At 01:32 PM 7/11/01 +1000, Justin Clift wrote:
> Standardised benchmarks would be nice to though.
Sure would be nice....
Here's some preliminary results of my testing...
System: RedHat Linux 7.1, kernel 2.4.3smp (i686)
Dual PIII 733 / 133FSB 1024MB RAM
2 18Gb 10000rpm U160 SCSI drives, hardware RAID1
added to /etc/sysctl.conf:
kernel.shmmax = 268435456
kernel.shmall = 268435456
PG 7.1.2
postgresql-7.1.2-4PGDG binary rpm set
relevant entries from postgresql.conf:
max_connections = 32
sort_mem = 4096
shared_buffers = 32000
fsync = true
The only benchmark I have is almost useless but I'll put it here anyway...
time ./pg_regress --schedule=parallel_schedule
(using postmaster on Unix socket, default port)
...
...
======================
All 76 tests passed.
======================
real 0m40.784s
user 0m1.980s
sys 0m1.220s
Adam.