Scaling up to 256 GFLOPS, Mercury Computer Systems’ VantageRT HCD systems double the processing power previously available in a PCI-based system.
Originally designed for medical imaging applications, VantageRT systems now meet application requirements of additional markets including semiconductor wafer inspection, ground-based radar, and image exploitation.
The new system extends Mercury’s PCI-based VantageRT product family, increasing compute density by combining the RACE++ interconnect architecture with four Motorola 500-MHz PowerPC 7410 microprocessors on each PCI module.
The VantageRT HCD system scales up to 64 G4 processors working together in an industrial PC chassis delivering up to 256 billion floating-point operations per second (GFLOPS) of processing power.
The scalability of the VantageRT HCD system is enabled by the RACE++ switch fabric. Each processor gets a connection to the RACE++ interconnect, which provides multiple, concurrent communications between processors. Each module has three over-the-top RACE++ ports, maintaining the balance between processing power and interprocessor communication bandwidth.
The VantageRT HCD compute nodes are identical to those used in Mercury’s VME and CompactPCI systems, so they run the same real-time software. It is supported by Mercury’s real-time software environment, including an integrated development environment, multiprocessor communications libraries, and algorithm libraries optimized for the AltiVec vector-processing engine.
The VantageRT HCD is available 60 days upon receipt of order.