When configuring a task on VALDI, you can now filter available devices such as GPUs by their technical specifications.
You can filter on any combination of the following properties:
- Minimum number of CPU cores
- Minimum RAM
- GPU presence
- GPU chip
- Minimum VRAM
- Minimum CUDA version
- Maximum CUDA version
As you tweak your requirements, you are shown in real-time the exact number of devices currently available on the network which meet your needs.
In addition, VALDI has also made it possible to pass command-line parameters to your Docker container, when it is run on a network device.
These updates should make it even faster and easier to deploy HPC applications on VALDI, the world’s first distributed cloud computing platform.