From nobody Sat Oct 21 09:17:11 2000
X-From-Line: ddyer@bigfoot.com Fri Oct 20 19:06:21 2000
Received: from lsmls02.we.mediaone.net ([24.130.1.15])
	by junk.nocrew.org with esmtp (Exim 3.12 #1 (Debian))
	for lars@nocrew.org
	id 13mfca-0003tJ-00; Fri, 20 Oct 2000 19:06:20 +0200
Received: from turnip.nowhere (we-24-130-9-112.we.mediaone.net [24.130.9.112])
	by lsmls02.we.mediaone.net (8.8.7/8.8.7) with ESMTP id KAA23715
	for <lars@nocrew.org>; Fri, 20 Oct 2000 10:06:16 -0700 (PDT)
X-Gnus-Mail-Source: file:/var/spool/mail/lars
Message-Id: <5.0.0.25.1.20001020095730.02e50c40@pop.we.mediaone.net>
X-Sender: ddyer@pop.we.mediaone.net
X-Mailer: QUALCOMM Windows Eudora Version 5.0
Date: Fri, 20 Oct 2000 10:08:05 -0700
To: lars brinkhoff <lars@nocrew.org>
From: Dave Dyer <ddyer@bigfoot.com>
Subject: Re: Foonly processors
In-Reply-To: <85d7gvrkab.fsf@junk.nocrew.org>
References: <Dave Dyer's message of "Tue, 17 Oct 2000 10:12:23 -0700">
 <5.0.0.25.1.20001017100546.035b5080@pop.we.mediaone.net>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Lines: 44
Xref: junk.nocrew.org personal:4405


>
>> F-1 has essentially essentuially the same cache and memory
>> management as the KL.
>
>Did it have the same cache and page table size as well?

I believe so.   The page size was certainly the same.  The cache
was 4-way associative and 4k overall I think.

>Hmm, I used the past tense there, but maybe the F-1 still exists
>somewhere?

It could be in a garage somewhere.  I think Karol Brandt had
posession last.  You might also try Gary Demos for news.  I don't
have current email addresses for either of them.  YOu could try
Tom McMahon (tlm@microsoft.com), he used to be my source for
Foonly gossip.

>> The F-1 actually operated in KA-10 mode, because the cash to finish
>> development wasn't avaialable, and it was immediately pushed into
>> production work by the Triple-I movie group.  
>
>When did it start production work?

Sometimes near 1978.  It was used in a major way in the production
of TRON.


>> Likewise, F-1 had hardware support for the full complement of
>> floating point and extended instructions, but some were never
>> actually coded.
>
>Does that include the G-float and string instructions?

I'm not familiar with G-float, but I suspect the microcode engine
was a pretty good turing machine simulator, and there was plenty
of free microcode space.

I think your timeline is inverted here; the F-1 predated the 
extended string instructions, which were not present in the 
first KL models.  I definitely could be wrong about this.