Re: how is system geared down? -- MIDRANGE-L

It's just my instinct telling me this but I suspect that the microcode
for the 2290 with Northstar is simply inserting the proper number of
No-Ops instructions to achieve the proper lack of speed.  It's very
likely that this is based upon some hardware componant so that swapping
of chips and such is extremely unlikely to have any affect (other than
the voiding of warranty!)

  The primary reason for the same Processor being used at different
performance levels is economy of scale-make 10000 of one thing instead
of 2000 of five different things. They will want the same for the bus,
the memory DIMMs, IOPs, IOAs, and other hardware componants so making
them run different speeds also compounds design effort, manufacturing
effort, spare parts, upgrades etc. 
  Similairly IBM would not want to have to support more different
microcode componants than they already do (one reason for V5R1 to be
last version on 400 and 500 models), so it's likely the microcode for
all Northstar processors is the same and there is some small hardware
thing in there telling that microcode how fast (or not) to run.  Isolate
that little bugger and you have found the 'Holy Grail' of iSeries
performance. (Much like Rusty Wallace often rips out his Rev Limiter
chip during the race. He just doesn't have to worry about voiding his
warranty!)

 - Larry

> Steve Richter wrote:
> 
> Based on the numbers cited in prev msgs:
>  "the model 170-2160 has a 100 Mhz Apache processor and a CPW rating
> of 114, while the model 170-2290 has a 200 Mhz Northstar processor and
> a CPW rating of only 73"
> it is obvious that the system is being geared down to meet the cpw
> targets.
> 
> Does anyone know how, where?
> 
> Is it slow memory, a slow bus or just a deliberately slowed clock?
> 
> Are the memory chips used in a power pc rs6000 usable in a power pc
> as400?
> 
> What about swapping memory chips between similar model as400s that
> have different cpw ratings? Any volunteers to do that?
> 
> Would the performance results of a power pc assembler pgm, patched in
> using sst, that uses register only operations ( no memory access ) to
> count to a trillion show the actual clock speed of the cpu?

-- 
Larry Bolhuis           | Cogito ergo mercari iSeries 
Arbor Solutions, Inc.   |
(616) 451-2500          |               (I think, therefore I buy iSeries.)
(616) 451-2571 -fax     |
lbolhuis@arbsol.com     |        #3  1951-2001
+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---