Tianhe1A

Tianhe-1A is a supercomputer developed by the National University of Defense Technology (NUDT) of China. It was installed at the National Supercomputing Center in Tianjin and completed in 2010. In mid-2010 it was ranked No. 1 on the TOP500 list, becoming the fastest known supercomputer by LINPACK benchmarks at that time.

The system employs a heterogeneous architecture that combines Intel Xeon CPUs with NVIDIA Tesla GPU accelerators, connected by a high-speed interconnect known as THExpress-2. This CPU-GPU hybrid design aimed to maximize computational density by offloading parallel workloads to the GPUs while using the CPUs for orchestration and more general tasks. The compute nodes are organized in a scalable layout, with a topology and software stack optimized for high-throughput scientific computing and large-scale simulations.

Performance and impact: On the LINPACK benchmark, Tianhe-1A demonstrated sustained performance of about 2.5 to 2.6 petaflops, making it the fastest supercomputer in the world for a period after its introduction. The system highlighted China’s advances in high-performance computing and contributed to the broader adoption of GPU acceleration in top-tier HPC systems. It was subsequently surpassed by newer systems on the TOP500 list, but remains a notable milestone in the history of supercomputing for its scale and architectural approach.