This repository contains benchmarks for LFI. Each benchmark contains a Makefile
with a bench target that can be used to build and run the benchmark.
There is also a top-level Makefile with the following targets: build,
bench, consolidate, plot. When using consolidate and plot, you should
set MACHINE to an identifier for your current machine (such as aarch64/m2),
so that the data and plot get placed in the appropriate directory in results.
Otherwise the generic name current will be used.
In order to run the benchmarks, you should first install an LFI toolchain, and
then modify the compiler definitions in toolchains as required.
The default configurations have built the following toolchains and placed them in your LFI_ROOT directory:
aarch64-lfi-clang/aarch64-lfi-stores-clang/aarch64-native-clang/
``
The install-lfi-prebuilt.sh script can be used to automatically download
prebuilt versions of these toolchains.
You can find instructions for building these toolchains here.
For running the benchmarks you will also need the lfi-run tool from the LFI
repository.
By default, LFI_ROOT defaults to /opt, but can be set with an environment variable.
export LFI_ROOT=/path/to/lfi