Address-stride assisted approximate load value prediction in GPUs
Address-stride assisted approximate load value prediction in GPUs
复制标题
GPU 中的地址跨距辅助近似负载值预测
DOI：
10.1145/3330345.3330362
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Jog, Adwait
中科院分区：
文献类型：
--
作者：
Wang, Haonan;Ibrahim, Mohamed;Mittal, Sparsh;Jog, Adwait
价值预测有望显着提高性能和能源效率。但是，如果预测值不正确，则会由于执行回滚而观察到显著的性能开销。为了解决这些开销，引入了值近似，其利用了这样的观察，即只要由于值误预测而导致的应用级质量损失对于用户来说是可接受的，就不需要回滚。然而，在图形处理单元（GPU）的上下文中，我们的评估表明，现有的近似值预测器在提高预测精度方面不是最佳的，因为它们不考虑内存请求顺序，这是确定值预测精度的关键特征。结果，总体数据移动减少益处被封顶，因为有必要限制预测值的百分比（即，为此，我们提出了一个新的地址步幅辅助近似值预测器（ASAP），显式地考虑内存地址和它们的请求顺序信息，以提供高的值预测精度。我们利用我们的新观察，内存请求地址之间的步幅和它们对应的数据值之间的步幅在几个应用程序中高度相关。因此，ASAP只预测那些在地址中具有规则步幅的请求的值。我们在各种GPGPU应用程序上进行ASAP评估。结果表明，ASAP可以显着提高价值预测精度比以前提出的机制在相同的覆盖范围，或可以实现更高的覆盖率（导致更高的性能/能量改善）在一个固定的错误阈值。
Value prediction holds the promise of significantly improving the performance and energy efficiency. However, if the values are predicted incorrectly, significant performance overheads are observed due to execution rollbacks. To address these overheads, value approximation is introduced, which leverages the observation that the rollbacks are not necessary as long as the application-level loss in quality due to value misprediction is acceptable to the user. However, in the context of Graphics Processing Units (GPUs), our evaluations show that the existing approximate value predictors are not optimal in improving the prediction accuracy as they do not consider memory request order, a key characteristic in determining the accuracy of value prediction. As a result, the overall data movement reduction benefits are capped as it is necessary to limit the percentage of predicted values (i.e., prediction coverage) for an acceptable value of application-level error.To this end, we propose a new Address-Stride Assisted Approximate Value Predictor (ASAP) that explicitly considers the memory addresses and their request order information so as to provide high value prediction accuracy. We take advantage of our new observation that the stride between memory request addresses and the stride between their corresponding data values are highly correlated in several applications. Therefore, ASAP predicts the values only for those requests that have regular strides in their addresses. We evaluate ASAP on a diverse set of GPGPU applications. The results show that ASAP can significantly improve the value prediction accuracy over the previously proposed mechanisms at the same coverage, or can achieve higher coverage (leading to higher performance/energy improvements) under a fixed error threshold.
登录
查看更多内容
FlexJava：对安全和模块化近似编程的语言支持
DOI：
10.1145/2786805.2786807
发表时间：
2015
期刊：
Proceedings of the 2015 10th Joint Meeting on Foundations of Software Engineering
影响因子：
--
作者：
Jongse Park;H. Esmaeilzadeh;Xin Zhang;M. Naik;William R. Harris
通讯作者：
William R. Harris
流水线处理器的加载指令单元
影响因子：
1.3
作者：
R. Eickemeyer;S. Vassiliadis
通讯作者：
S. Vassiliadis
DOI：
10.1109/hpca.2018.00030
发表时间：
2018-02
期刊：
2018 IEEE International Symposium on High Performance Computer Architecture (HPCA)
影响因子：
--
作者：
Haonan Wang;Fan Luo;M. Ibrahim;Onur Kayiran;Adwait Jog
通讯作者：
Haonan Wang;Fan Luo;M. Ibrahim;Onur Kayiran;Adwait Jog
基于上下文的值预测器的实现
DOI：
--
发表时间：
1997
期刊：
影响因子：
--
作者：
Yiannakis Sazeides;James E. Smith
通讯作者：
James E. Smith
使用基于数据流的上下文进行准确的价值预测
DOI：
--
发表时间：
2001
期刊：
Proceedings 2001 International Conference on Parallel Architectures and Compilation Techniques
影响因子：
--
作者：
Renju Thomas;M. Franklin
通讯作者：
M. Franklin