In addition to generative AI applications, new multiple model AI approaches are engineered to enhance prediction and accuracy in many industry use cases like advanced fraud detection, image processing and retail automation, according to IBM.
LinuxONE Emperor 5 also comes with advanced security features specifically designed for the AI threat landscape. The platform extends IBM’s end-to-end cybersecurity approach through confidential computing, a technique where data is encrypted even when in memory. The platform also comes with NIST-standardized post-quantum cryptographic algorithms.
The mainframe also comes with a hardware security module and support for confidential containers, and integrated with Red Hat OpenShift Container Platform to ensure sensitive AI models and data remain protected.
With IBM Secure Execution, LinuxONE Emperor 5 utilizes security features rooted in hardware to isolate workloads, providing protection against insider threats and unauthorized access from within the workload provisioning infrastructure.
On the software side, the AI Toolkit for IBM LinuxONE has been updated and optimized for Telum II, developer productivity and AI deployment. AI Toolkit for LinuxONE includes the latest version of the IBM Z Deep Learning Compiler, which allows the use of the on-chip AI Accelerator on Telum II processor.
The toolkit has enhancements and optimizations for popular open-source frameworks, including PyTorch, TensorFlow, TensorFlow Serving, NVIDIA Triton Inference Server, and SnapML. It also supports a variety of databases, including IBM Db2, MongoDB, EDB, and Oracle.