On Wed, 6 Mar 2019 at 21:08, Joseph Reichman <[email protected]> wrote:

> From what I remember there used to be a list of instruction clock speeds in
> the principle of operation I wonder if that's still available

The important issue is what you would do with such information were it
available. Are you trying to make your program run as fast as
possible? Looking at existing code for dodgy instructions or
sequences? Just historical curiosity...? (If the latter, there are old
Functional Characteristics books for most of the 1970s machines online
at bitsavers.org: http://bitsavers.org/pdf/ibm/360/funcChar/  or
http://bitsavers.org/pdf/ibm/370/funcChar .)

First, design your application logic sensibly.

Then, as has been discussed here a number of times, if you want the
best code possible for a given processor generation, use a high-level
language (typically C) and tell the compiler what machine you will be
running on. People with both great experience and access to IBM
information that you and I don't have design and configure the code
generation in the IBM compilers, and it's almost a sure bet that it'll
run faster than anything you or I can write.

You can certainly look at the generated code if you're interested in
what the compiler thinks is is optimal. But resist the temptation to
feed a tiny snippet of code to the compiler and then clone the
assembler result in your program. The compiler code generation can
take a larger view of your program and generate overall better code
than that resulting from a bunch of single statements glued together.

Tony H.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to