HPE is expanding its machine learning capabilities portfolio considerably. The tech giant introduced the HPE Machine Learning Development System and Swarm Learning framework, two appliances for developing machine learning models.
The introductions were made possible by integrating Determined AI, an AI and ML specialist acquired last year. The first release to integrate the technology is the HPE Machine Learning Development System. The physical appliance allows users to train and build ML models on a large scale.
The HPE Machine Learning Development System combines HPE’s existing Machine Learning Development environment with processing power, accelerators and onboard networking. According to HPE, the system allows companies to significantly accelerate the development of ML models.
The platform answers needs on the physical side of the ML development process, such as installing large amounts of parallel processors and using specialised processing power. Everything is provided in a compact and pre-installed environment, allowing companies to start developing their ML models immediately upon connection. In addition, the system simplifies the management of required infrastructure.
Inside the system, you’ll find an HPE Apollo 6500 Gen10 system with eight Nvidia A100 80 GB GPUs. In addition, the platform features HPE ProLiant DL325 servers and a 1 Gbps Ethernet Aruba CX 6300 switch. Networking and storage functionality is made possible by Nvidia Quantum InfiniBand and HPE Performance Cluster Management monitoring and management software.
HPE Swarming Learning framework
In addition to the Machine Learning Development System, HPE is introducing the Swarm Learning framework. This is a dedicated framework for machine learning functionality in distributed and edge systems. The framework provides specialized containers that can be integrated into I-models using the HPE Swarm API. This technology allows an organization to learn from existing AI models without actually sharing data.
Collaboration with Qualcomm
HPE also announced an extension of its partnership with chip manufacturer Qualcomm. The partners primarily focus on producing new technology for large-scale AI inferencing. For this purpose, HPE is introducing a new version of its Edgeline EL8000 Converged Edge systems in combination with the Qualcomm Cloud AI 100 accelerator. This should enable AI inferencing in edge locations. The application will become available in August 2022.