Computing-Kernels Performance Prediction Using data flow Analysis and Microbenchmarking
نویسندگان
چکیده
On modern multi-core processors, the growing gap between memory size, bandwidth and latency compared to computing capability makes the memory hierarchy predominant for performance. The Microkernel-Description-Language based Performance Evaluation Framework, MDL-PEF, accurately predicts optimized inner-loops performance depending on the loop’s data access. The MDL-PEF approach revolves around a data flow description language, MDL. A static analysis step extracts the data flow structures of the assembly code. Then the predictor uses pattern matching against an MDL-Microkernel database for predicting performance. Finally, MDL-PEF provides an automatic tool to initialize a pattern matching database for the target architecture. The overall system can predict the kernel performance on different platforms and optimizations, helping the user choose the best architecture for a given kernel. Preliminary experiments, with a 56 elements database, predict the innermost loop throughput of 636 binary loops of the NAS benchmarks with an average 10% of relative error. The performance predictor is part of the Modular Assembly Quality Analyzer and Optimizer (MAQAO) performance tool framework. Future works will extend MDLPEF to other architecture paradigms and more complex control flows such as outer loops.
منابع مشابه
Prediction of Stage–Discharge Relationship in Bushehr Mand River Using Hybrid Fuzzy and GMDH Methods
Understanding the Stage–Discharge relationship is of great importance in the management and planning of water resources, as well as the design of hydraulic structures, the organization of rivers, and the planning of flood warning systems. With the advancement of science and increasing the speed of computing, new methods called intelligent systems have been introduced, the use of which can be a ...
متن کاملEnsemble Kernel Learning Model for Prediction of Time Series Based on the Support Vector Regression and Meta Heuristic Search
In this paper, a method for predicting time series is presented. Time series prediction is a process which predicted future system values based on information obtained from past and present data points. Time series prediction models are widely used in various fields of engineering, economics, etc. The main purpose of using different models for time series prediction is to make the forecast with...
متن کاملA method of performance estimation for axial-flow turbines based on losses prediction
The main objective in this paper is creating a method for one-dimensional modeling of multi stage axial flow turbine. The calculation used in this technique is based on common thermodynamics and aerodynamics principles in a mean stream line analyses. In this approach, loss models have to be used to determine the entropy increase across each section in the turbine stage. Finally, the analysis an...
متن کاملHybrid Models Performance Assessment to Predict Flow of Gamasyab River
Awareness of the level of river flow and its fluctuations at different times is one of the significant factor to achieve sustainable development for water resource issues. Therefore, the present study two hybrid models, Wavelet- Adaptive Neural Fuzzy Interference System (WANFIS) and Wavelet- Artificial Neural Network (WANN) are used for flow prediction of Gamasyab River (Nahavand, Hamedan, Iran...
متن کاملHybrid Models Performance Assessment to Predict Flow of Gamasyab River
Awareness of the level of river flow and its fluctuations at different times is one of the significant factor to achieve sustainable development for water resource issues. Therefore, the present study two hybrid models, Wavelet- Adaptive Neural Fuzzy Interference System (WANFIS) and Wavelet- Artificial Neural Network (WANN) are used for flow prediction of Gamasyab River (Nahavand, Hamedan, Iran...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2012