Imperial College London > Talks@ee.imperial > CAS Talks > A Multi-Codec Framework to Enhance Data Channels in FPGA Streaming Systems

A Multi-Codec Framework to Enhance Data Channels in FPGA Streaming Systems

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Grigorios Mingas.

We propose a framework to apply general-purpose data transforms to a data channel between a stateless FPGA kernel and a CPU , such as increasing effective bandwidth by using spare CPU and FPGA resources on a heterogeneous CPU -FPGA system. Compute-intensive transforms (codecs) such as compression are applied in real-time while maintaining the simplicity of a high-level abstraction for a PCI -express channel. Real-time encoding/decoding is accelerated by splitting the data stream into segments and having multiple codec threads processing different segments. Multi-threading on the CPU allows the overlap of encoding, data transmission and decoding thereby functioning as a software pipeline. We demonstrate this using 4 compression codecs across 3 data types, showing speed-ups ranging from 1.26x to 1.60x in a Maxeler-based system. For example, we achieve a 1.51x speed-up for delta compression, increasing the effective total input/output bandwidth of the data channel from 2509MB/s to 3790MB/s.

This talk is part of the CAS Talks series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

Changes to Talks@imperial | Privacy and Publicity