SC21 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

Accurate Throughput Prediction of Basic Blocks on Recent Intel Microarchitectures

Authors: Andreas Abel and Jan Reineke (Saarland University)

Abstract: Tools to predict the throughput of basic blocks on a specific microarchitecture are useful to optimize software performance and to build optimizing compilers. In recent work, several such tools have been proposed. The accuracy of their predictions, however, has been shown to be relatively low. To a significant degree, these inaccuracies are due to elements of the pipelines of recent CPUs that are not taken into account by previous tools. We propose a new parametric pipeline model that is significantly more detailed than previous models. Based on this model, we develop a simulator for predicting the throughput of basic blocks.

Our tool supports all Intel Core microarchitecture generations released in the last decade. We evaluate it on an improved version of the BHive benchmark suite. On many recent microarchitectures, its predictions are more accurate than the predictions of state-of-the-art tools by more than an order of magnitude.

Best Poster Finalist (BP): no

Poster: PDF
Poster summary: PDF

Back to Poster Archive Listing