA_INSTALL_PATH/computeprof/doc
about.txt
Changelog.txt
computeprof.html
computeprof.qch
computeprof.qhc
Compute_Visual_Profiler_User_Guide.pdf
help.png
How to Setup Personal CUDA SDK Environment
# Create a directory in your home directory
cd
mkdir cuda-sdk
cd cuda-sdk
# Make sure to invoke the setup script
setup cuda
# Create a skeleton of the read-only SDK
ln -s $CUDA_INSTALL_PATH/../shared .
mkdir C
cd C
ln -s $CUDA_INSTALL_PATH/../C/* .
ls
bin@ doc@ Makefile@ Samples.html@ tools@
common@ lib@ releaseNotesData@ src@
# Remove link to bin as makefile will try to write your program there
rm bin
rm: remove symbolic link `bin'? y
# Create a place for your experiments and copy a few projects
mkdir my-src
cp -a src/vectorAdd my-src
cp -a src/matrixMul my-src
# Build and run
cd my-src/vectorAdd/
make
../../bin/linux/release/vectorAdd
Vector addition
PASSED
Press ENTER to exit...
How to Setup the OpenCL Environment
Follow the same steps as for the CUDA Environment, except for the $CUDA_INSTALL_PATH/../C that should be replaced with $CUDA_INSTALL_PATH/../OpenCL. The rest is the same.
# Start in the cuda-sdk directory created above
cd ~/cuda-sdk
# Make sure to invoke the setup script
setup cuda
# Create the skeleton of the read-only SDK
mkdir OpenCL
cd OpenCL
ln -s $CUDA_INSTALL_PATH/../OpenCL/* .
bin@ common@ doc@ Makefile@ releaseNotesData@ Samples.html@ src@
# Remove link to bin as makefile will try to write your program there
rm bin
rm: remove symbolic link `bin'? y
# Create a place for your experiments and copy a few projects
mkdir my-src
cp -a src/oclVectorAdd src/oclMatVecMul my-src/
# Build and run (make sure you are in the directory of the example while running it)
cd my-src/oclVectorAdd/
make
../../bin/linux/release/oclVectorAdd
../../bin/linux/release/oclVectorAdd Starting...