华为发布AI新技术

Core Viewpoint - Huawei has launched Flex:ai AI container software, which utilizes computing power slicing technology to divide a single GPU/NPU computing power card into multiple virtual computing units, achieving a slicing granularity of up to 10% [1] Group 1 - Flex:ai allows a single card to simultaneously support multiple AI workloads [1] - The software can aggregate idle XPU computing power across nodes within a cluster to form a "shared computing pool" [1] Group 2 - Flex:ai will be open-sourced in the Magic Engine community following its release [2]