XDF Silicon Valley and the Future of Heterogenous Acceleration

//XDF Silicon Valley and the Future of Heterogenous Acceleration

XDF Silicon Valley and the Future of Heterogenous Acceleration

Falcon Computing is extremely proud of its close partnership with Xilinx and the opportunity to be part of the Alveo Partner Ecosystem as the leading provider of an HPC platform and Life Sciences end-to-end solutions as highlighted during the release of Alveo for the data center. Being part of the Alveo Ecosystem also gave us an opportunity to showcase a demo of our Accelerated Genomics solution at the Alveo room. In there, we were able to connect and learn from analysts, partners, acceleration experts and a wave of software developers that attended XDF for the first time. From these conversations, we put our thoughts on the top two overarching themes in heterogeneous computing. These are also themes that continue shape our strategy to fulfill our mission of enabling heterogeneous hardware acceleration across the hybrid cloud.

Software developers will lead demand for data center acceleration
We met a lot of software developers that were attending XDF for the first time. To no surprise, the majority came wanting to learn more about adding ML and mapping custom applications on FPGAs. According to a demographic study by the Evans Data Corporation, there are currently about 23M developers in the world out of which about 6.5M are already adopting ML and AI in their projects. The adoption of Cognitive Era applications and the computational challenges that data scientists encounter in application such as image analysis and genomic sequencing are going to be the leading demand drivers. In the past, software developers and data scientists heavily depended on hardware architects to push their applications to the hardware layer. As their applications continue to evolve at an extremely fast pace, these group of users will need solutions that simplify their path to acceleration and solutions that reduce full-reliance on hardware acceleration experts to appropriately map their application to specialized hardware.

End-to-end solutions will drive early adoption, long term adoption will demand horizontal platforms
For software developers to get a taste of FPGA acceleration, they will adopt solutions that offer painless set-up and quickly show value on very specific use cases. As such, it is no surprise that the Alveo demo room was filled with partners that provide end-to-end solutions for tightly defined use cases such as video streaming, database acceleration and Falcon’s own genomics sequencing. As new users get first-hand experience of the benefits of FPGAs, there will be an overarching need to adopt acceleration across a wider range of processes that make complex applications to run. At this point, users will require platforms rather than tools. Using a multi-stack platform, users will simplify application mapping, resource runtime management and application optimization among others. The requirements for these platforms will become even more complex as there is a concurrent adoption of high-end CPUs and GPUs.

XDF San Jose was a great opportunity to learn from software developers about their needs while being able to show our HPC and Life Science solutions that support the Alveo accelerator cards across public and private clouds. We look forward to XDF Beijing and continuing to develop products that enable a simpler path to heterogenous computing.

By |2018-10-17T18:39:13+00:00October 12th, 2018|Blog|0 Comments