異構多核心平台的日漸流行，其性能優越於同構性系統。為了達到最佳的性能，，只能實現如果任務都可以準確地分發到合適的設備。 OpenCL的任務可以調度，充分利用所有可用的設備在一個系統中。如果我們允許調度決策系統和應用程序分配給設備的基礎上分析數據庫和設備負荷，我們都可以提高整個系統的吞吐量並減少單獨的應用程序運行時。
在本文中，我們提出了我們的方法，在對異構平台的支持狀態，以最大限度地提高整體吞吐量。我們的方法的關鍵是準確地估計時，應用程序將在一個給定的移動設備歷史運行信息的基礎上完成執行。此外，我們的調度機制，包括針對OpenCL應用程序狀態化的支持，使我們能夠更有效的調度決策。我們的系統上運行的應用程序與多核CPU和GPU的OpenCL一套評估我們的方法。我們的實驗結果表明，我們的負載均衡機制其效能相對於Round-Robin排班有1.61的效能提升。我們的計劃是有效的調度支持在異構環境中的狀態和無狀態的OpenCL工作。

Heterogeneous multi-core platforms are increasingly popular due to their perceived superior performance over homogenous systems. The best performance, however, can only be achieved if tasks can be accurately distributed to the right devices. OpenCL tasks can be dispatched to take advantage of all the available devices in a system. If we instead allow the system to make scheduling decisions and assign applications to devices based on profiling database and device loading, we can both increase overall system throughput and decrease individual application runtimes.

In this paper, we present our methodologies in supporting stateful over heterogeneous platforms in order to maximize overall throughput. The key to our approach is accurately estimating when an application would finish execution on a given device based on historical runtime information. In addition, our scheduling mechanism includes stateful support for OpenCL applications, allowing us to make scheduling decisions more effective. We evaluate our approach with a set of OpenCL applications running on a system with multicore CPUs and GPUs. The result of our experiment shows that our load-balancing mechanism is 1.61 speedup over Round-Robin mechanism. Our schemes are effective in supporting the dispatching of stateful and stateless OpenCL tasks over heterogeneous environments.

[1] Khronos OpenCL Working Group std., The OpenCL Specification Version 1.1,
Khronos OpenCL Working Group Std., Sep. 2010.
[2] AMD, “AMD developer central - OpenCL zone.” [Online]. Available:
http://developer.amd.com/zones/openclzone/Pages/default.aspx
[3] Nvidia, “NVIDIA developer zone - OpenCL.” [Online]. Available: http:
//developer.nvidia.com/opencl
[4] Intel, “Intel SDK for OpenCL applications 2012 release.” [Online]. Available:
http://software.intel.com/en-us/articles/opencl-release-notes/
[5] B. Gaster, L. Howes, D. R. Kaeli, P. Mistry, and D. Schaa, Heterogeneous Com-
puting with OpenCL, 1st ed. Morgan Kaufmann, August 2011.
[6] C.-K. Chen, Y.-H. Chang, Y.-T. Chen, C.-C. Yang, and J.-K. Lee, “Switching
supports for stateful object remoting on network processors,” in The Journal of
Supercomputing, vol. 40, 2007, pp. 281–298.
[7] J. Lee, J. Kim, S. Seo, S. Kim, J. Park, H. Kim, T. T. Dao, Y. Cho, S. J. Seo, S. H.
Lee, S. M. Cho, H. J. Song, S.-B. Suh, and J.-D. Choi, “An OpenCL framework for heterogeneous multicores with local memory,” in PACT ’10: Proceedings of the 19th international conference on Parallel architectures and compilation
techniques, 2010, pp. 193–204.
[8] J. Kim, S. Seo, J. Lee, J. Nah, G. Jo, and J. Lee, “SnuCL: An OpenCL frame-
work for heterogeneous CPU/GPU clusters,” in ICS ’12: Proceedings of the 26th
International Conference on Supercomputing, 2012, pp. 341–352.
[9] M. Y. Wu, S. Hariri, and H. Topcuouglu, “Performance-Effective and Low-
Complexity Task Scheduling for Heterogeneous Computing,” in IEEE Trans
on Parallel and Distributed Systems IEEE Trans on Parallel and Distributed
Systems, vol. 13, 2002, pp. 260–274.
[10] Y. Kwok and I. Ahmad, “Dynamic Critical-Path Scheduling: An Effective Tech-
nique for Allocating Task Graphs to Multiprocessors,” in IEEE Trans. Parallel
and Distributed System, vol. 7, no. 5, May 1996, pp. 506–521.
[11] M. Wu, W. Shu, and J. Gu, “Local Search for DAG Scheduling and Task As-
signment,” in Proc. 1997 Intl Conf. Parallel Processing, 1997, pp. 174–180.