14 Nov RESEARCH PAPER: Xilinx Reconfigurable Acceleration Stack Targets Machine Learning, Data, Analytics, & Video Streaming
Hyperscale cloud, e-commerce, and social networking datacenters are increasingly facing the challenge of accelerating workloads with complex data types such as 4K video and natural languages, the processing of which often outstrips the capacity of traditional CPUs. This issue is especially acute in the so-called “Super Seven” datacenter companies (Alibaba, Amazon, Baidu, Facebook, Google, Microsoft, and Tencent), where these new applications can demand many thousands of accelerated application servers.
You can download the paper here.
Table of Contents
- Xilinx Strategy for Datacenter Accelerators
- Acceleration for Machine Learning
- Acceleration for Video Transcoding
- Acceleration for Data Analytics
- The Value of Hardware Reconfigurability
- Figure 1: Xilinx Acceleration Stack
- Figure 2: Autonomous Driving Sensors & Processors
- Figure 3: Single Server with Xilinx vs Rack of CPUs
- Figure 4: FPGA Acceleration for SQL Queries