That architecture showed that the synergy between mainland-designed software and chips would realise significant results, according to the report.
FP8, known as floating-point 8, is a format that reduces precision to speed up AI training and inference by using less memory and bandwidth. UE8M0, another 8-bit format, is said to increase training efficiency, which reduces hardware requirements by cutting memory use by up to 75 per cent.
The V3.1’s data format represents “a key pillar for building an independent and controllable AI ecosystem” for China, as the country establishes a “healthy ecosystem for the coordinated development of hardware, software and applications in the field of AI computing”, according to a report on Sunday by Haitong Securities.
