Search
Moore Threads MTT S5000 parameters revealed for the first time: 80GB VRAM, single-card computing power of 1 petaFLOPS
2 day ago /
Read about 0 minute
Author：小编
Following the release of Zhipu's new-generation large language model GLM-5, Moore Threads swiftly announced that its flagship AI training and inference all-in-one full-function GPU, the MTT S5000, has completed Day-0 full-process adaptation and verification, with immediate support available. The MTT S5000 is a full-function GPU smart computing card specifically designed for large-scale model training, inference, and high-performance computing. It utilizes the fourth-generation MUSA architecture 'Pinghu' and natively supports mainstream frameworks such as PyTorch, Megatron-LM, vLLM, and SGLang.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
