Lablup, an all-in-one enterprise development and service platform company for artificial intelligence development, announced its participation in Supercomputing 2024 (SC24), held in Atlanta, USA, from November 17 to 23.

SC24 is the largest supercomputing conference, hosted annually since 1988 by the Association for Computing Machinery (ACM) and IEEE Computer Society, where the latest research and technology in high-performance computing (HPC), networking, storage, and analytics are shared.

Participating in the exhibition for the second consecutive year, Lablup will unveil version 24.12 of its AI infrastructure platform Backend.AI at this event. Additionally, they will showcase demos of PALI, an inference solution based on Backend.AI 24.12, optimized for various AI semiconductors. Since NVIDIA's Grace CPU was unveiled in 2021, Lablup promptly released a new Backend.AI version compatible with the Grace architecture, completed verification for the GH200 Grace Hopper Superchip, and prepared for NVIDIA's Blackwell platform to be released in 2024. Furthermore, through a six-month collaboration partnership with Intel, they optimized Backend.AI to maximize the performance of Intel Gaudi 2 and Intel Gaudi 3 AI accelerator platforms.

At SC24, Lablup will demonstrate AI model operations and development on Intel Gaudi 2 and Intel Gaudi 3 AI accelerator platforms and NVIDIA Grace Hopper Superchips using PALI. Simultaneously, through various booth sessions with partners, they will introduce the foundational technologies for operating, optimizing, and managing large-scale inference services with PALI.

CEO Jeongkyu Shin stated, "Lablup is a company that accelerates computing. The AI field, which demands massive computing power, is becoming closely integrated with supercomputing," adding, "At SC24, where the frontiers of computing including AI, accelerated computing, storage, networks, and quantum computers unfold, we will share Lablup's advanced technology with the world."