Converged Compute
On the Arcfra Enterprise Cloud Platform, hosts in the cluster can be flexibly configured with GPUs of varying numbers and models. Arcfra supports GPU passthrough and vGPU, as well as technologies like MIG (Multi-Instance GPU) and MPS (Multi-Process Service), allowing for flexible GPU partitioning to enhance GPU utilization and AI application performance in virtualized and containerized environments.
Converged Workloads
With Arcfra Kubernetes Engine, you can use Arcfra Enterprise Cloud Platform to support both virtualized and containerized AI applications. Integrated deployment allows for efficient resource allocation, meeting diverse needs for performance, security, scalability, and agility across different AI workloads, thereby improving resource utilization and reducing overall costs.
Converged Storage
With its robust storage capabilities, Arcfra provides exceptional performance and stability for AI applications. Features like I/O localization, cold/hot data tiering, Boost mode, RDMA support, and volume pinning meet AI storage needs. Arcfra’s storage technology adapts to various hardware options, supports flexible scaling, and helps balance business requirements with cost considerations.