[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Disk images: from //e to Mac via PC?



m.kelsey@spam.eecs.wsu.edu wrote:

>supertimer@aol.com (Supertimer) writes:
>
>> m.kelsey@spam.eecs.wsu.edu wrote:
>> 
>>>supertimer@aol.com (Supertimer) writes:
>>>
>>>So I'm to believe the IIgs can sustain transfer rates of 230Kbit/sec?
>>>My stock //e (1Mhz) couldn't even pump out data at a full 19200 baud
>>>over null modem.  My stock IIgs can barely reach to 48Kbit out of
>>>56Kbit on upload.  I find it difficult to believe a stock IIgs can
>>>sustain 230Kbit unless an accelerator were in place!
>> 
>> In this case, CPU speed is not the limiting factor, but the serial
>> hardware.  The standard IIgs is 2.6-2.8Mhz, right?  Yet you can
>> do 48kbit...put an 8Mhz ZipChip in a IIe and it CAN'T go 19200
>> even though the CPU speed is faster than the GS.
>
>The Apricorn .... could pull 38400 with the //e, however, my //e
>could crank data out a 19200, but receive was substantially slower.

It could be that Proterm simply limits to 19200 on a IIe. Remember,
as David Empson said, the Disk II controller transfers at 250kbits
per second SERIAL and WITHOUT DMA.

The main limit to speed in most IIes is the Super Serial Card's
COM chip...the same chip used in most IIe serial cards...

>If I set the GS serial port to 19200 it sends and receives at
>nearly the peak rate.  This must be due to a major improvement
>in hardware.

Simply the fact being the GS serial chip is better.  Not only
that, its disk controller chip, the IWM, is better.  It can do
500kbits per second according to David Empson.

>> For the same reason, an AppleTalked GS can link at 230kbit.
>> When AppleTaking, the GS uses the hardware characteristics
>> of the serial chip and motherboard to transfer at these high
>> speeds.  Timing is critical, so many interrupts are shut off...this
>> messes up things like 3200 color mode.
>> 
>> Remember...the GS can pump MUCH faster than 230kbit
>> with an SCSI card...or even with a floppy controller (which is
>> serial).  The serial hardware is the limiting factor, not the CPU
>> speed (except when doing error checking stuff like Zmodem).
>
>That is a good point, I hadn't considered DMA as possibility for 
>onboard dedicated hardware.  From what I've benchmarked, my IIgs 
>doesn't pull more than 150K/sec from hard drives capable of 600K/sec
>on a faster PC.  While this value is far greater than 230Kbit, it
>is using DMA.  Knowing the IIgs can handle such transfer rates it
>opens some new light on the subject.

As David Empson said, running in ProDOS 8 is going to limit
the transfer rate.  Run in GS/OS with the Apple SCSI driver will
result in much faster transfers.  Apple wrote a demo that pulled
successive full screen 320x200 graphics frames from the hard
disk at 32 frames per second to show off the speed of the Apple
High Speed SCSI card during its intro...that's something like
980 kilobytes per second...close to the 1MB per second limit
of this card.

>Now this brings up another important point, I stumbled onto some
>TOPS cards which I have heard rumoured to be AppleTalk compliant.
>
>If this is the case, this would be a nifty method preferred over
>serial link.  The problem is, where to find TOPS software, the
>proper connectors, and if the dad-gum things are even AppleTalk
>complaint!

Like David Empson says, the IIgs uses nothing magical to get the
high transfer rate required by AppleTalk...it simply dedicates most
of its CPU time transferring data into and out of the serial chip...

It is just the hardware characteristics of the motherboard (to clarify,
this means that it has a serial chip on the motherboard capable of
this kind of transfer)...