Cause of high CPU consumption

Question

I created the following test setting:
CREATE TABLE test_p (
        id           NUMBER
      , rand         NUMBER
      , id_pad         VARCHAR2(100)
      , vc_pad          VARCHAR2(50)
) NOLOGGING;
INSERT /*+ APPEND */ INTO test_p (
      id
    , rand
    , id_pad
    , vc_pad
SELECT rownum
     , trunc(dbms_random.value(0,5000))
     , rpad(rownum, 100)
     , rpad('x', 50)
FROM dual
CONNECT BY LEVEL <= 10000000;
CREATE INDEX idx_p_p1 ON test_p (id);
ALTER TABLE test_p ADD CONSTRAINT pk_p PRIMARY KEY (id) USING INDEX idx_p_p1;Now I executed the following statement on several servers (10.2.0.4; most of the servers Solaris 10)
SELECT count(p.rand)
FROM test_p p;One server showed an execution time of 16sec with 13sec "CPU used by this session" and 10sec "db file scattered read". Several retests showed about the same pattern: high CPU usage though there was a low load on the system (24 CPUs and run queue of around 10). I ran the same test on other systems (with less CPUs if this does matter) - the percentage of CPU time / execution time was muuuuch lower on all other systems.
What can be the cause of the high percentage of CPU time?
- because the CPUs are slower on the system with 24CPUs???
- because there might be swaping / Paging???
- because context switches might be more expensive on systems with more CPUs???

Answer 1

Server       execution   CPU     scattered reads
                time    time     time    
Server A     16,2862     13,917     10,2995      
Server b     57,7479     8,4802     52,9839      
Server c     21,2614     3,4114     19,2308       
Server d     41,0842     17,6861     35,039      
Server e     23,6073     5,8746     15,2809      
Server f     46,2391     15,1703     39,9993