Actually, the 1800 factor gives a slightly conservative number, so 14 on a
box that is supposed to be capped at 18 MIPS is probably about right. If
you have the box I think you do, I would have said it is probably 16 MIPS,
so the 14 is "reasonably" close. When I developed the 1800 number we had
some good reasons to keep the MIPS estimates on the low side. I've been
running rexxcps on small systems from the early P/390s on and have a fairly
decent record of various emulated and non-emulated systems. Using rexxcps
only measures the cpu speed of a single processor, but for typical workloads
it does give a pretty good indication of the relative processor speeds. If
anyone is interested, the current IBM zPDT based systems provide between 150
and 200 rexxcps based MIPS per enabled processor on current IBM System X
servers, depending on the processor speed. As usual, MIPS is as meaningful
or meaningless as you want it to be.
Mike Hammock
--------------------------------------------------
From: "Fran Hensler" <f...@zvm.sru.edu>
Sent: Saturday, June 11, 2011 7:45 AM
To: <IBMVM@LISTSERV.UARK.EDU>
Subject: Re: REXXCPS EXEC
On Fri, 10 Jun 2011 10:25:05 -0400 Mike Hammock said:
I have found in the past that dividing the REXX CPS number by 1800 gives
an
approximation(!) of the general MIPS rating.
Mike -
I'm running a FLEX-ES box that is knee capped at 18 MIPS. I added
your formula to the end of REXXCPS. Here are the results on a
completely idle system:
----- REXXCPS 2.2 -- Measuring REXX clauses/second -----
REXX version is: REXX370 4.02 01 Dec 1998
System is: CMS
Averaging: 10 measures of 30 iterations
CP IND = AVGPROC-011% 01
Performance: 24794 REXX clauses per second
24794/1800= 14 MIPS
Ready; T=11.08/11.22 06:39:22
rexxcps 1000
----- REXXCPS 2.2 -- Measuring REXX clauses/second -----
REXX version is: REXX370 4.02 01 Dec 1998
System is: CMS
Averaging: 1000 measures of 30 iterations
CP IND = AVGPROC-100% 01
Performance: 23699 REXX clauses per second
23699/1800= 13 MIPS
Ready; T=1133.43/1143.71 07:38:38
I ran REXXCPS at 30 and 1000 three times and each time the 30
measures indicated 14 MIPS while 1000 indicated 13 MIPS.
At the end of August we will downgrade to an 8 MIP box. It will
be interesting to see what REXXCPS reports then.
/Fran Hensler at Slippery Rock University of Pennsylvania USA for 48 years
mailto:f...@zvm.sru.edu http://zvm.sru.edu/~fjh +1.724.794.6172
"Yes, Virginia, there is a Slippery Rock"
--------------------------------------------------------------------------