We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
Hacker News·Jun 28·5 min read
TOP500 at ISC'26: We Have a New Number 1 – By George Cozma
TL;DR
A gap existed in the TOP500 list with no Chinese supercomputers for nine years. The LineShine Supercomputer, powered by the LX2 Armv9 CPU, was developed with 304 active cores and high-bandwidth memory.
✦ Why It Matters
Engineers can explore the implications of Armv9 architecture and high-bandwidth memory in supercomputing applications.
Key Takeaways
- Consider Armv9 architecture for future projects.
- Explore high-bandwidth memory solutions for performance.
- Analyze the implications of increased core counts.
- Stay updated on global supercomputing trends.
Full Summary
For nearly a decade, no Chinese supercomputers appeared on the TOP500 list, highlighting a competitive gap in high-performance computing. The LineShine Supercomputer, based in Shenzhen, China, features the LX2 CPU, which is Armv9-compliant and includes advanced features like Scalable Vector Extension 2 (SVE2) and Scalable Matrix Extension (SME).
Each LX2 chip has 304 active cores running at 1.55 GHz, delivering 60.3 teraflops (TFLOP/s) of double-precision floating-point performance while consuming 690 watts. Additionally, it incorporates 32 GB of high-bandwidth memory with a bandwidth of 4 terabytes per second (TB/s) and is supported by 256 GB of DDR5 memory.
This development signifies a major leap in China's supercomputing capabilities and could influence future designs in high-performance computing systems.