Thread Private Variable Access Optimization Technique for Sunway High-Performance Multi-core Processors

The primary way to achieve thread-level parallelism on the Sunway high-performance multicore processor is to use the OpenMP programming technique. To address the problem of low parallelism efficiency caused by slow access to thread private variables in the compilation of Sunway OpenMP programs, this paper proposes a thread private variable access technique based on privileged instructions. The privileged instruction-based thread-private variable access technique centralizes the implementation of thread-private variables at the compiler level, eliminating the model switching overhead of invoking OS core processing and improving the speed of accessing thread-private variables. On the Sunway 1621 server platform, NPB3.3-OMP and SPEC OMP2012 achieved 6.2% and 6.8% running efficiency gains, respectively. The results show that the techniques proposed in this paper can provide technical support for giving full play to the advantages of Sunway’s high-performance multi-core processors.