Path: utzoo!utgpu!news-server.csri.toronto.edu!eecg.toronto.edu!leblanc
Newsgroups: comp.sys.amiga
From: leblanc@eecg.toronto.edu (Marcel LeBlanc)
Subject: Re: New GVP Controller questions
Message-ID: <1990Sep21.163715.3423@jarvis.csri.toronto.edu>
References: <9009201538.AA28646@ucbvax.Berkeley.EDU> <12291@ogicse.ogi.edu>
Date: 21 Sep 90 20:37:15 GMT
Lines: 23

jmeissen@ogicse.ogi.edu (John Meissen) writes:
>>5.  Is there any inherant advantage/disadvantage to having the memory
>>    expansion on the controller board, and do I lose any 2090A functionality 
>>    that I might need later?  Given GVPs reputation for quality
>>    design I would suspect not, but I would rather ask than be sorry later.

>Yes, the controller can DMA into the on-board memory without competing
>for the Zorro BUS (so I have been told). This makes transfers faster than
>if all the expansion memory was on another card.

I think the person you spoke to may not have been very clear.  Unless GVP
has completely redesigned their SCSI controller, it DOES NOT DMA to the
Amiga's memory (the 2090A does).  The GVP controller that I have writes to a
cache (hence their misleading 'DMA to on-board memory' claim), which must
then be copied to it's final destination by the CPU.  The GVP controller is
not in the same speed league as the 2090A controller.  Dave Haynie recently
posted an excellent explanation of the DMA and non-DMA techniques used by
popular HD controllers for the Amiga.  I can mail you a copy if you would
like.

Marcel A. LeBlanc  --  Electrical Eng. Computer Group, Univ. of Toronto
-----------------------------------------------------------------------
leblanc@eecg.toronto.edu		else: uunet!utcsri!eecg!leblanc