Acceleration technology solutions company specializing in big-data applications is developing a way to speed up computational genomics for the bioscience field. The computational cost of genomic sequencing is vast, with many constraints in its current systems: on-premise data centers are limited by fixed space, energy budget, performance bottleneck; in the cloud, long upload time, security, high cost for compute and storage are major issues.
The company leveraged Inspur’s F10A FPGA accelerator with its compiler tool to run acceleration algorithms on standard, best-practices GATK pipeline. The solution demonstrated outstanding results: aggressive acceleration, rapid turn-around times, and a scalable approach. It had the ability to quickly support new code bases/releases or migrate across hardware platforms. It enabled the use of FPGA accelerators without expertise in FPGA design or OpenCL languages. The solution was able to eventually scale to hybrid, heterogeneous clouds (CPU, FPGA and other), and showed continued improvement to performance and support of emerging hardware.
Inspur’s F10A, as demonstrated, provided the combined solution a significant boost to the performance, speed and efficiency. 10 x improvements to the GATK pipeline were observed, with a 3.4 x speed-up of computational processing. This solution provides a considerable speed boost to compute and time intensive life sciences applications.