added comments to example_map.sh and further clarification in the README

> NOTE: `HIP_VISIBLE_DEVICES` must be set.

> NOTE: On Lyra, the current Slurm doesn't easily allow for fine-grained process/thread placement so an example mapping script is also included in this repo. It can be modifed and called "in front of" `hello_jobstep` (or any other executable really). The script uses `numactl` to map hardware threads and GPUs to node-local MPI ranks.

> [OPTIONAL]: On Lyra, the current Slurm doesn't easily allow for fine-grained process/thread placement so an example mapping script is also included in this repo. It can be modifed and called "in front of" `hello_jobstep` (or any other executable really). The script uses `numactl` to map hardware threads and GPUs to node-local MPI ranks. NOTE: You will need to use the `srun` argument `--ntasks_per_gpu` with this script.

#!/bin/bash

export APP=$1

#------------------------------------------------------

# You'll need to read in more command line args if your

# executable takes arguments

#------------------------------------------------------

APP=$1

lrank=$(($SLURM_PROCID % 4))

#------------------------------------------------------

# The number of node-local MPI ranks

# The `--ntasks_per_node` flag to srun should be used

#------------------------------------------------------

lrank=$(($SLURM_PROCID % $SLURM_NTASKS_PER_NODE))

#------------------------------------------------------

# Ideally, the number of hardware threads set below

# for each rank with numactl should be the same as

# OMP_NUM_THREADS

#------------------------------------------------------

export OMP_NUM_THREADS=4

export OMP_PLACES=cores

#------------------------------------------------------

# Set hardware threads and GPUs for each node-local

# MPI rank. NOTE: For more than 4 MPI ranks per node, 

# additional cases would need to be added.

#------------------------------------------------------

case ${lrank} in

[0])

export HIP_VISIBLE_DEVICES=0,1

export HIP_VISIBLE_DEVICES=0

numactl --physcpubind=64,65,66,67 $APP

  ;;

[1])

export HIP_VISIBLE_DEVICES=2,3

export HIP_VISIBLE_DEVICES=1

numactl --physcpubind=68,69,70,71 $APP

  ;;

[2])

export HIP_VISIBLE_DEVICES=2

numactl --physcpubind=72,73,74,75 $APP

  ;;

[3])

export HIP_VISIBLE_DEVICES=3

numactl --physcpubind=76,77,78,79 $APP

  ;;

esac