Traccc as-a-Service

Traccc as-a-Service#

Main objective: run traccc as-a-Service. Getting this working includes creating three main components:

  1. a shared library of traccc and writing a standalone version with the essential pieces of the code included

  2. a custom backend using the standalone version above to launch the Triton server

  3. a client to send data to the server

A minimal description of how to build a working version is detailed below. For more in-depth descriptions of each step in the process, consult the git repo of this project.

Running out of the box#

Get the code#

Simply clone the repository with

git clone --recurse-submodules git@github.com:milescb/traccc-aaS.git

Docker#

A docker built for the triton server can be found at docker.io/milescb/triton-server:latest. To run this do

shifter --module=gpu --image=milescb/tritonserver:latest

or use your favorite docker application and mount the appropriate directories.

Note

An image has been built with the custom backend pre-installed at docker.io/milescb/traccc-aas:v1.1. To run this, open the image, then run the server with

tritonserver --model-repository=$MODEL_REPO

To see how this was built, consult the Dockerfile in the traccc-aaS repository.

Shared Library on NERSC#

To run out of the box on NERSC, an installation of traccc and the backend can be found at /global/cfs/projectdirs/m3443/data/traccc-aaS/software/prod/ver_09152024/install. To set up the environment, run the docker then set the following environment variables

export DATADIR=/global/cfs/projectdirs/m3443/data/traccc-aaS/data
export INSTALLDIR=/global/cfs/projectdirs/m3443/data/traccc-aaS/software/prod/ver_09152024/install
export PATH=$INSTALLDIR/bin:$PATH
export LD_LIBRARY_PATH=$INSTALLDIR/lib:$LD_LIBRARY_PATH

Then the server can be launched with

tritonserver --model-repository=$INSTALLDIR/models

Note

If you do not have access to NERSC computing resources, then you will need to build traccc from source. Follow the instructions in the traccc repo then update INSTALLDIR and DATADIR appropriately.

To avoid having to build from source, deploy the server with the docker image docker.io/milescb/tritonserver:latest as directed above.

Building the backend#

First, enter the docker and set environment variables as documented above. Then run

cd backend/traccc-gpu && mkdir build install && cd build
cmake -B . -S ../ \
    -DCMAKE_INSTALL_PREFIX=../install/

cmake --build . --target install -- -j20

Then, the server can be launched as above:

tritonserver --model-repository=../../models

Run a simple client#

Once the server is launched, run the model via:

cd client && python TracccTritonClient.py

This launches a simple gRPC client which makes one inference request to the deployed server. To test the scalability of the model and obtain performance metrics, the perf_analyzer tool can be used. For more information on how to obtain metrics and make plots, consult the traccc-aaS-performance repository.