After switching to 4 CPUs server selects work 8x slower

[Przemyslaw Dargiel]

I've swiched from a 2xPII 800 with 1Gb RAM and WinNT 4.0 to a new
SunFire3800 with 4xUltraSpacIII 750MHz and 4GB Ram with Solaris 8.
After that a exemplary selecty wich executed it self for 5 sec. now
executes for 40 sec.
WinNT is a 24/7 server and SHOW PROCESSLIST shows at most 10
processes at a time. At this time the processors usage is at 100% for
each. The new machine is currently in test phase. Although there is
one application working on it but it can be ignored. SHOW PROCESSLIST
sometimes shows a single query. The same query on the same table
structures and with the same portion of data executes 8x slower on
Solaris 8 and the processor usage is at max 20%. I've tryed
changeing the settings in my.cnf according to
http://dev.mysql.com/doc/mysql/en/server-parameters.html but the
effects ware minimal - from 40 to 39 sec. at time to time. Finaly I've
used my-large.cnf from the distro catalog. I've checked the indexes in
the tables, they are the same. I did an analysis of the table and
found no changes. I've checked the local and remote querys, no
changes.
I've tried useing HIGH_PRIRITY select with no effect. What's strange
is that on solaris the processor usage is at max 20-25% and IDLE rest
of the time.

What wrong is with my configuration?
Thanks in advance.

Details:
Server 1:
2xPentium III 800Mhz
1Gb Ram
WinNT4.0
Mysql 4.0.15-nt (MYISAM Engine)
Apache with PHP and a oracle client installed

Server 2: SunFire 3800
4x UltraSparcIII 750Mhz
4Gb Ram
SunOS 5.8
Mysql 4.0.21-max-log (MYISAM Engine)
Apache with PHP and a oracle client installed

my.cnf from SunFire3800
[client]
port = 3306
socket = /tmp/mysql.sock
[mysqld]
port = 3306
socket = /tmp/mysql.sock
skip-locking
key_buffer = 256M
max_allowed_packet = 1M
table_cache = 256
sort_buffer_size = 1M
read_buffer_size = 1M
myisam_sort_buffer_size = 64M
thread_cache = 8
query_cache_size= 16M
# Try number of CPU's*2 for thread_concurrency
thread_concurrency = 8
default-character-set=latin2
log-bin
server-id = 1
[mysqldump]
quick
max_allowed_packet = 16M
[mysql]
no-auto-rehash
[isamchk]
key_buffer = 128M
sort_buffer_size = 128M
read_buffer = 2M
write_buffer = 2M
[myisamchk]
key_buffer = 128M
sort_buffer_size = 128M
read_buffer = 2M
write_buffer = 2M
[mysqlhotcopy]
interactive-timeout


The select I've used:(this select returns single records, meanly 3-5)

select o.id osoba_id, o.imie, o.nazwisko, gt.symbol gt_symbol, o.tel_s, o.kom_s,
count(distinct gzd2.id) liczba_zadan
from gt_zadanie gzd, gt_osoby_gt gog, gts_gt ggt,
osoby o, gt_user_terminarz gut, gts_cykl_pracy gcp, gts_gt gt
left join gts_kompetencja as gtsk on(gzd.rodzaj_sprzetu=gtsk.rodzaj_sprzetu and gzd.produkt=gtsk.produkt and gzd.typ=gtsk.typ_sprzetu and gzd.rodzaj_zlecenia=gtsk.rodzaj_zlecenia)
join gt_user_komp as guk on(gtsk.id=guk.komp and guk.user=gut.user)
left join miejscowosc as ma on(gzd.id_miejscowoscA=ma.id_miejscowosc and gzd.obszar=ma.id_obszar)
left join gmina as ga on(ma.id_gmina=ga.id_gmina and ma.id_obszar=ga.id_obszar)
join gt_user_powiat as gup on(gog.user=gup.user and gup.id_powiat=ga.id_powiat)
left join gt_przydzial as gprz2 on(gprz2.osoba=o.id)
left join gt_zadanie as gzd2 on(gprz2.zadanie=gzd2.id and gzd2.status=3)
where ((unix_timestamp(gut.data)+time_to_sec(gcp.start)<=unix_timestamp(gzd.termin_od)
and unix_timestamp(gut.data)+time_to_sec(gcp.start)+gcp.godziny*3600>=unix_timestamp(gzd.termin_od))
or
(unix_timestamp(gut.data)+time_to_sec(gcp.start)<=unix_timestamp(gzd.termin_do)
and unix_timestamp(gut.data)+time_to_sec(gcp.start)+gcp.godziny*3600>=unix_timestamp(gzd.termin_do))
or
(unix_timestamp(gut.data)+time_to_sec(gcp.start)>=unix_timestamp(gzd.termin_od)
and unix_timestamp(gut.data)+time_to_sec(gcp.start)+gcp.godziny*3600<=unix_timestamp(gzd.termin_do))
)
and gut.user=o.id
and gut.cykl=gcp.id
and ((gzd.typ_zlecenia=ggt.typ_zlecenia and gzd.typ_zlecenia!=4) or (gzd.typ_zlecenia=4))
and ggt.id=gog.gt
and gog.user=o.id
and gog.gt=gt.id
and gt.id_obszar=gzd.obszar
and gzd.id=11991
and gcp.dyzur=''
and o.id not in (2182,2184)
group by o.id
order by o.nazwisko


its explain is the same on both servers:

+-------+--------+----------------------------------------------------+---------
-------+---------+-----------------------+------+-------------------------------
--+
| table | type | possible_keys | key
| key_len | ref | rows | Extra
|
+-------+--------+----------------------------------------------------+---------
-------+---------+-----------------------+------+-------------------------------
--+
| gzd | const | PRIMARY,id,obszar,typ_zlecenia | PRIMARY
| 4 | const | 1 | Using temporary; Using filesor
t |
| gt | ref | PRIMARY,id_obszar | id_obsza
r | 5 | const | 12 | Using where
|
| gtsk | ref | rodzaj_sprzetu,typ_sprzetu,rodzaj_zlecenia,produkt | produkt
| 5 | const | 4 |
|
| guk | ref | user,komp | komp
| 5 | gtsk.id | 8 | Using where
|
| gut | ref | user,cykl | user
| 5 | guk.user | 106 | Using where
|
| gcp | eq_ref | PRIMARY,dyzur | PRIMARY
| 4 | gut.cykl | 1 | Using where
|
| o | eq_ref | PRIMARY,id | PRIMARY
| 4 | gut.user | 1 | Using where
|
| gog | ref | user,gt | user
| 5 | o.id | 1 | Using where
|
| ggt | eq_ref | PRIMARY,typ_zlecenia | PRIMARY
| 4 | gog.gt | 1 | Using where
|
| ma | ref | ID_OBSZAR,ID_MIEJSCOWOSC | ID_MIEJS
COWOSC | 5 | const | 2 |
|
| ga | ref | ID_GMINA | ID_GMINA
| 5 | ma.ID_GMINA | 2 |
|
| gup | ref | user_powiat,user,id_powiat | user_pow
iat | 10 | gog.user,ga.ID_POWIAT | 1 | Using where; Using index
|
| gprz2 | ref | osoba | osoba
| 5 | o.id | 112 |
|
| gzd2 | eq_ref | PRIMARY,id,status | PRIMARY
| 4 | gprz2.zadanie | 1 | Using where
|
+-------+--------+----------------------------------------------------+---------
-------+---------+-----------------------+------+-------------------------------
--+
14 rows in set (0.10 sec)