🚀 go-pugleaf

In article <830101240snz@phaedrus.demon.co.uk>, Neville Wilford
<jnw@phaedrus.demon.co.uk> writes

>In article <Yr8IhBBZHvdxEwEf@wookie.demon.co.uk>
>           wookie@wookie.demon.co.uk "John Beardmore" writes:
>
>>Among the complicating factors are:
>>
>> the need for 16 rather than 8 sim holders
>>
>> non standard motherbaord
>>
>> non standard metalwork ( AST )
>>
>> the machine belongs to a client; tactless to carve it up but OK to 
>> upgrade it.
>>
>>
>>Do you know of any modern motherbaords that will take 16 off 1 meg 
>>simms as well as possibly some 72s at the same time ?  If so, I'd
>>certainly think about it.
>
>I have a 386DX/25 relegated to a secondary role for the same reason.
>It has 16 x 30-pin SIMM sockets currently populated with 16 x 1MB
>SIMMS.  This seemed like an advantage when I bought the machine a few
>years ago.
>
>I did consider one of the "make-it-486" type upgrades, but not for very
>long.  Amongst the cards in the machine are an Adaptec 1542 bus-mastering
>SCSI HA and an Allied Telesis AT1500 ethernet card, also DMA bus
>mastering.  This is one of the relatively few '386 boards which can
>handle DMA transfers properly.

I don't know much about 386 boards in general but the AST boards I've
used have had no DMA problems as far as I know.


>I found a major reason not to upgrade in the Linux Ethernet-HOWTO
>discussing PIO vs shared memory vs DMA ethernet card types:
>
> "And a reason for not using ANY DMA device is using a Cyrix 486
>  processor designed for plug-in replacement of a 386: these
>  processors must flush their cache with each DMA cycle."
>
>Anyone know if this still applies to the newer generation of upgrade
>chips?

Well, the Make-It Cyrix 486oid I'm using with NT here seems OK. There is
a 1542 on the bus which I've used for a few odd jobs like loading OSs
and stuff. I havn't recognised any problems. Nothing else in the box
uses DMA.


Cheers, J/.
-- 
John Beardmore