Empowering AI Everywhere: Democratizing PyTorch with Intel...

Sponsored Session: Empowering AI Everywhere: Democratizing PyTorch with Intel AIPCs & More - Fan Zhao & Eikan Wang, Intel PyTorch has become a cornerstone for AI development, powering everything from inference pipelines to autonomous agents. Traditionally, these workloads required high-end systems equipped with discrete GPUs, often limiting accessibility and flexibility. Scaling inference at this pace demands more than powerful chips. It requires an open inference-first approach, where sustained throughput, and system-level efficiency matter far more than peak compute. This presentation explores how advancements in AIPCs are democratizing PyTorch, enabling seamless development and deployment across diverse platforms. By supporting both Windows and Linux environments, and integrating capabilities for both integrated and discrete GPUs, AIPCs are breaking down barriers to entry, making AI development more accessible than ever. Over the past year, significant enhancements have been made to support PyTorch on AIPCs, including features like torch.compile, comprehensive ATen operations, low-precision computation, and torch.distributed. These improvements have notably enhanced the user experience, particularly on Windows, allowing developers to leverage PyTorch’s full potential without the need for specialized hardware. Join us as we demonstrate how AIPCs empower developers to natively build and run sophisticated AI applications directly on everyday devices, such as laptops and desktops. This shift not only simplifies the setup process but also opens new avenues for innovation, enabling AI to thrive in environments previously considered out of reach.