Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!usc!cs.utexas.edu!uunet!van-bc!
From: lphillips@lpami.wimsey.bc.ca (Larry Phillips)
Newsgroups: comp.sys.amiga
Subject: Re: fastmemfirst
Message-ID: <684@lpami.wimsey.bc.ca>
Date: 1 Aug 89 11:45:19 GMT
Lines: 64
Return-Path: <lpami!lphillips>
To: van-bc!rnews

In <477@tardis.Tymnet.COM>, jms@tardis.Tymnet.COM (Joe Smith) writes:y
>In article <11353@mcdphx.phx.mcd.mot.com> stan@teroach.phx.mcd.mot.com (Stan Fisher) writes:
>>   When using the Super Agnes chip, therefore elimating the C00000 RAM,
>>it is no longer necessary to run fastmemfirst in the startup-sequence, in
>>order to conserve Chip RAM.  Was I dreaming this?   Was it the Super Agnes
> ^^^^^^^^^^^^^^^^^^^^^^^^^^
>          ^^^^^^^^ -- This assumption is false.

No it isn't. Your premises are wrong.

>>alone that changed this? or was it the use of SetCPU 1.5 too?  Seems to me
>>irregardless of Super/old Agnes, a program requesting "either type of RAM"
>>could be given chip, when fast is available.
>
>A plain A2000 with a 2 megs of expansion RAM has 3 types of memory:
>  512K of chip RAM, which is slow due to access by the custom chips.
>  512K of non-chip RAM at C00000, which is slow due to being on the chip bus.
> 2048K of expansion RAM, which is fast due to not being accessable by the 
> custom chips.

This is true, except that a 2000 with 2 megs of expansion ram and the 'obese
Agnus' has only two types of memory, CHIP and FAST. The chunk that was FAST
only by virtue of being on the CHIP bus, addressed at $C00000 is now pure CHIP,
and is no longer addressed at $C00000, but is instead readdressed to be
contiguous with the original CHIP memory.

>In a setup like that, the C00000 RAM would be near the head of the list
>of available memory because it is detected before the autoconfigure expansion
>RAM.  This is not nice, since requests for "fast" RAM would go to the slow
>non-chip memory instead of the fast non-chip memory.  FastMemFirst (also
>known as SlowMemLast) puts the slow-fast RAM at the end of the list where
>it is less likely to get used.
>
>FastMemFirst has no effect if you don't have any expansion memory.
>FastMemFirst has no effect if you don't have any memory at C00000 (an A500).

FastMemFirst has no effect on a 500 with the second 512K and an 'obese Agnus'
(Jay Miner likes to call it the 'Gordo Agnus'), since there is no $C00000,
HALF-FAST memory.

FastMemFirst has no effect on a 2000 with a Gordo Agnus.

>The request for "either type of RAM" tells the system to search the fast-RAM
>free list first, then search the chip-RAM free list only if the search of
>the fast-RAM list failed.  The searching of the fast-RAM list first is
>always done - regardless of whether FastMemFirst has been done or not.
>
>In summary: FastMemFirst is a kludge to make systems with the old Agnus and
>real expansion memory run faster.  It is not needed if you have then new
>1-meg Agnus.

In summary, the Gordo Agnus is a fully engineered _solution_ to the kludge of
having $C00000 memory. Hmm... so why are you disagreeing right up until the
last statement?

-larry

--
"So what the hell are we going to do with a Sun?" - Darlene Phillips -
+-----------------------------------------------------------------------+ 
|   //   Larry Phillips                                                 |
| \X/    lphillips@lpami.wimsey.bc.ca -or- uunet!van-bc!lpami!lphillips |
|        COMPUSERVE: 76703,4322  -or-  76703.4322@compuserve.com        |
+-----------------------------------------------------------------------+