Skip to contentSkip to conversation

benchmarks 2.5.40-mm1 and 2.5.39 on quad xeon

DISCUSSION [email protected] 1 participant 1 message Oct 7, 2002

lore
Share & tools
lore
#0 lore raw

Fork improvements continue in the 2.5 tree.

Seconds to build autoconf-2.53 12 times:
seconds
2.5.39 780.3
2.5.40-mm1 750.5 4% faster

ext[23] filesystems built with e2fsprogs-1.29.

postgres database create times 4-8% faster.

kernel - ext3   Create-ui   Create   Create-emb
2.5.39          326.4       311.7    333.7     seconds
2.5.40-mm1      300.9       299.2    318.9     

dbench 64 on ext2 13% faster.

dbench 64 avg 5 runs
2.5.39 201.40 mb/sec
2.5.40-mm1 227.55

dbench 192       avg 5 runs
2.5.39           189.57 mb/sec
2.5.40-mm1       193.55

tbench 192 11% faster.

tbench 192
2.5.39 118.17
2.5.40-mm1 131.39

In the 120% memory pressure test there has been
a 66% improvement since 2.5.38:

qsbench    h:mm:ss   
2.5.38     1:57:48
2.5.39     1:30:49

2.5.40-mm1 1:10:40

All tests on quad xeon with 3.75 gb ram.

Signature · 2 lines
Randy Hron
http://home.earthlink.net/~rwhron/kernel/bigbox.html