Go to file
2022-01-04 13:51:53 +01:00
asm Update force kernel to be integrated 2021-11-08 14:05:29 +01:00
src Porting atom velocity memory layout to AoS, porting velocity integration to CUDA, adding measurements + logbook update 2022-01-01 18:18:12 +01:00
util Add cache sets histogram script, update gather plot script and add results for icx32 2021-07-29 17:01:30 +02:00
.gitignore Porting atom velocity memory layout to AoS, porting velocity integration to CUDA, adding measurements + logbook update 2022-01-01 18:18:12 +01:00
config.mk Fixed linker error by putting includes and cuda function in extern 'C' 2021-11-11 14:49:29 +01:00
evaluate_cpu_omp_perf.sh Reminder for likwid perf measurements 2022-01-04 13:51:53 +01:00
evaluate_gpu_perf_per_thread.sh Also pinnend neighbor-struct memory, added additional performance measurements, added nvprof result to logbook 2021-12-18 15:58:56 +01:00
include_CLANG.mk Streamline build system 2021-03-24 08:43:44 +01:00
include_GCC.mk Restructure timing and instrumentation. Add performance metric. 2021-04-15 14:55:02 +02:00
include_ICC.mk Restructure timing and instrumentation. Add performance metric. 2021-04-15 14:55:02 +02:00
include_LIKWID.mk Add LIKWID Option. Allow to overwrite with asm variant. 2021-06-11 09:38:34 +02:00
include_NVCC.mk Fixed linker error by putting includes and cuda function in extern 'C' 2021-11-11 14:49:29 +01:00
LICENSE Switch License to LGPL3 2020-08-19 10:47:40 +02:00
Makefile getTimeStamp() couldn't get linked 2021-11-11 08:03:56 +01:00
README.md Streamline build system 2021-03-24 08:43:44 +01:00

MD-Bench

A simple, sequential C implementation of the Mantevo miniMD benchmark in less than 1000 LOC.

Build

  1. Open config.mk and edit the TAG value according to the tool chain used. Currently supported is GCC, CLANG (LLVM), and ICC (Intel).
  2. Change DATA_LAYOUT and DATA_TYPE if desired in config.mk.
  3. Open and adapt the compiler flags in <include_<TOOLCHAIN>.mk, e.g. in include_ICC.mk for the Intel tool chain.
  4. Build the binary calling make.

You can clean intermediate build results with make clean, and all build results with make distclean. You have to call make clean before make if you changed the build settings.

Configuration

Currently all settings apart from the options described below are hard-coded in main.c.

Run the benchmark

Without any options 200 steps with system size 32x32x32 is used.

The default can be changed using the following options:

-n / --nsteps <int>:  set number of timesteps for simulation
-nx/-ny/-nz <int>:    set linear dimension of systembox in x/y/z direction