AITOPAI热报
IndustryOfficial

Moore Threads 2112卡GPU集群商用后利用率达92.9%

Moore Threads' 2,112-GPU Wuxi Cluster Reaches 92.9% Utilization After Going Commercial

官方一手Pandaily@[email protected] (Pandaily)
Blog/Media
Read the original
Why we picked it

Moore Threads的2112卡S5000集群商用数据出来了，跑DeepSeek V4 Flash能到H800的92%，利用率92.9%，看看国产GPU实际表现。

Moore Threads位于无锡惠山的MTT S5000集群共部署2112张GPU，为55家客户提供算力服务。商用后该集群利用率达到92.9%。在DeepSeek V4 Flash推理测试中，其性能达到H800方案的92%。这是国产GPU集群在商业化运营阶段交出的具体数据。

Original · Pandaily

Moore Threads' 2,112-GPU Wuxi Cluster Reaches 92.9% Utilization After Going Commercial

Moore Threads' MTT S5000 cluster in Wuxi runs 2,112 GPUs for 55 customers at 92.9% utilization, with DeepSeek V4 Flash inference tests reaching 92% of an H800 setup.

#Moore Threads#MTT S5000#DeepSeek#国产GPU#算力集群

Keep reading

Recent stories with the same tag or section
ModelsUS vs ChinaOfficialSelected15:49
Moore Threads MTT S5000 Matches Mainstream GPUs on Embodied RL Training

The non-obvious point is that curve correlation, not raw speed, is becoming the acceptance metric for swapping GPUs into RL pipelines — a compatibility benchmark Moore Threads can reuse across workloads. Teams building robotics or embodied RL stacks in China and its export markets gain a concrete alternative to CUDA-based setups. Readers evaluating domestic GPU swaps should check the r=0.976 correlation and 92% dual-arm success rate against their own tasks.

#Moore Threads#MTT S5000#RLinf#reinforcement learning
Event
官方一手pandaily@[email protected] (Pandaily)2 sourcesSource