Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!cbatt!ihnp4!qantel!lll-lcc!pyramid!amiga!dale
From: dale@amiga.UUCP (Dale Luck)
Newsgroups: net.micro.amiga
Subject: Re: LIFE on a 68020 (Re: Tom)
Message-ID: <1573@amiga.amiga.UUCP>
Date: Sat, 27-Sep-86 18:41:45 EDT
Article-I.D.: amiga.1573
Posted: Sat Sep 27 18:41:45 1986
Date-Received: Tue, 30-Sep-86 07:31:20 EDT
References: <8609221813.AA04648@cory.Berkeley.EDU> <1137@hoptoad.uucp>
Reply-To: dale@tooter.UUCP (Dale Luck)
Organization: Commodore-Amiga Inc., 983 University Ave #D, Los Gatos CA 95030
Lines: 36

In article <1137@hoptoad.uucp> gnu@hoptoad.uucp (John Gilmore) writes:
>In article <8609221813.AA04648@cory.Berkeley.EDU>, dillon@CORY.BERKELEY.EDU (Matt Dillon) writes:
>> 	With its cache, a 68020 doing simple blits would use most of bandwidth
>> (100% for simple copies).  Since a 68020 usually has a faster clock than a
>> 68000, it would easily beat out the Amiga's blitter.   Additionaly, you
>> wouldn't need multiple blitter passes.
>> 	Of course, the disadvantage is that it uses the processor... the system
>> would basically stop while you're doing the blit.  But hell, you only live
>> once.
>
>It doesn't really matter whether the CPU is doing it or the blitter is
>doing it -- if ANYBODY is burning up 100% of the memory bandwidth, the
>system will "basically stop" -- or stop doing anything but the blit.
>
>I think the Amiga blitter can be configured to steal basically all the
>memory cycles (preempt the processor).  The CPU could also do a "wait"
>instruction to pause until the next interrupt, test if the blitter
>is done, and jump back to the wait if not.  While in a WAIT, the CPU
>does not use any bus bandwidth.  I don't know if the memory arbitration
>on the Amiga is smart enough to notice, though.
>-- 

Remember you guys and gals that the amiga has two seperate buses, the
blitter/chip bus and the 68k bus. While the blitter is running the 68k
is free to use all the bandwidth of the expansion bus/roms and the
kickstart memory. The blitter does not slow anybody down. You can be
getting ready for the next blit while the blitter is busy churning up
the bits from the previous blit. This fact is used extensively in the
graphics library.
The access time of non-agnus dma devices (like the 68k) is 4 clock ticks,
like the 68000. The blitter can get at the memory at twice that speed.

Hope this clears it up, or maybe makes the discusion even more complex.

Dale Luck
Commodore-Amiga, Inc.