Skip to content
forked from nchong/cudahook

Intercepting CUDA runtime calls with LD_PRELOAD

Notifications You must be signed in to change notification settings

kllmagn/cudahook

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

cudahook

A library for intercepting CUDA runtime calls with LD_PRELOAD and dlsym.

Compiling

We use hook mechanisms that are specific to GNU/Linux.

$ make

should work, although you need to edit the CUDAPATH variable in the Makefile if CUDA is not installed to /usr/local/cuda

Using

$ LD_PRELOAD=/path/to/cudahook/libcudahook.so ./your-program

will run the CUDA program and intercept the following runtime calls:

  • cudaConfigureCall
  • cudaLaunch
  • cudaRegisterFunction
  • cudaSetupArgument

At the moment, we use this information to print out information about invoked kernels.

About

Intercepting CUDA runtime calls with LD_PRELOAD

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 92.4%
  • Makefile 7.6%