Benchmarking different algorithms (was Beginning/End of week)

[Joe Boyle]

The results arn't really that surprising ( he said now that the results are
in ). I seem to recall that * and / calculations are achieved by simply
shifting register contents left and right; whereas char conversion is more
complex; it even states in the PH doc'n that character processing should be
kept to a minimum for performance reasons :)

Regards, Joe.

This e-mail and all information contained in it is confidential and may be
legally privileged. If you are not the intended recipient, your access to
this e-mail is unauthorized. Any use, dissemination, distribution,
publication or copying by you of this e-mail or any of the information
contained within it is prohibited and may be unlawful. Do not open any
attachments, delete it immediately from your system and notify the sender
promptly by e-mail that you have done so. The content of this 
e-mail and any attachments sent with it may have been altered without the
consent or knowledge of the author.

-----Original Message-----
From: powerh-l-admin@lists.sowder.com
[mailto:powerh-l-admin@lists.sowder.com] On Behalf Of
Jari.Kaljunen@teliasonera.com
Sent: 10 November 2004 16:09
To: powerh-l@lists.sowder.com
Subject: SV: Benchmarking different algorithms (was Beginning/End of week)

And on OpenVms:

1:
define d_day_1st date = floor(regdat / 100)*100 + 1

versus

2:
define d_day_1st date = nconv(asc(regdat,8)[1:6] + "01")

the winner is 1 with 10,5 seconds with 100000 records, alt 2. took 13,9
seconds.

/Jari


-----Ursprungligt meddelande-----
Från: powerh-l-admin@lists.sowder.com
[mailto:powerh-l-admin@lists.sowder.com]För Latimer, Richard
Skickat: den 10 november 2004 01:31
Till: David Morrison - Corporate; powerh-l@lists.sowder.com
Ämne: RE: Benchmarking different algorithms (was Beginning/End of week)


This has now got my curiosity going ... 

I have just conducted two back to back tests as follows:
Test 1:
set report limit 100000                 
access flight_data                      
DEFINE d_days_1st num*6 &               
   = DAYS(FLOOR(dte_mov / 100)*100) + 1 
set subfile name test1out keep          
rep summary idy_mov dte_mov d_days_1st  

Test 2:
set report limit 100000                       
access flight_data                                      
DEFINE d_days_1st num*6 &                               
   =days(nconvert(substring(cconvert(dte_mov,8),1,6) + "01"))
set subfile name test2out keep                          
rep summary idy_mov dte_mov d_days_1st              

Both were compiled and run sequentially in batch mode, no other users were
active at the time.
Test 1 - elapsed time 1 minute 57 seconds, 109 seconds processing unit time
reported in log
Test 2 - elapsed time 2 minutes 36 seconds, 151 seconds processing unit time
reported in log

Test done using 6.07E on AS/400 model 30S

Looks like a win for maths on the IBM ...

cheers
Richard

  > 


-- 
**********************************************************************
This electronic message together with any attachments is confidential. If
you receive it in error: (i) you must not use, disclose, copy or retain
it; (ii) please contact the sender immediately by reply email and then
delete the emails. Views expressed in this email may not be those of the
Airways Corporation of New Zealand Limited
**********************************************************************

= = = = = = = = = = = = = = = = = = = = = = = = = = = =
Mailing list: powerh-l@lists.sowder.com
Subscribe: "subscribe" in message body to powerh-l-request@lists.sowder.com
Unsubscribe: "unsubscribe <password>" in message body to
powerh-l-request@lists.sowder.com
http://lists.sowder.com/mailman/listinfo/powerh-l
This list is closed, thus to post to the list you must be a subscriber.