Using GStreamer for Seamless Off-loading Audio Processing to a DSP - Ruud Derwig, Synopsys
This session explains how off-loading audio processing from an application CPU to an audio DSP is made easy using GStreamer. Despite the compute power of multicore application CPUs, the SoC design trend is heterogeneous with specialized subsystems and cores. For power efficiency and hardware cost such SoCs are optimal, for developers they are a pain. In the homogenous case SMP-Linux hides complexities, in the heterogeneous case developers must deal with different tools, shared memory (coherency), multiple OSes, optimization of DSP code, etc. Solutions as remoteproc are a first step in simplifying the use of the different cores on a SoC. We take this basic management and control a step further by leveraging the domain specifics of audio processing. The complexities of off-loading are hidden inside GStreamer elements, while retaining the flexible, plug-and-play graph creation of GStreamer.
Software & System Architect, Synopsys
Ruud Derwig has 15+ years of experience with software architectures for consumer electronics products. Ruud has been working at Philips Corporate Research, NXP Semiconductors, and is currently Software Architect in Synopsys. He represented NXP and Philips in the architecture group of the CE Linux Forum (CELF), chaired the Audio Video Graphics working group of CELF, and does some of the organization of ELC-Europe. Recently, he's been developing the software for the SoundWave audio subsystem.
Attendance numbers do not account for private attendees. Get there early!
Remove this from your schedule?
This session is full and you may not be able to get back in.