HPL calendar
Supercomputers fill hundreds of racks, run millions of cores in unison, and consume electricity measured in megawatts. The June 2026 list of TOP500, which ranks these systems using the HPL (Linpack) test, brought a change at the top. The top three are as follows:
- LineShine: The system located at the Shenzhen National Supercomputing Center in China took the top spot with 2.198 exaflops of HPL performance.
US Lawrence figures
It has more than 13.7 million cores; its theoretical peak performance is 2.736 exaflops. By using LingKun processors and LingQi infrastructure, it brought a different architecture to a summit dominated by AMD, Intel, and NVIDIA.
- El Capitan: The system at Lawrence Livermore National Laboratory in the US, the leader of the previous list, is second with 1.809 exaflops.
AMD EPYC measurements
Built on the HPE Cray EX255a; it includes fourth-generation AMD EPYC processors and Instinct MI300A accelerators that combine the CPU and GPU in the same package. The core count exceeds 11 million, and the theoretical peak performance is approximately 2.8 exaflops.
Oak Ridge remarks
It was developed for nuclear stockpile management simulations; its power consumption is approximately 29.7 megawatts.
- Frontier: The system at Oak Ridge National Laboratory became a milestone of the exascale era by surpassing the exaflop threshold in 2022; it is third with 1.353 exaflops.
The phrase “most powerful” depends on the test: the ranking is based on HPL performance and can change with different benchmarks.