Ascend 910C Cluster Hits 9,000 PFLOPS
pstAsiatech · x · 2026-07-12
Sources report that a training cluster consisting of 11,520 Huawei Ascend 910C chips has been deployed in Guangdong, boasting a nominal compute power of 9,000 PFLOPS.
The deployment was reportedly completed by China Telecom, with the goal of "breaking the constraints of overseas high-end compute supply." However, the original text notes that the announcement did not provide any training efficiency benchmarks comparing it directly to Nvidia.
More from Infra
- 12 KV Cache Reduction Techniques Every AI Engineer Should Understand, Explained — blaizedsouza · 2026-09-11
- The shadow GPU capacity market is formalizing, with Meta selling excess compute to outside buyers — DavidLinthicum · 2026-09-11
- Engram's random reads don't suit SSDs; CPU-memory over NVLink could serve all 72 GPUs — bookwormengr · 2026-09-11
- 80% of the DIY LLM inference hype posters have already quit — it's brutally hard systems work — abhijithneil · 2026-09-11
- Hugging Face's Ultra Scale Playbook: a free book on training LLMs on GPU clusters — mdancho84 · 2026-09-11
- Is inference latency becoming the biggest bottleneck for production AI agents? — Euphoric_Sea632 · 2026-09-11