Running the examples🔗
vAccel comes with a set of examples for common acceleration operations like image classification, matrix multiplication etc. Besides providing common acceleration functionality, the examples are meant as a demonstration of how to use the different operations of the vAccel API.
The bare minimum you have to configure in order to run the examples - or any vAccel application - is the backend plugin that will be used. You can do this by setting the VACCEL_PLUGINS environment variable:
This will select the NoOp dummy plugin as a backend.
You are now ready to run any example.
For example, assuming vAccel is installed in /usr/local, you can run an image classification operation with:
$ classify /usr/local/share/vaccel/images/example.jpg 1
Initialized session with id: 1
classification tags: This is a dummy classification tag!
classification imagename: This is a dummy imgname!
By enabling debug level logging, you can get more details on what is happening in the background:
The output will now be:
$ classify /usr/local/share/vaccel/images/example.jpg 1
2025.04.03-15:44:04.61 - <debug> Initializing vAccel
2025.04.03-15:44:04.61 - <info> vAccel 0.6.1-194-19056528
2025.04.03-15:44:04.61 - <debug> Config:
2025.04.03-15:44:04.61 - <debug> plugins = libvaccel-noop.so
2025.04.03-15:44:04.61 - <debug> log_level = debug
2025.04.03-15:44:04.61 - <debug> log_file = (null)
2025.04.03-15:44:04.61 - <debug> profiling_enabled = false
2025.04.03-15:44:04.61 - <debug> version_ignore = false
2025.04.03-15:44:04.61 - <debug> Created top-level rundir: /run/user/1002/vaccel/VC0Gxz
2025.04.03-15:44:04.61 - <info> Registered plugin noop 0.6.1-194-19056528
2025.04.03-15:44:04.61 - <debug> Registered op noop from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op blas_sgemm from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op image_classify from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op image_detect from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op image_segment from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op image_pose from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op image_depth from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op exec from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op tf_session_load from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op tf_session_run from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op tf_session_delete from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op minmax from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op fpga_arraycopy from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op fpga_vectoradd from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op fpga_parallel from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op fpga_mmult from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op exec_with_resource from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op torch_jitload_forward from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op torch_sgemm from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op opencv from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op tflite_session_load from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op tflite_session_run from plugin noop
2025.04.03-15:44:04.61 - <debug> Registered op tflite_session_delete from plugin noop
2025.04.03-15:44:04.61 - <debug> Loaded plugin noop from libvaccel-noop.so
2025.04.03-15:44:04.62 - <debug> New rundir for session 1: /run/user/1002/vaccel/VC0Gxz/session.1
2025.04.03-15:44:04.62 - <debug> Initialized session 1
Initialized session with id: 1
2025.04.03-15:44:04.62 - <debug> session:1 Looking for plugin implementing VACCEL_OP_IMAGE_CLASSIFY
2025.04.03-15:44:04.62 - <debug> Returning func from hint plugin noop
2025.04.03-15:44:04.62 - <debug> Found implementation in noop plugin
2025.04.03-15:44:04.62 - <debug> [noop] Calling Image classification for session 1
2025.04.03-15:44:04.62 - <debug> [noop] Dumping arguments for Image classification:
2025.04.03-15:44:04.62 - <debug> [noop] model: (null)
2025.04.03-15:44:04.62 - <debug> [noop] len_img: 79281
2025.04.03-15:44:04.62 - <debug> [noop] len_out_text: 512
2025.04.03-15:44:04.62 - <debug> [noop] len_out_imgname: 512
2025.04.03-15:44:04.62 - <debug> [noop] will return a dummy result
2025.04.03-15:44:04.62 - <debug> [noop] will return a dummy result
classification tags: This is a dummy classification tag!
classification imagename: This is a dummy imgname!
2025.04.03-15:44:04.62 - <debug> Released session 1
2025.04.03-15:44:04.62 - <debug> Cleaning up vAccel
2025.04.03-15:44:04.62 - <debug> Cleaning up sessions
2025.04.03-15:44:04.62 - <debug> Cleaning up resources
2025.04.03-15:44:04.62 - <debug> Cleaning up plugins
2025.04.03-15:44:04.62 - <debug> Unregistered plugin noop
You can find sample invocation commands for all the examples in the run-examples script. You can also clone the repo and run the script yourself with:
git clone https://github.com/nubificus/vaccel
cd vaccel
# Replace '/usr/local' with the vAccel installation prefix
./scripts/run-examples.sh /usr/local
If you have already built vAccel from source, simply run:
in your vaccel directory.
You can find more information on the environment variables you can use to configure vAccel here.