Load data with cpimport and colxml Job is significantly slower than forcing in parallel cpimport

it's observed that load 1TB data with cpimport mode m1 and colxml Job id
is done nearly twice slower than
loading tables with the same data in parallel simultaneously via script

Actually when use cpimport with colxml job it's observed that tables are loaded one after another;
While run cpimport with colxml job is customized with more usability trends
it would be nice to have similar load rates to the exec time of forcing manually cpimport in parallel

achieved results from 1TB data load:

load method                                                                          load time

cpimport mode  m1 with  colxml [1TB]:                                  4 hours 1 min 40 sec

cpimport mode  m1 without colxml,  force parallel [1TB] :      2 hours 12 min 21 sec

used scripts :

Used script  to load with cpimport with and colxml  Job

${PCOLXML} $LOAD_DB -j${RAN}

${PCPIMPORT}  -m1 -j${RAN}

Used script  to run multiple cpimport jobs simultaneously

Loop over the tpc-ds tables : ${PCPIMPORT} -m1  ${LOAD_DB} ${i%.*}  insert-data-tables/data/${LOAD_DB}/${i}  &

PCPIMPORT=/home/mariadb-user/mariadb/columnstore/bin/cpimport

PCOLXML=/home/mariadb-user/mariadb/columnstore/bin/colxml

Updated with last load time results with Scale Factor 1000 [1TB data], obtained on Intel Packet with optane drives