Scalability of continuation-based fine-grained multithreading in handling multiple I/O requests on FUCE

Multi-tasking operating systems not only handle concurrent tasksbut also have concurrency and parallelism of various granularities in themselves. We claim that operating systems need to be developed based on a computation model that can deal with concurrency and parallelism of various granularities. In order to investigate this claim, we have been developing an operating system called CEFOS based on a dataflow-based computation model. A program for CEFOS consists of zero-waitthreads, each of which runs to completion without suspension once started. Synchronization between zero-wait threads is performed in a dataflow manner according to their continuation relations. Handler routines for I/O devices are also realized with zero-wait threads and executed under the continuation-based multithreading mechanism. We can eliminate "interrupts" that interfere with the execution ofinstruction streams in typical conventional approaches, and we can naturally handle concurrency and exploit parallelism in programs even for I/O-centric computation. In this paper, after introducing our model and our operating system based on the model, we discuss implementation issues on Fuce, which is a continuation-based multithreading processor dedicated to fine-grained multithreading. Then we evaluate the scalability of our system with the number of execution units and I/O devices.