oneAPI for FPGA

Whole genome sequencing is key in the field of genomics, but since it takes a long time to execute calculations, parallelization and offloading using GPU are applied to accelerate it.

The most widely used Genome Analysis Toolkit (GATK) is written in Java, so it is impossible to directly apply oneAPI to port it to FPGA, but the previous step, aligning/mapping, is written in C/C++, BWA-MEM and DRAGMAP using SYCL. It can be ported to FPGA.

In particular, it is economical and convenient to use in the form of a Programmable Acceleration Card (PAC) installed in the server rather than using the FPGA alone because it has to be used with GATK, and CPU work is also essential.

SYCL was conveniently ported using the Acceleration Stack for use in PAC, and execution speed was improved by applying parallelization and pipelining.

×


Watch the oneAPI DevSummit hosted by UXL:

Watch Now