NETLIB-HPL

    # Build HPL using AOCC and AOCL math libraries - Spack will autodetect the AMD CPU generation and use appropriate flags
$ spack install hpl %aocc +openmp ^amdblis threads=openmp ^openmpi fabrics=cma,ucx

    #! /bin/bash
# Load HPL into environment
# NOTE: If you have built multiple versions of HPL with Spack you may need to be
# more specific about which version to load. Spack will complain if your request
# is ambiguous and could refer to multiple packages.
# Please see: (https://spack.readthedocs.io/en/latest/basic_usage.html#ambiguous-specs)
spack load hpl %aocc

### Performance settings ###
# System level tunings
echo 3 > /proc/sys/vm/drop_caches   # Clear caches to maximize available RAM
echo 1 > /proc/sys/vm/compact_memory # Rearrange RAM usage to maximise the size of free blocks
echo 0 > /proc/sys/kernel/numa_balancing # Prevent kernel from migrating threads overzealously
echo 'always' > /sys/kernel/mm/transparent_hugepage/enabled # Enable hugepages for better TLB usage
echo 'always' > /sys/kernel/mm/transparent_hugepage/defrag  # Enable page defragmentation and coalescing

# Capture system specification and OpenMP settings
CORES_PER_L3CACHE=8
NUM_CORES=$(nproc)
# Use 4 threads per MPI rank  - this means 2 ranks per CPU L3cache (Zen 2+) or 1 rank per L3 (Zen 1), If using an "F" part (e.g 75F3) also ensure that OMP_NUM_THREADS is set appropriately
# (Recommended OMP_NUM_THREADS= #cores per L3 cache)
export OMP_NUM_THREADS=4   
export OMP_PROC_BIND=TRUE  # bind threads to specific resources
export OMP_PLACES="cores"   # bind threads to cores

# AMDBlis (BLAS layer) optimizations
export BLIS_JC_NT=1  # (No outer loop parallelization)
export BLIS_IC_NT=$OMP_NUM_THREADS # (# of 2nd level threads – one per core in the shared L3 cache domain):
export BLIS_JR_NT=1 # (No 4th level threads)
export BLIS_IR_NT=1 # (No 5th level threads)

# MPI settings
MPI_RANKS=$(( $NUM_CORES / $OMP_NUM_THREADS ))
RANKS_PER_L3CACHE=$(( $CORES_PER_L3CACHE / $OMP_NUM_THREADS ))
MPI_OPTS=”-np $MPI_RANKS --bind-to core --map-by ppr:$RANKS_PER_L3CACHE:l3cache:pe=$OMP_NUM_THREADS ”

# Running HPL
mpirun $MPI_OPTS xhpl

    HPLinpack benchmark input file
Innovative Computing Laboratory, University of Tennessee
HPL.out     output file name (if any)
6           device out (6=stdout,7=stderr,file)
1           # of problems sizes (N)
430080      Ns  <--- Modify this to change the memory footprint
1           # of NBs
456         # NBs
0           MAP process mapping (0=Row-,1=Column-major)
1           # of process grids (P x Q)
8           Ps <--- Set Ps and Qs to a suitable grid size
8           Qs <--- make sure that Ps * Qs == #MPI Ranks
16.0        threshold
1           # of panel fact<
1           PFACTs (0=left, 1=Crout, 2=Right)
1           # of recursive stopping criterium
4           NBMINs (>= 1)
1           # of panels in recursion
2           NDIVs
1           # of recursive panel fact.
1           RFACTs (0=left, 1=Crout, 2=Right)
1           # of broadcast
3           BCASTs (0=1rg,1=1rM,2=2rg,3=2rM,4=Lng,5=LnM)
1           # of lookahead depth
0           DEPTHs (>=0)
2           SWAP (0=bin-exch,1=long,2=mix)
64          swapping threshold
0           L1 in (0=transposed,1=no-transposed) form
0           U in (0=transposed,1=no-transposed) form
1           Equilibration (0=no,1=yes)
8           memory alignment in double (> 0)

    $ chmod +x ./run_hpl.sh   # make the script executable
$ ./run_hpl.sh