for f in 5 6; do { time for i in {1..10}; do blastn -query NC_012808.1.fna -subject NC_012808.1.fna -outfmt "$f" > "$f"_file.tmp; done ; } 2> "$f"_time.txt; done
Here user+sys of 10x blastn of this genome against itself was ca. 37s for xml and 30s for tabular. I presume the difference is mainly related to IO. The xml output is 26M whereas the tabular output is 384K.
I doubt it takes that much longer. Could you perhaps provide more details?