This is a benchmark for testing instruction throughput, GFlops, latency, supports arm, x86.
This benchmark only support Cmake build system and require Cmake version upper than 3.15.2, you can compiler the benchmark fellow the step:
-
clone or download the project
-
choose a test plantform, the current default processor is aarch64. If the processor is other architecture, only need to modify CMakeLists.txt.
-
if the processor architecture is armv7
<set(MGEPEAK_ARCH "armv7")> -
if the processor architecture is x86
<set(MGEPEAK_ARCH "x86")> -
A gcc or clang compier should find by camke through PATH env
mkdir -p build && cd build cmake .. make
-
after build, the executable file megpeak is stored in build drector
if you compiler the project and get the megpeak.
bash ./megpeak