Docker images meant to build versions with the simpletrack benchmark for nvidia GPU and intel runtimes.
Docker images meant to build versions with the simpletrack benchmark for nvidia GPU and intel and pocl runtimes. To be specify custom number of particles and turns use the environment variables:
```
~$ docker run --rm -e PARTICLES=100 -e TURNS=5 gitlab-registry.cern.ch/hep-workloads-gpu/...
```
* __Nvidia__: To run the benchmark through the container image entrypoint make sure to enable GPU passthrough.
* __Nvidia__: To run the benchmark through the container image entrypoint make sure to enable GPU passthrough.
...
@@ -13,3 +17,15 @@ Docker images meant to build versions with the simpletrack benchmark for nvidia
...
@@ -13,3 +17,15 @@ Docker images meant to build versions with the simpletrack benchmark for nvidia
```
```
~$ docker run --rm --device /dev/dri:/dev/dri gitlab-registry.cern.ch/hep-benchmarks/hep-workloads-gpu/lhc/simpletrack:intel-latest
~$ docker run --rm --device /dev/dri:/dev/dri gitlab-registry.cern.ch/hep-benchmarks/hep-workloads-gpu/lhc/simpletrack:intel-latest
```
```
* __POCL__: Portable Computing Language with built ICD extensions of OpenCL. Example execution using the cpu device:
```
~$ docker run --rm gitlab-registry.cern.ch/hep-benchmarks/hep-workloads-gpu/lhc/simpletrack:pocl-latest
```
To check for available devices:
```
~$ docker run --rm --entrypoint "/usr/bin/clinfo" gitlab-registry.cern.ch/hep-benchmarks/hep-workloads-gpu/lhc/simpletrack:pocl-latest