Intel Cilk Research Articles

Intel Cilk Plus for complex parallel algorithms: “Enormous Fast Fourier Transforms” (EFFT) library

In this paper we demonstrate the methodology for parallelizing the computation of large one-dimensional discrete fast Fourier transforms (DFFTs) on multi-core Intel Xeon processors. DFFTs based on the recursive Cooley–Tukey method have to control cache utilization, memory bandwidth and vector hardware usage, and at the same time scale across multiple threads or compute nodes. Our method builds on a single-threaded Intel Math Kernel Library (MKL) implementation of real-to-complex DFFT, and uses the Intel Cilk Plus framework for thread parallelism. We demonstrate the ability of Intel Cilk Plus to handle parallel recursion with nested loop-centric parallelism without tuning the code to the number of cores or cache metrics. The result of our work is a library called EFFT that performs 1D DFTs of size 2N for N ≥ 21 faster than the corresponding Intel MKL parallel DFT implementation by up to 1.5 × , and faster than FFTW by up to 2.5x. The code of EFFT is available for free download under the GPLv3 license. This work provides a new efficient DFFT implementation, and at the same time demonstrates an educational example of how computer science problems with complex parallel patterns can be optimized for high performance using the Intel Cilk Plus framework.

Steal Locally, Share Globally

In a general-purpose computing system, several parallel applications run simultaneously on the same platform. Even if each application is highly tuned for that specific platform, additional performance issues are arising in such a dynamic environment in which multiple applications compete for the resources. Different scheduling and resource management techniques have been proposed either at operating system or user level to improve the performance of concurrent workloads. In this paper, we propose a task-based strategy called "Steal Locally, Share Globally" implemented in the runtime of our parallel programming model GPRM (Glasgow Parallel Reduction Machine). We have chosen a state-of-the-art manycore parallel machine, the Intel Xeon Phi, to compare GPRM with some well-known parallel programming models, OpenMP, Intel Cilk Plus and Intel TBB, in both single-programming and multiprogramming scenarios. We show that GPRM not only performs well for single workloads, but also outperforms the other models for multiprogramming workloads. There are three considerations regarding our task-based scheme: (i) It is implemented inside the parallel framework, not as a separate layer; (ii) It improves the performance without the need to change the number of threads for each application (iii) It can be further tuned and improved, not only for the GPRM applications, but for other equivalent parallel programming models.

Scientific computations on multi-core systems using different programming frameworks

Numerical linear algebra is one of the most important forms of scientific computation. The basic computations in numerical linear algebra are matrix computations and linear systems solution. These computations are used as kernels in many computational problems. This study demonstrates the parallelisation of these scientific computations using multi-core programming frameworks. Specifically, the frameworks examined here are Pthreads, OpenMP, Intel Cilk Plus, Intel TBB, SWARM, and FastFlow. A unified and exploratory performance evaluation and a qualitative study of these frameworks are also presented for parallel scientific computations with several parameters. The OpenMP and SWARM models produce good results running in parallel with compiler optimisation when implementing matrix operations at large and medium scales, whereas the remaining models do not perform as well for some matrix operations. The qualitative results show that the OpenMP, Cilk Plus, TBB, and SWARM frameworks require minimal programming effort, whereas the other models require advanced programming skills and experience. Finally, based on an extended study, general conclusions regarding the programming models and matrix operations for some parameters were obtained.

Multi-Core Program Optimization: Parallel Sorting Algorithms in Intel Cilk Plus

New performance leaps has been achieved with multiprogramming and multi-core systems. Present parallel programming techniques and environment needs significant changes in programs to accomplish parallelism and also constitute complex, confusing and error-prone constructs and rules. Intel Cilk Plus is a C based computing system that presents a straight forward and well-structured model for the development, verification and analysis of multicore and parallel programming. In this article, two programs are developed using Intel Cilk Plus. Two sequential sorting programs in C/C++ language are converted to multi-core programs in Intel Cilk Plus framework to achieve parallelism and better performance. Converted program in Cilk Plus is then checked for various conditions using tools of Cilk and after that, comparison of performance and speedup achieved over the single-core sequential program is discussed and reported.

Energy-efficient work-stealing language runtimes

Work stealing is a promising approach to constructing multithreaded program runtimes of parallel programming languages. This paper presents HERMES, an energy-efficient work-stealing language runtime. The key insight is that threads in a work-stealing environment -- thieves and victims - have varying impacts on the overall program running time, and a coordination of their execution "tempo" can lead to energy efficiency with minimal performance loss. The centerpiece of HERMES is two complementary algorithms to coordinate thread tempo: the workpath-sensitive algorithm determines tempo for each thread based on thief-victim relationships on the execution path, whereas the workload-sensitive algorithm selects appropriate tempo based on the size of work-stealing deques. We construct HERMES on top of Intel Cilk Plus's runtime, and implement tempo adjustment through standard Dynamic Voltage and Frequency Scaling (DVFS). Benchmarks running on HERMES demonstrate an average of 11-12% energy savings with an average of 3-4% performance loss through meter-based measurements over commercial CPUs.

Energy-efficient work-stealing language runtimes

Work stealing is a promising approach to constructing multithreaded program runtimes of parallel programming languages. This paper presents HERMES, an energy-efficient work-stealing language runtime. The key insight is that threads in a work-stealing environment -- thieves and victims - have varying impacts on the overall program running time, and a coordination of their execution "tempo" can lead to energy efficiency with minimal performance loss. The centerpiece of HERMES is two complementary algorithms to coordinate thread tempo: the workpath-sensitive algorithm determines tempo for each thread based on thief-victim relationships on the execution path, whereas the workload-sensitive algorithm selects appropriate tempo based on the size of work-stealing deques. We construct HERMES on top of Intel Cilk Plus's runtime, and implement tempo adjustment through standard Dynamic Voltage and Frequency Scaling (DVFS). Benchmarks running on HERMES demonstrate an average of 11-12% energy savings with an average of 3-4% performance loss through meter-based measurements over commercial CPUs.

Composable Parallel Patterns with Intel Cilk Plus

Intel Cilk Plus extends C and C++ to enable writing composable deterministic parallel software that can exploit both the thread and vector parallelism commonly available in modern hardware.

Application of parallel ant colony algorithm based on TBB and Cilk++ in path optimization

An Ant Colony Algorithm (ACA) with rollback mechanism was proposed for obtaining optimized path,which traverses all necessary nodes in real road network.In case of the large scale real road network,the convergence speed of traditional sequential ACA was quite slow.Therefore,two parallel ACAs based on Intel Threading Building Blocks (TBB) and Cilk++ parallel programming model were designed separately.Both of them were easier to operate with simple instructions than complicated thread triggering and critical resource boundary recognition.Therefore,these two models were easier to develop and more applicable compared with WinAPI multi-threaded based parallel ACA (pACA).The experimental results show that pACA based TBB is nearly identical with pACA based on WinAPI in terms of efficiency,while pACA based on Cilk++ exhibits higher efficiency and better speed-up than pACA based on WinAPI in dual-core system.