NVIDIA cuOpt™ is a GPU-accelerated solver that uses heuristics and metaheuristics to solve complex vehicle routing problem variants with a wide range of constraints. Using NVIDIA’s cutting-edge GPUs and parallel computing capabilities, cuOpt delivers significant speedups and efficiency improvements over traditional CPU-based optimization methods. It’s particularly useful for applications requiring rapid solutions to large-scale, complex problems, such as route optimization, supply chain logistics, and resource allocation.
Oracle Cloud Infrastructure (OCI) offers a range of NVIDIA GPU instances in both virtual machine (VM) and bare metal, including the NVIDIA L40S GPUs and NVIDIA A10 , A100 , and H100 Tensor Core GPUs. These shapes are optimized for high-performance computing (HPC) and AI. NVIDIA H100 and NVIDIA A100 GPUs are ideal for running compute-intensive applications like NVIDIA cuOpt.

NVIDIA cuOpt on Oracle Cloud Marketplace

OCI now offers NVIDIA cuOpt as a Marketplace image, simplifying the deployment process. You can quickly set up any NVIDIA GPU instance within your OCI tenancy to harness the power of cuOpt for your optimization needs. This integration allows you to utilize OCI’s robust cloud environment, combined with NVIDIA’s GPU acceleration, to tackle demanding optimization tasks efficiently.

You can deploy the Marketplace image on any NVIDIA GPU instance with Oracle Linux.

 

Deploy NVIDIA cuOpt using Helm on OKE

In addition to deploying cuOpt on OCI instances powered by NVIDIA, you can also deploy it on Oracle Kubernetes Engine (OKE) using Helm. This method allows for more scalable and containerized deployments, making it easier to manage and orchestrate multiple instances of cuOpt in a Kubernetes environment and integrate it with your applications that already run in OKE.

Architecture diagram for deploying cuOpt to OKE on OCI

Figure 1: Architecture diagram for deploying cuOpt to OKE on OCI

Conclusion

Whether you choose to deploy NVIDIA cuOpt on an OCI VM instance accelerated by NVIDIA GPUs such as the NVIDIA A100, NVIDIA H100 GPU s or use   Helm on OKE, you can take advantage of the powerful GPU-accelerated optimization capabilities it offers. By employing OCI, you gain access to a robust, scalable, and secure cloud environment that complements the efficiency of NVIDIA cuOpt. These deployment options provide the flexibility to match your specific infrastructure and scaling needs, ensuring that you can tackle complex optimization problems with speed and reliability. Explore the power of NVIDIA cuOpt on Oracle Cloud Infrastructure and streamline your optimization processes today.

For more information, see the following resources: