Introduction

hsa-micro-benchmarks is a set of micro-benchmarks designed for HSA platforms to profile the detailed system behavior. This is still an on-going work and not well-tested yet. If you have any problems, feel free to post an issue.

Pre-requests

This sample is modified from the original CLOC sample code. Please make sure you have set up the HSA environment before using this sample.

Batch testing

Each file in the folder test_set is a test set of micro-benchmarks with some arguments in the begining lines of the file. Those arguments like TIMES are parsed and processed by script in order to execute properly.

Arguments of Micro-benchmarks

• TIMES: Is the number of repeated times to minimize the overhead of each loop iteration. It is similar to the concept of loop-unrolling.

Execution

1. Run ./run_set.sh and the result will be shown on the screen as well as dumped to output folder
2. Execute head ./output/* to see the result manually

• Run ./run_set.sh [TEST SET FILE] to run a single test. Ex: ./run_set.sh ./test_set/ld_st
• Run ./run_set.sh [TEST SET DIRECTORY] to run tests in that folder. Ex: ./run_set.sh ./test_set

Normal Build an Test

• make to build
• make test to build and run
• make clean to clean the generated files