Description
The Purdue Anvil AI system has 21 nodes each with four NVIDIA 80GB H100 SXM GPUs to support machine learning and artificial intelligence applications.
RP Description
Anvil AI is a compute cluster consisting of 21 nodes, each with four NVIDIA H100 GPUs and Intel Xeon Platinum 8468 processors with 96 cores and 1 TB of memory. It is particularly well suited for large-model work that needs high GPU memory and fast interconnect, and is often used for deep learning training, fine-tuning and large-scale inference. It includes a great deal of AI and machine learning software, with CUDA-enabled framework builds and container support.
Jobs Information
You can run jobs at different sizes and durations on Anvil AI. The following lists the different queues that you can submit to, describing how many nodes you get, how long you can run, the type of resources you get, and the average wait time.
Jobs are submitted through the Slurm scheduler. Run batch jobs with sbatch, start interactive work with srun or salloc, or launch an application through Open OnDemand.
Submit to the ai partition with -p ai, and give a GPU allocation with -A - AI work is not charged against a CPU allocation. No more than 12 GPUs may be in use per user and 32 per allocation at one time.
Per-queue wall-clock limits are in the queue table; if you do not set one, the partition default applies, which scontrol show partition reports.
For more information about running jobs on Anvil, see [Anvil Running Jobs].