Macnica has shared that at the NAB Show in Las Vegas, which starts April 19, the company will showcase new capabilities for its MEP100 SmartNIC, introducing support for both 100GbE and 25GbE media flows on a single card, along with expanded support for GPU-based processing, Windows-based applications and real-time metadata.
These enhancements are designed to provide broadcasters and product developers greater flexibility to scale SMPTE ST 2110 IP workflows based on channel density, infrastructure and application requirements, while integrating more easily into modern software-driven environments.
“Never before has this kind of flexibility been offered in a SmartNIC,” said Andrew Starks, Director of Product Marketing at Macnica. “Our customers are not locked into a single network speed, and we are continuing to expand the platform to better support GPU processing, software-based production tools and emerging workflow requirements.”
At 25GbE, the MEP100 supports workflows such as graphics generation, video server playout, including applications leveraging low-latency JPEG-XS compression. At 100GbE, the same card enables high-density, uncompressed workflows for switching, multi-channel processing and large-scale live production environments.
In addition to expanded network flexibility, the MEP100 introduces enhanced support for GPU-accelerated workflows through GPUDirect, enabling efficient transfer of media streams into GPU memory for processing. On Windows platforms, new DirectShow filter support allows integration with applications such as vMix, simplifying adoption within existing production environments. The platform also adds support for SMPTE ST 2110-41 Fast Metadata, enabling the transport of time-aligned metadata alongside video and audio streams.
“These additions are part of a continued effort to make ST 2110 workflows more accessible to software-based production and processing environments,” said Starks. “It’s about giving developers and end users practical ways to build and scale systems using familiar tools and modern compute platforms.”
— Reporting by Brian Galante
