Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!uunet!microsoft!brianw
From: brianw@microsoft.UUCP (Brian Willoughby)
Newsgroups: comp.sys.apple
Subject: Re: Speed options for the GS
Summary: 1 MHz Apple II Bus
Message-ID: <8081@microsoft.UUCP>
Date: 17 Oct 89 05:29:15 GMT
References: <8910152111.aa20779@SMOKE.BRL.MIL>
Reply-To: brianw@microsoft.UUCP (Brian Willoughby)
Organization: Microsoft Corp., Redmond WA
Lines: 80

In article <8910152111.aa20779@SMOKE.BRL.MIL> CHRIS@MSUS1.BITNET writes:
>
>> I'm not sure exactly what you're asking, but the TransWarp GS always
>> slows down to 1MHz when any slot device is addressed, just as the IIGS
>> itself does.  I haven't heard any plans to change this.
>
>I don't think your correct there.  I have a device that will only operate with
>the GS at normal speed.  If it always does that, why would me device work.
>
>Chris

A little clarification is needed.  The Apple Bus transfers data in 1 MHz
cycles.  Most Apple peripherals don't have their data ready any quicker,
and when you are writing data to a peripheral, they will certainly ignore
data that doesn't stick around for the full cycle.  Therefore, all Apple
II accelerators operate at 1 MHz FOR THE SINGLE INSTANT THAT THE DATA
TRANSFER CYCLE IS HAPPENING.  This is true for the TransWarp II Plus,
TransWarp GS and the 2.8 MHz standard GS.

Fine. However, some peripheral drivers execute in fast RAM, but were
written to execute on the 1 MHz Apple.  For this reason, all Apple II
compatible systems which operate above 1 MHz need a method of executing
this timed software at the proper speed, especially for direct CPU
controlled hardware like disk devices.

The IIgs designers implemented a defeatable feature where accessing the
'motor on' address causes the CPU to run at 1 MHz, and 'motor off'
resumes the speed selected in the control panel.  Unfortunately, this
doesn;t work if:
A) The device doesn't even use Disk ][ -style 'motor' control signals.
B) The device is in a slot other than 4,5,6 or 7.
C) The 'motor' sense for that slot is turned off (I don't have a IIgs,
so I don't know if this can happen under normal situations).

The TransWarp designers at AE have implemented a different solution (at
least for the II Plus version).  If ANY slot address is accessed (not
just the 'motor' switches), AND the slot is switched to slow on their
special slot speed DIP switch - then the card stays at 1 MHz for a
pre-determined number of cycles.  AE must have done a little research
om determining what the minimum time would be for the worst case, because
it is a long time even by the slow computing speed of humans (i.e. I have
enough time to type a Ctrl-G 'Bell' character three timess at the 1 MHz
speed, after accessing a card manually from the monitor, before the
TransWarp returns to 3.58 MHz).

This is a very confusing point, which even the Tech Support people at AE,
Laser and other Apple II product companies have a hard time
understanding.  The $Cxxx space is ALWAYS accessed with a 1 MHz clock
cycle.  The only choice allowed is whether to switch back to high speed
or not while executing the driver software.  On an interesting side note,
my UDC card (almost) works with the switch at full speed.  That's because
ALL of the UDC SmartPort driver software is in the $Cxxx space, so the
TransWarp has to slow down to read it.  Every access - both code and data
- are on 1 MHz cycles.  Except, of course, when the sector is copied
to/from fast RAM.  But the UDC buffers each sector in on-board RAM which
is in the $Cxxx space also, so the timing critical routines still execute
at 1 MHz.  The only thing the slow setting would do in this case is delay
the switch back the full speed AFTER the SmartPort driver is exited.  In
case you're wondering, I say it -> (almost) <- works because I can't get
the UDC to live with my TransWarp, so when it eventually fails in the
fast position I'm assuming that its for the same reason that it
eventually fails in the slow position (try explaining that to a Laser or
AE Tech Supportee - I tried with both and only accomplished a sizable
donation to MCI, but no help with the hardware).

You've heard that GS/OS generated drivers can be faster than the
ROM-based drivers?  I've just explained why.

P.S. I've received many doubting responses concerning the information in
my IIgs -related postings (since I'm a Plus owner) but if the curious
IIgs owner would simply purchase the A.W. tech refs, they might know as
much (perhaps more) about their machine as I do.  Hey, I don't work for
Addison Wesley, but since I'm researching into the purchase of a IIgs, I
have found that the manuals are the best way to find out about a machine.

Brian Willoughby
UUCP:           ...!{tikal, sun, uunet, elwood}!microsoft!brianw
InterNet:       microsoft!brianw@uunet.UU.NET
  or:           microsoft!brianw@Sun.COM
Bitnet          brianw@microsoft.UUCP