runtime_eval.sh

#!/bin/bash -e
# runtime_eval.sh: Runtime evaluation for amphi. Written by S. Wendzel, 2025/2026.
# Tested using macOS.
# This script measures the time it takes to let amphi compress data using
# different chunk sizes, dictionary sizes.

export LANG=C
export CLANG=$LANG
# configuration parameters
CHUNK_SIZE_LIST="4 5 6 7 8 9 10 12 15 20 25 30 35 40 45 50"
DICT_SIZE_LIST="10000 50000 100000 200000 500000 1000000 2000000 4000000 " #5000000 8388606" #max
aDICT=empty.dict_a # for incr
A_FILE=A #temporary file containing only "A" characters
MSG_FILE=msg.file
MESSAGE_SIZES="4 8 10 12 14 18 22 26 30 34 38 42 50"

perl -e 'print "A"x40000000' > $A_FILE;
chmod +t amphi

echo "Run this script as many times as you want experiments to be executed. By default, each configuration is used once!" >&2
echo "MessageSize;DictFileSize;ChunkSize;ComputationTime"
for MESSAGE_SIZE in $MESSAGE_SIZES ; do
    #echo "MESSAGE_SIZE=$MESSAGE_SIZE"

    # The following dd if=$A_FILE is the worst-case setting, as we search for "AAAA..."
    # but such a string cannot be found in the dictionary, at all. For the best-case
    # setting, we need to search for a string composed with if=/dev/zero.
    # Thus,
    #   worst-case setting: if=$A_FILE
    #   best-case setting:  if=/dev/zero
    dd if=/dev/zero of=$MSG_FILE count=$MESSAGE_SIZE 2>/dev/null
    for DICTSIZE in $DICT_SIZE_LIST ; do
        #echo "   DICTSIZE=$DICTSIZE"
        dd if=/dev/zero of=$aDICT count=$DICTSIZE 2>/dev/null
        for CHUNK_SIZE in $CHUNK_SIZE_LIST ; do
            #echo "     Chunk size: ${CHUNK_SIZE}:"
            # amphi chunk-size  dict-file    msg-file  compr-file ["incr"] [verbose]
            TIMEDIFF=`./amphi $CHUNK_SIZE $aDICT ${MSG_FILE} compr.out | awk '{print $3}'`
            echo "$MESSAGE_SIZE;$DICTSIZE;$CHUNK_SIZE;$TIMEDIFF"
        done
    done
done