Nowdays we are to step forward from multi-core to many-core computing paradigm.
To realize powerful computing power within low power consumption in many-core processor environment, we fatefully encounter with some issues like
- virtualization techniques
- context switching expanse increasing according to the number of processor cores when mission critical job is running
- distributed shared memory issues
- thread synchronization issues
- third tier to parallelism requiring new API
- code portability
- task-aware core cluster local power management algorithm
- threaded and cascading interrupt handlers for enormously increasing trap and interrupts according to increasing number of processor cores
- processor core routing algorithm to utilize bus in a most efficient way.
- and more
We are doing research on the many-core processor architecture and operating system itself and focusing on how to leverage our many-core platform hundreds and thousands times efficient and powerful than legacy multi-core platforms.