MASA: a multithreaded processor architecture for parallel symbolic computing
ISCA '88 Proceedings of the 15th Annual International Symposium on Computer architecture
ISCA '89 Proceedings of the 16th annual international symposium on Computer architecture
APRIL: a processor architecture for multiprocessing
ISCA '90 Proceedings of the 17th annual international symposium on Computer Architecture
Performance Tradeoffs in Multithreaded Processors
IEEE Transactions on Parallel and Distributed Systems
WAITING ALGORITHMS FOR SYNCHRONIZATION IN LARGE-SCALE MULTIPROCESSORS
WAITING ALGORITHMS FOR SYNCHRONIZATION IN LARGE-SCALE MULTIPROCESSORS
Hi-index | 0.00 |
As cache based shared-memory multiprocessors are scaled (the number of processors are increased), there will be an increase in the network latency, due to cache coherence and synchronization delays. These parameters can cause a significant drop in processor utilization. Once a process has remote procedure call or need data from remote side, the processor has to be idle or make context switch to load the next process in which the CPU (Central Processing Unit) time is wasted.By maintaining the multiple threads in the hardware, and switching among them, the multithreaded processor can overlap the waiting time for the synchronization delay so that it decreases the processor idle time.This correspondence describes the multithreaded processor architecture, in which there are a number of hardware contexts per processor. It uses coarsegrain method to schedule threads and two-phase waiting algorithm to synchronize the waiting threads. From this architecture, we can study how many hardware contexts are needed, so the processor utilization can be increased. Moreover, we can study the effect on the utilization by changing the cache miss ratio.