Acceleration experts show the future of HPC
A race of researchers and developers all over the world
High Performance Computing (HPC) utilizes supercomputers to perform large-scale computations such as computer simulation.
The history of supercomputers is long, and has evolved greatly since "CDC 6600" designed by Seymour Clay, called "The Father of Supercomputing" appeared in 1964. As well as improving the performance of devices by miniaturizing and architecture improvement, its computing power has dramatically improved through large-scale parallelization and software optimization.
Meanwhile, the increase in power consumption accompanying the increase in computing power and the expansion of the application fields of supercomputers, a different view point is required to improve performance compared to past performance improvements. Japanese companies and research institutes are at the forefront of driving evolution of supercomputers. In "Top 500" supercomputers competing for the calculation performance per power consumption, Japanese companies dominated the top rank. "Top 500" supercomputers competing for performance on big data processing, the supercomputer "K" gained first place for the sixth consecutive term. Japanese companies and research institutions are demonstrating their competitive edge.
For greater performance improvements, architecture changes such as many cores, large-scale clusters that utilize level parallel computing have been advancing, and benchmarking and software libraries to be used on the new systems are numerous. Fixstars as a professional of optimization / parallel processing is able to optimize the software.
We obtained close to theoretical performance by transplanting / accelerating "HPCG", a benchmark for large scale linear equation on "PEZY-SC".
The supercomputer "K" won first place in the world for the sixth consecutive year in the performance ranking "Top 500" on big data processing.
"K" won the first place in the world
We built an environment for distributed deep learning, and benchmarked the acceleration of "ChainerMN" in a high-power computing cloud environment where GPU nodes can be used.
According to customer's request (required specification), we port and optimize the existing software to the target hardware.
We investigate / propose hardware specifications from the viewpoint of software development (processing performance, ease of development, etc.).
We optimize your algorithm for the target processor including supporting the development of algorithms themselves.
We support software implementation and performance evaluation at the preliminary development stage, and performance estimation on target hardware and software development.
We investigate whether your software fully utilizes the hardware potential. We also report on the problems identified and provide proposals for future improvement.
In addition to this, we support customers in a wide range of stages from research to product development.
We offer various solutions useful in the HPC field.
We utilize optimization and parallel processing technology to accelerate applications and provide professional services to realize massive computational processing in a shorter time.
We combine optimization technology making full use of parallel processing technology and characteristics of various accelerators with a quantum computer showing overwhelming performance to provide a new high speed solution.