Along with generative AI purposes, new a number of mannequin AI approaches are engineered to boost prediction and accuracy in lots of business use instances like superior fraud detection, picture processing and retail automation, in accordance with IBM.
LinuxONE Emperor 5 additionally comes with superior safety features particularly designed for the AI risk panorama. The platform extends IBM’s end-to-end cybersecurity method by way of confidential computing, a way the place information is encrypted even when in reminiscence. The platform additionally comes with NIST-standardized post-quantum cryptographic algorithms.
The mainframe additionally comes with a {hardware} safety module and assist for confidential containers, and built-in with Purple Hat OpenShift Container Platform to make sure delicate AI fashions and information stay protected.
With IBM Safe Execution, LinuxONE Emperor 5 makes use of safety features rooted in {hardware} to isolate workloads, offering safety towards insider threats and unauthorized entry from throughout the workload provisioning infrastructure.
On the software program aspect, the AI Toolkit for IBM LinuxONE has been up to date and optimized for Telum II, developer productiveness and AI deployment. AI Toolkit for LinuxONE contains the newest model of the IBM Z Deep Studying Compiler, which permits using the on-chip AI Accelerator on Telum II processor.
The toolkit has enhancements and optimizations for common open-source frameworks, together with PyTorch, TensorFlow, TensorFlow Serving, NVIDIA Triton Inference Server, and SnapML. It additionally helps a wide range of databases, together with IBM Db2, MongoDB, EDB, and Oracle.
