HOS - Huygens On Speed

GPU-based ultrasound field simulator with interactive user interface.

Homepage: http://heim.ifi.uio.no/jpaasen/huygens/

Released under the GPL license.

Now there is also a WebGL-implmentation of this simulator. You can find it here.

What do you need to make it run on speed?

• CUDA capable GPU
• CUDA Toolkit 3.2 or greater

Build with cmake and make

1. Create "build" folder
2. Enter build folder and type: > cmake ..
3. Then make the project with: > make
4. Run program with: > ./HuygensOnSpeedGlutApp/HuygensOnSpeedApp

Or short: > cmake .. ; make ; ./HuygensOnSpeedGlutApp/HuygensOnSpeedApp

Build with cmake and Visual Studio

1. Generate Visual Studio 20XY solution file using cmake in terminal or GUI.
2. Open solution file and build (F5).

Build options

Build CPU version

Set CMAKE arguments CALC_WITH and DISP_WITH to either "CUDA" or "CPU". Valid combinations are:

• CALC_WITH="CUDA" and DISP_WITH="CUDA"
• CALC_WITH="CPU" and DISP_WITH="CUDA"
• CALC_WITH="CPU" and DISP_WITH="CPU" Command line example:
  > cmake .. -DCALC_WITH="CPU" -DDISP_WITH="CPU"

Info

Info for Linux users

Remember to have libGL, libGlew, libGLU and libglut in lib/linker path. To grab glew use: sudo apt-get install libglew1.6-dev To grab glu use: sudo apt-get install libglu1-mesa-dev The other two is usually install together with CUDA. To se if you have a lib installed use: apt-file search

Info for Windows users

Generate solution file using cmake. Project dependencies (libs) are found in the CUDA Toolkit and SDK if CALC_WITH="CUDA" or DISP_WITH="CPU" and should be located automatically by cmake. If both CALC_WITH and DISP_WITH are "CPU", the project uses files found in the local include and lib folder.

Info for Mac users

Same build instructions as for linux: mkdir build; cd build; cmake ..; make

Glew have to be install to make it run. > brew install glew If you don't have brew you can get it here Or just paste the following line in the terminal: ruby -e "$(curl -fsSL https://raw.github.com/Homebrew/homebrew/go/install)"

Known issues on mac

Have to force close the application (ctrl + z in the terminal and right-klick + force quit on the ui icon)

Info about the projects

HuygensOnSpeed

This includes the simulator code and data structures for point sources, observation points, Paint tools and etc. It includes both a GPU and CPU implementation. For display, a CUDA-OpenGL binding is used. Now there is also a display class for CPU and OpenGL only.

Build instructions (windows)

The following build instructions is based on this code beeing build as a static library HuygensOnSpeed(.lib/.a).

• Additional include directories: "$(CUDA_PATH)\include";"$(NVSDKCOMPUTE_ROOT)\C\common\inc"
• Additional lib dependencies: cudart(.lib) cublas(.lib)
• Additional lib dirs: "$(NVSDKCOMPUTE_ROOT)/C/common/lib/$(PlatformName)"; "$(CUDA_PATH)/lib/$(PlatformName)"

HuygensOnSpeedGlutApp

This is the Paint-UI used to draw lines of source points.

Build instructions (windows)

• Additional include directories: "$(CUDA_PATH)\include";"$(NVSDKCOMPUTE_ROOT)\C\common\inc"
• Additional lib dependencies: HuygensOnSpeed(.lib) glew(64.lib)
• Additional lib dirs: "$(NVSDKCOMPUTE_ROOT)/C/common/lib/$(PlatformName)";"$(SolutionDir)$(PlatformName)/$(ConfigurationName)";"$(CUDA_PATH)/lib/$(PlatformName)"

After compilation (on windows, maybe on Linux too?), the glut app will require the following dynamic libraries from the cuda sdk in order to run:

• cudart(64.dll)
• cublas(64.dll)
• freeglut(.dll)
• glew(64.dll)

MatlabInterface

Functionality for calling HuygensOnSpeed from Matlab.

Build instructions

See own Readme file in sub folder. Building mex-interface has only been tested on windows 7 64bit.

Author

Jon Petter Åsen - jon.p.asen@ntnu.no