[Bug-tar] GNU tar, star and BSD tar speed comparision +script

[Jan Psota]

Concatenated output of tcp script (below).
I replaced 'tar' with 'star' by hand. Have fun :-).

1. Tested on Athlon 1500MHz, RAM 512MB, disk speed 50MB/s, reiserfs
best time of 3 repetitions, /tmp/drv, 82M, 4105 files
program operation       real    user    system  speed
bsdtar  create          0.71    0.10    0.60    102253 KB/s
star    create          1.18    0.02    1.05     61532 KB/s
gnutar  create          0.67    0.01    0.65    108370 KB/s

bsdtar  list            0.05    0.02    0.02   1452160 KB/s
star    list            0.59    0.02    0.46    123064 KB/s
gnutar  list            0.22    0.02    0.19    330036 KB/s

bsdtar  extract         0.74    0.08    0.65     98118 KB/s
star    extract         1.29    0.03    1.16     56285 KB/s
gnutar  extract         0.68    0.04    0.64    106776 KB/s

bsdtar  compare         unsupported
star    compare         1.04    0.22    0.71     69815 KB/s
gnutar  compare         0.59    0.16    0.43    123064 KB/s

best time of 3 repetitions, /home/jasiu, 2,1G, 8377 files
program operation       real    user    system  speed
bsdtar  create          119.78  1.55    28.81     17461 KB/s
star    create          120.44  0.26    40.49     17365 KB/s
gnutar  create          118.34  0.49    28.43     17674 KB/s

bsdtar  list              0.21  0.06     0.14   9960157 KB/s
star    list             37.46  0.18    17.96     55836 KB/s
gnutar  list             36.02  0.29     8.17     58068 KB/s

bsdtar  extract         148.44  3.54    28.94     14090 KB/s
star    extract         149.99  0.40    38.77     13945 KB/s
gnutar  extract         147.57  0.34    27.94     14173 KB/s

bsdtar  compare         unsupported
star    compare         119.52  6.01    26.80     17500 KB/s
gnutar  compare         615.32  3.84    16.94      3399 KB/s

2. Tested on Athlon 64 2x2.2GHz, RAM 2GB, disk 60MB/s, reiserfs
best time of 3 repetitions, src="" 202M, 1890 files
program operation       real    user    system  speed
bsdtar  create          0.34    0.03    0.30    596447 KB/s
star    create          0.43    0.00    0.58    471609 KB/s
gnutar  create          0.32    0.00    0.31    633812 KB/s

bsdtar  list            0.01    0.00    0.01  20282000 KB/s
star    list            0.31    0.00    0.20    654258 KB/s
gnutar  list            0.11    0.01    0.10   1843818 KB/s

bsdtar  extract         0.36    0.02    0.33    563388 KB/s
star    extract         0.44    0.03    0.56    460954 KB/s
gnutar  extract         0.32    0.01    0.30    633812 KB/s

bsdtar  compare         -
star    compare         0.37    0.13    0.36    548162 KB/s
gnutar  compare         0.45    0.20    0.24    450711 KB/s

best time of 3 repetitions, /home, 3,0G, 7443 files
program operation       real    user    system  speed
bsdtar  create           77.98  0.56    11.60     39130 KB/s
star    create           72.78  0.13    13.40     41923 KB/s
gnutar  create           79.02  0.20    11.87     38613 KB/s

bsdtar  list              0.05  0.02     0.02  61025140 KB/s
star    list             48.02  0.06     4.45     63541 KB/s
gnutar  list             47.82  0.12     3.00     63807 KB/s

bsdtar  extract         127.53  0.65    13.15     23925 KB/s
star    extract         132.70  0.11    10.82     22993 KB/s
gnutar  extract         132.04  0.13    12.62     23108 KB/s

bsdtar  compare         -
star    compare         108.33  1.17     7.90     28166 KB/s
gnutar  compare         240.63  3.27     7.06     12680 KB/s

--
Jan Psota

-------------------------------------tcp-------------------------------------
#!/bin/sh
# tar comparision program
# 2007-10-23 Jan Psota

n=3                                     # number of repetitions
TAR=(bsdtar tar gnutar)                 # TApeArchivers to compare
OPN=(create list extract compare)       # operations

test $# -ge 2 || {
        echo -e "usage:\t$0 tmp_dir src_dir...

TCP stands for Tar Comparision Program here.
It currently compares: BSD tar, star, and GNU tar in archive creation,
listing, extraction and archive-to-extracted comparision.
Tcp prints best time of n=$n repetitions.

Tcp will create temporary archive named tmp_dir/tcp.tar and tmp_dir/tmp
to store extracted files (after normal exit they will be removed).
Tcp does not check filesystem tmp_dir is on for free space, so make sure
there is two times more space in tmp_dir than biggest of src_dirs uses.
Tcp expects 'tar' linked to 'star' because of star's programmer's wish ;-)
[-no-fsync and other...], so original GNU tar should be renamed to 'gnutar'
for tests.
Do not use white space in arguments."
        exit 0
}
tmp=$1; shift
SRC=""> dst=$tmp/tcp.tar
dst_path=$tmp/tmp
test -e $dst -o -e /tmp/tcp \
        && { echo "$dst or /tmp/tcp exists, exiting"; exit 1; }
mkdir $dst_path || exit 2

use_times ()
{

        LC_ALL=C awk -vN=$n -vL="`du -k $dst`" -vOFS="\t" -vORS="" '
                /real|user|sys/ { T[$1] = $2 }
                /sys/ { S=(T["real"]>0) ? S=(L+0)/T["real"] : 0
                        print "",T["real"], T["user"], T["sys"]
                        printf "%10.1d KB/s\n", S }' \
                /tmp/tcp | sort | head -1
}

for src in address@hidden; do
        test -d $src || { echo "'$src' is not a directory"; continue; }
        echo -e "best time of $n repetitions, src="" archive=$dst,\n"\
"\textract to $dst_path, "`du -sh $src | awk '{print $1}'`",\
" `find $src | wc -l`" files"
        echo -e "program\toperation\treal\tuser\tsystem\tspeed"
        let op_num=0
        for op in "cf $dst -C $src ." "tf $dst" "xf $dst -C $dst_path" \
                "f $dst -C $dst_path --diff"; do
                let tar_num=0
                for tar in address@hidden; do
                        cat /dev/null > /tmp/tcp
                        echo -en "$tar\t${OPN[op_num]}\t"
                        for ((i=1; i<=$n; i++)); do
                                echo $op | grep -q ^cf && rm -f $dst
                                echo $op | grep -q ^xf &&
                                        { chmod -R u+w $dst_path
                                        rm -rf $dst_path; mkdir $dst_path; }
                                sync
                                if echo $op | grep -q ^f; then  # op == compare
                                        time -p $tar $op > /dev/null
                                else    # op in (create | list | extract)
                                        time -p $tar $op > /dev/null || exit 3
                                fi 2>> /tmp/tcp
                        done
                        use_times
                        let tar_num++
                done
                let op_num++
                echo
        done
done
rm -rf $dst_path $dst /tmp/tcp