Gerhard Postpischil wrote:
Edward Jaffe wrote:
An instruction timings book is not intended for use by management.
Agreed, but some managers consider any activity not directly related
to productive work (e.g., cranking out mindless code) to be suspect,
and grounds for discipline, unless of course it's their idea to check
into performance of a particular application.
In our organization, "management" developed a sophisticated measurement
program to compare code fragments for frequently-executed code paths. It
really works.The results are usually worth the trouble it takes to
construct the abstract test cases to be compared.
We also have a program, based on the above technology, that is supposed
to produce instruction timing tables. I recently discovered it has major
flaws because the person who developed it didn't understand the effects
of various pipeline interlocks. The results are "garbage" and not to be
trusted. :-( (Eventually, someone--perhaps I-- will clean it up and
make it useful again.)
The point is, having access to IBM's instruction timings document would
save us much time. And, the information provided would be guaranteed
accurate by the hardware manufacturer.
--
Edward E Jaffe
Phoenix Software International, Inc
5200 W Century Blvd, Suite 800
Los Angeles, CA 90045
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
