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.
This sample is modified from the original CLOC sample code. Please make sure you have set up the HSA environment before using this sample.
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.
- TIMES: Is the number of repeated times to minimize the overhead of each loop iteration. It is similar to the concept of loop-unrolling.
- Run
./run_set.sh
and the result will be shown on the screen as well as dumped to output folder - 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
make
to buildmake test
to build and runmake clean
to clean the generated files