Aonix has announced the release of PERC Ultra SMP with support for Concurrent's RedHawk real-time Linux and associated NightStar advanced Linux debugging and analysis tools. PERC Ultra, Aonix's flagship product, targets the same time-critical applications such as simulation and training, data acquisition, imaging, and process control that RedHawk currently serves.

PERC Ultra SMP, a variant of the Aonix family of deterministic and hard real-time virtual machines and Java development tools, provides new ability to execute real-time applications effectively in multi-processor and multi-core systems. The promise of faster execution enabled by multi-core systems is realized by the combination of RedHawk Linux and Ultra SMP where support for the latest Intel and AMD quad-core processors and chipsets provides for more computing power in smaller packages than previously possible.

Aonix and Concurrent have collaborated to bring this joint solution to market in response to requirements of a contractor in the military sector. Ultimately the combined solution will be used in mission planning and weapon systems aboard U.S. Navy warships. Earlier versions of PERC Ultra for single processors running on RedHawk Linux were previously deployed as part of the Aegis Weapon System Open Architecture Program. It was selected to enhance the capabilities and service life of the U.S. Navy's premier surface combat system while reducing costs. Projects like these benefit from the ability to address multi-core hardware platforms that provide faster, more efficient execution of real-time applications.

Ultra SMP sets the CPU affinity for each Java thread to an assigned CPU or core in the RedHawk Linux system. This provides precise control over which Java threads are running and where they run in the PERC Ultra SMP virtual machine, regardless of any scheduling activities associated with Linux itself. Only the threads that the PERC Ultra SMP VM activates can be run on the Linux kernel, allowing the PERC VM to maintain strict real-time prioritization and priority inheritance within Java applications.

As with PERC Ultra, the SMP version allows higher priority Java threads to immediately interrupt garbage collection to respond to time-critical events and subsequently resume garbage collection and lower priority Java tasks when those higher priority tasks are complete. Deployed as a highly parallel process, the garbage collector within PERC uses idle processors for garbage collection, working in tandem so that one processor can scan stack memory in search of pointers to live objects while another processor relocates previously identified live objects.

The PERC Ultra SMP Virtual Machine provides hooks into Concurrent's NightStar tools to allow user visibility into the execution of individual Java threads, just as it can for native threads running on RedHawk Linux. Java and native threads are displayed together in a synchronized view of all system activity, including system calls, interrupts, and exceptions, along with user-defined event logging. Users can monitor CPU usage, memory use, and detailed process and thread information to analyze the efficiency of their multi-threaded applications and identify ways to improve performance.