NEC's World’s Fastest Vector Supercomputer

NEC Corporation today announced the worldwide launch and availability of the SX series model “SX-8,” the world’s most powerful vector supercomputer with a peak processing performance of 65TFLOPS (TFLOPS: one trillion floating point operations per second).

Inheriting vector architecture, whose excellent sustained performance has been demonstrated by the Earth Simulator, the new product combines enhanced CPUs, and memory and I/O processing performance. A further-enhanced, single-chip, vector processor, realized by the SX-6, also contributes to the greatly improved price performance and space saving offered by the SX-8.

The SX-8, with its dedicated CPUs, large-scale memory, and high data transfer rate between memory and CPUs, delivers much higher sustained performance than scalar supercomputers with a number of general-purpose CPUs. This is particularly effective in fields that require large-scale and ultra high-speed computing of massive data, such as meteorological forecasting, environmental simulations and automotive crash analysis.

Features of the new product are described below.

1. World’s highest computing performance of 65TFLOPS The single-node model (includes up to 8 CPUs) achieves a peak vector performance of 128GFLOPS (GFLOPS: one billion floating point operations per second), while the multi-node model achieves the world’s fastest peak vector performance of 65TFLOPS when configured with 512 nodes. In addition, a 262TB/s (36.8TB/s in SX-6) high peak data transfer rate between CPU(s) and memory is realized, and it also boasts an enlarged memory capacity of up to 64TB (16TB in SX-6).

2. Further-enhanced single-chip, vector processor The vector processor (vector and scalar units) is integrated into a single chip by applying leading-edge CMOS technology with 90-nanometer (nanometer: 10-9 meter) copper interconnects and the most advanced LSI design technology. Pipelines of the vector unit, the central part of a vector processor, operate at a 2GHz clock frequency, which is double the speed of the SX-6, and realize a peak vector performance of 16GFLOPS per CPU. Moreover, hardware support of the vector square root operation achieves a sustained performance six times higher than that of the SX-6.

3. Excellent space-saving and power-saving The SX-8 achieves a reduction in space by 25% and power consumption by 50% as compared with conventional models. This was achieved by applying high-density packaging technology in which processor(s) and memory are implemented on a single module.

4. Software environment suitable for large-scale, multi-node system The basic software, “SUPER-UX,” maintains upward compatibility with the existing SX series, and achieves further scalability expansion due to enhanced I/O processing and MPI. Global File System (GFS), a high-speed, file sharing system among multiple nodes on a system or among different systems, has also been enhanced. By utilizing GFS, users can access the shared files with a high performance close to that of a local disc, and can also take advantage of the high performance functions of GFS from third party platforms such as SUN and HP as they do with NFS (Network File System).