Thread fork/join techniques for multi-level parallelism exploitation in NUMA multiprocessors

Abstract

This paper presents some techniques for efficient thread forking and joining in parallel execution environments, taking into consideration the physical structure of NUMA machines and the support for multi-level parallelization and processor grouping. Two work generation schemes and one join mechanism are designed, implemented, evaluated and compared with the ones used in the IFUX MP library, an efficient implementation which supports a single level of parallelism.Supporting multiple levels of parallelism is a current research goal, both in shared and distributed memory machines. Our proposals include a first work generation scheme (GWD, or global work descriptor) which supports multiple levels of parallelism, but not processor grouping. The second work generation scheme (LWD, or local work descriptor) has been designed to support multiple levels of parallelism and processor grouping. Processor grouping is needed to distribute processors among different parts of the computation and maintain the working set of each processor across different parallel constructs.The mechanisms are evaluated using synthetic benchmarks, two SPEC95fp applications and one NAS application. The performance evaluation concludes that: i) the overhead of the proposed mechanisms is similar to the overhead of the existing ones when exploiting a single level of parallelism, and ii) a remarkable improvement in performance is obtained for applications that have multiple levels of parallelism. The comparison with the traditional single-level parallelism exploitation gives an improvement in the range of 3065% for these applications.

Automatic Generation of OpenMP Directives and Its Application to Computational Fluid Dynamics Codes

“…We attribute the performance degradation in the directive implementation of LU to less data locality and larger synchronization overhead in the 1-D pipeline used in the OpenMP version as compared to the 2-D pipeline used in the MPI version. This is consistent with the result of a study from [12].…”

“…The pipeline algorithm is used for parallelizing the NAS benchmark LU in Sect. 4.1 and also described in [12].…”

Performance of OSCAR Multigrain Parallelizing Compiler on SMP Servers

“…NANOS compiler [12] based on Parafrase2 has been trying to exploit the multi-level parallelism including the coarse grain parallelism by using extended OpenMP API. The OSCAR multigrain parallelizing compiler [13] exploits the coarse grain task parallelism among loops, subroutines and basic blocks [14], and the near fine grain parallelism among statements inside a basic block [15] in addition to the conventional loop parallelism among iterations.…”

Cache Optimization for Coarse Grain Task Parallel Processing Using Inter-Array Padding

“…NANOS compiler [3] uses multi level parallelism by using the extended OpenMP API. PROMIS compiler [4] integrates loop level parallelism and instruction level parallelism using a common intermediate language.…”