On Thu, Feb 3, 2011 at 10:59 AM, Micheal Butz <[email protected]>wrote:
> Edward Jaffes calculation about shifting bits seemed like something I would > do in Assembler > > Regardless if I subtracted a formated T > TOD which is 26 bytes and just look at the microseconds portion the last 6 > digits wouldn't that be accurate as well > The problem is a matter of obtaining sufficient number of digits to get an accurate answer. Please consider the case where start time where the last 6 digits of start time is 999999 and the last 6 digits of end time is 000001. A difference of just 2 microsecond as Tom pointed out. However, subtracting just the last 6 digits will result in a negative number and an inaccurate result. > > Sent from my iPhone > > > On Feb 3, 2011, at 1:42 PM, Tom Harper <[email protected]> wrote: > > Michael, >> >> Your comment doesn't seem relevant to me. Even if it takes two >> microseconds, it can span a time boundary which you are not including. More >> puzzling to me is why don't you want to get your calculation correct? >> >> Tom Harper >> >> -----Original Message----- >> From: IBM Mainframe Discussion List [mailto:[email protected]] On >> Behalf Of Micheal Butz >> Sent: Thursday, February 03, 2011 12:39 PM >> To: [email protected] >> Subject: Re: STCK vs TIMUSED >> >> If the code that I marking either Wall or CPU takes longer than 1 >> second there is big trouble >> >> Sent from my iPhone >> >> On Feb 3, 2011, at 1:21 PM, Sam Siegel <[email protected]> wrote: >> >> On Thu, Feb 3, 2011 at 10:11 AM, Micheal Butz <[email protected] >>> >>>> wrote: >>>> >>> >>> I am doing this in Rexx exec processing a SVC dump after calling >>>> BLSUXTOD. To convert the double word to a 26 character string I use >>>> the >>>> last 6 digits I understand I can bypass this conversion and just do >>>> the >>>> subtraction as you said on the double word regardless wouldn't the >>>> results >>>> be the same. ??? >>>> >>>> By using the right most digits there can be a situation where the >>>> end-time >>>> >>> appears to occur prior to the start-time. Consider the following: >>> >>> Start time: 1 999 998 >>> End time: 2 000 003 >>> >>> When all digits are considered, end-time is greater than start- >>> time. When >>> only 6 digits are considered, start-time appears to be greater than >>> end-time. >>> >>> To get an accurate time enough digits need to be used to ensure that >>> end-time is greater than start-time when doing the subtractions. >>> >>> Using all digits as Ed has suggested pretty much guarantees end-time >>> will be >>> great than start-time. >>> >>> >>> >>> Sent from my iPhone >>>> >>>> On Feb 3, 2011, at 11:57 AM, Edward Jaffe <[email protected] >>>> >>>>> >>>>> wrote: >>>> >>>> On 2/3/2011 8:41 AM, Micheal Butz wrote: >>>> >>>>> >>>>> I do a substr to get the last 6 digits of the date/time. The >>>>>> microseconds >>>>>> portion. And substract. Value = after - before >>>>>> >>>>>> >>>>> To subtract TOD values, use unsigned binary doubleword arithmetic. >>>>> If you >>>>> want resolution to microseconds, shift both doubleword values >>>>> right by 12 >>>>> bits before subtracting. >>>>> >>>>> -- >>>>> Edward E Jaffe >>>>> Phoenix Software International, Inc >>>>> 831 Parkview Drive North >>>>> El Segundo, CA 90245 >>>>> 310-338-0400 x318 >>>>> [email protected] >>>>> http://www.phoenixsoftware.com/ >>>>> >>>>> >> ---------------------------------------------------------------------- >> For IBM-MAIN subscribe / signoff / archive access instructions, >> send email to [email protected] with the message: GET IBM-MAIN INFO >> Search the archives at http://bama.ua.edu/archives/ibm-main.html >> > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: GET IBM-MAIN INFO > Search the archives at http://bama.ua.edu/archives/ibm-main.html > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
