INFINITIX AI-Stack PaaS Product Offers Optimal NVIDIA GPU Computing Efficiency

The Impact and Challenges of AI on industries

The impact and challenges of artificial intelligence (AI) continue to increase businesses' interest in AI investment. However, the talent shortage in industries has remained a difficult issue to resolve. The scarcity of talent leads to higher labor costs, which is detrimental to businesses' development. Generally, businesses gain a competitive advantage through either cost reduction or differentiation, and for AI technology, its core costs are talent and computing power. Unfortunately, the investment costs for these two resources are currently relatively high. GPU computing power is still expensive on the market, which raises the cost of learning AI and reduces the number of people who can afford to enter this field.

INFINITIX's Goal

The goal of INFINITIX is to make AI technology more accessible to more people. The company is dedicated to reducing the cost of the "AI adoption" process, allowing more customers to have the opportunity and ability to enter the AI domain. To meet the rapidly developing AI, high-performance computing, and data center markets, INFINITIX's "AI Computing Platform AI-Stack" helps organizations in various sectors, including enterprises, medical institutions, financial industries, general manufacturing, semiconductors, educational institutions, and government R&D units, build "self-built" or "local + cloud" AI cloud platforms. This enables researchers, AI engineers, professors, and students to focus on learning and professional development.

Features of the AI-Stack Platform

The AI-Stack platform provides a web-based interface and a unified user portal, allowing users to self-service create deep learning containers and perform daily operations through a graphical interface. It automatically allocates and deploys resources for different users, while also incorporating NVIDIA-optimized and widely used AI frameworks such as TensorFlow, Caffe, PyTorch, MXNet, RAPIDS, Matlab, Chainer, and TensorRT. Users can define custom AI images as needed, benefiting from an excellent framework expansion design that greatly reduces the time spent downloading deep learning and AI frameworks. Additionally, through system process control and resource application mechanisms, along with batch task creation and scheduling operations, both individuals and development teams can use GPU resources more efficiently.

Technical Support

In terms of technical support, AI-Stack is paired with NVIDIA Driver, NVIDIA CUDA (version 10.0 and above), NCCL, cuDNN, and other technologies to provide AI machine learning training services. It uses NVIDIA Docker and Kubernetes for GPU container management. Customers only need to prepare a Linux Ubuntu 18.04 environment to complete the installation and can easily set up a container within one minute, saving significant time and cost on environment setup. Furthermore, AI-Stack provides users with the Jupyter Notebook programming tool for development, accelerating AI development efficiency.

Hardware Support

In terms of hardware support, AI-Stack can manage NVIDIA GPUs from the A100, Tesla, Quadro, and GeForce series, and interface with external storage devices. Regardless of the GPU server brand or number of devices, AI-Stack can easily manage them through its software license. Whether it's one person using multiple cards, multiple people sharing a machine, or multiple people using multiple machines—whether within a single department, research institute, R&D center, or across departments, institutes, and public-facing applications—AI-Stack offers efficient collaboration and seamless management of AI development environments.

Future Developments

In the future, AI-Stack will also integrate NVIDIA Clara™. This will make it easier to use Clara for medical imaging, genomics research, patient monitoring, and drug development on the AI-Stack platform.

As a member of the NVIDIA Inception Program, INFINITIX is always closely monitoring NVIDIA product developments. AI-Stack aligns with the management node development direction of the NVIDIA DGX POD reference architecture, ensuring the latest NVIDIA GPU technology support. Therefore, AI-Stack is the best management platform for customers choosing to use GPUs for AI.