Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 7 Feb 1996 22:14:14 +0100
From:      se@zpr.uni-koeln.de (Stefan Esser)
To:        Peter Dufault <dufault@hda.hda.com>
Cc:        hasty@rah.star-gate.com, fwmiller@cs.umd.edu, hackers@freebsd.org
Subject:   Re: PCI programming
Message-ID:  <199602072114.AA22888@Sysiphos>
In-Reply-To: Peter Dufault <dufault@hda.hda.com> "Re: PCI programming" (Feb  7,  9:00)

next in thread | previous in thread | raw e-mail | index | archive | help
On Feb 7,  9:00, Peter Dufault wrote:
} Subject: Re: PCI programming
} We forwarded this gentleman's private e-mail to the list.
} In general we ought to avoid this, and invite the originator to
} send to the list himself.
} 
} >  Both these devices are Bus Masters..
} > neither contains  accessible ram.
} > you can't really pass data directly from one master to another.
} > 
} > you need a RAM scratch pad that both devices can access.
} 
} A while back I set it up to DMA directly from a VME disk array
} controller directly out a high speed multidrop card to an HDTV
} frame buffer.  The disk array controller was set up to generate
} non-ascending addresses.
} 
} I suspect the NCR PCI SCSI controller can do the same sort of thing
} since the NCR 725 can be set up that way, and a network card with
} a PCI visible FIFO could be the DMA target.

The NCR 53c8xx can do SCSI bus to memory block moves.

But it can't do DMA port writes to some Ethernet chip,
since it actively generates only memory accesses (it 
can rrespond to memory mapped or port I/O accesses to
its registers).

But since bus-master PCI Ethernet chips have identical
DMA capabilities, you'll be able to have the NCR fill
a sector buffer in DRAM and the Ethernet chip send it 
as soon as it is complete.

Neither chip has FIFOs that could hold a maximum size
Ethernet packet, and the speeds aren't matched well 
enough to directly connect one chip's output with the
other chip's input ...


Regards, STefan
-- 
 Stefan Esser, Zentrum fuer Paralleles Rechnen		Tel:	+49 221 4706021
 Universitaet zu Koeln, Weyertal 80, 50931 Koeln	FAX:	+49 221 4705160
 ==============================================================================
 http://www.zpr.uni-koeln.de/~se			  <se@ZPR.Uni-Koeln.DE>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199602072114.AA22888>