From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 21:49:08 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri, 27 Aug 2004 15:49:08 -0500 (CDT) Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> References: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> Message-ID: On Fri, 27 Aug 2004, Brad Parker wrote: > Many, many, many old ITS backup tapes are currently being read and after > some "cleansing" (removing personal info) will be available, I'm told, > on cd-rom. You can assume I'll be like a 4-year old kid camping on the > door step of the people who are doing that :-) Sorry to disappoint, but I have the fabled full dumps of the AI and MC KS10s, including "personal" data, and the LISPM source was not present in the dumps. I checked a year ago, at least. There is already a "free" ITS distribution out on the internet, but I forget where. From brad@heeltoe.com Fri Aug 27 22:06:34 2004 From: brad@heeltoe.com (Brad Parker) Date: Fri, 27 Aug 2004 17:06:34 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: Message from Daniel Seagraves of "Fri, 27 Aug 2004 15:49:08 CDT." Message-ID: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> Daniel Seagraves wrote: >On Fri, 27 Aug 2004, Brad Parker wrote: > >> Many, many, many old ITS backup tapes are currently being read and after >> some "cleansing" (removing personal info) will be available, I'm told, >> on cd-rom. You can assume I'll be like a 4-year old kid camping on the >> door step of the people who are doing that :-) > >Sorry to disappoint, but I have the fabled full dumps of the AI and MC >KS10s, including "personal" data, and the LISPM source was not present in >the dumps. I checked a year ago, at least. There is already a "free" ITS >distribution out on the internet, but I forget where. I was told (by the person who made the dumps originally) that there was something like a full pallet of tapes to be read. What I found on the net was, well, maybe 3-4 tapes worth. It looked more like a partial snap shot of what was on those machines at one point in time (i.e. one day). But not everything. I think these backup tapes span a large amount of time, like serveral years. Does that sounds more plausible? -brad From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 22:17:29 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri, 27 Aug 2004 16:17:29 -0500 (CDT) Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> References: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> Message-ID: On Fri, 27 Aug 2004, Brad Parker wrote: > I was told (by the person who made the dumps originally) that there was > something like a full pallet of tapes to be read. Possible. What I have are the hard-disk images of what was on the machines when they were turned off. > What I found on the net was, well, maybe 3-4 tapes worth. It looked > more like a partial snap shot of what was on those machines at one > point in time (i.e. one day). But not everything. I have the list of everything that was supposed to exist. I'll append it to the message. > I think these backup tapes span a large amount of time, like serveral years. GFR was a nightly (?) job, those tapes are probably GFR tapes. Maybe a year or two would cover a pallet. ** AI:LMFONT;-READ- -THIS- ** -*-Text-*- This directory is used by RG, MOON, BEE, ALAN, RMS, TK, H, MMCM, HIC, CENT, CWH, and JLK in connection with the Lisp Machine project. Relevant directories are: AI: LISPM; compiler, simulator, microcode, microassembler, etc. and nuclear system source and qfasl. LISPM1; cold loads and other miscellaneous files. LISPM2; overflow directory for LISPM ZWEI; Editor ZMAIL; Mail reading program LMCONS; console programs. LMDOC; documentation, etc., other than the manual itself. LMFILE; file computer programs LMFONT; Fonts (ast, kst, and qfasl files) LMIO; random I/O code LMIO1; randomer I/O code LMWIN; window system LMWIND; window system documentation LMLREC; LREC files. LMMAN; Lisp Machine manual. LMDEMO; Copies of qfasl files on MC:LMDEMO; LCADR; Software specific to CADR. (ucode, etc) LMPAT; Patch files for Lisp Machine minor system changes. LMPICT; Lisp Machine PICTure stuff LCHESS; Lisp Machine / Chess Machine chess program WOODS; The Woods Lunar rock program CONS; hardware (processor) CONS1; hardware (memory, backplane, etc.) CADRIO; hardware (random i/o devices, IOB) CADRDC; hardware (disk control) CONSWD; hardware (mem interface etc.) CADR; hardware (second) CADRWD; hardware (second machine wirelist) CADRPL; hardware (plot files) CADR1; hardware (second machine misc, bus interface, & c.) CADRM; hardware (memories) CADRMW; hardware (multiwire memory) CADRTV; hardware (tv controllers) CADRPT; hardware (part lists and purchase orders) CHAOS; hardware (chaosnet part of IOB) MC: LMFONT; Copies of fonts on AI:LMFONT; LMDEMO; Sources for demo files LMMAX; The primary MACSYMA directory for the LISPM version LMMAXQ; QFASL files for MACSYMA Relevant TAGS files are: LMDOC; LISPM TAGS - nuclear system LMMAN; .LMMAN TAGS - the manual ZWEI; ZWEI TAGS - new editor KWC; ENDGME TAGS - KWC's program LCHESS; CHESS TAGS - RG's program LMPICT; LT TAGS - vision program MC: MINIMA; MINIMA TAGS - Macsyma WOODS; WOODS TAGS - Lunar ** END ** From rjs@fdy2.demon.co.uk Fri Aug 27 23:20:01 2004 From: rjs@fdy2.demon.co.uk (Robert Swindells) Date: Fri, 27 Aug 2004 23:20:01 +0100 (BST) Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> (message from Brad Parker on Fri, 27 Aug 2004 17:28:25 -0400) Message-ID: <20040827222001.70D685A7@fdy2.demon.co.uk> Brad Parker wrote: >I know many people are interested in a new lisp-os or other forward >looking things. I am not (yet) interested in that. I'm still far too >interested in the old machines and running them. I love my lispms :-) My feeling is that starting from scratch is hard. I like playing with older stuff too. >I'm enjoying the notion of sucking the CADR netlist files into a >modern synthesis tool and using one of the cheap FPGA w/sdram eval >boards to fire up nuevo-cadr. There's certainly enough ram in most >modern parts to hold all the microcode. It's a sick idea but a fun >one... I would be amazed if you could do anything automatically. Creating a CPU in a HDL seems to be a fairly common student project at the moment, I only know of one microcoded one though. You use up the ram in a FPGA fairly quickly if your microcode is 64 bits wide, I have been doing the sums. Robert Swindells From rjs@fdy2.demon.co.uk Sat Aug 28 00:08:52 2004 From: rjs@fdy2.demon.co.uk (Robert Swindells) Date: Sat, 28 Aug 2004 00:08:52 +0100 (BST) Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272303.i7RN3hHx008299@mwave.heeltoe.com> (message from Brad Parker on Fri, 27 Aug 2004 19:03:42 -0400) Message-ID: <20040827230852.450275A7@fdy2.demon.co.uk> Brad Parker wrote: >Robert Swindells wrote: >> >>One other possible host would be Franz Lisp. >> >>It can be run on modern systems, but is probably easier to hack than >>MACLISP. >it's what I've been using on linux, trying to hack it back into a more >maclisp-ish style. It should be pretty close, particularly if you load lmhacks.l. What have you needed to change ? Robert From mikemac@mikemac.com Sat Aug 28 00:19:58 2004 From: mikemac@mikemac.com (Mike McDonald) Date: Fri, 27 Aug 2004 19:19:58 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: Your message of "Fri, 27 Aug 2004 17:28:25 EDT." <200408272128.i7RLSPvV007461@mwave.heeltoe.com> Message-ID: <200408272319.i7RNJwY29863@saturn.mikemac.com> >To: Daniel Seagraves >Date: Fri, 27 Aug 2004 17:28:25 -0400 >From: Brad Parker >I'm enjoying the notion of sucking the CADR netlist files into a modern >synthesis tool and using one of the cheap FPGA w/sdram eval boards to >fire up nuevo-cadr. There's certainly enough ram in most modern parts >to hold all the microcode. It's a sick idea but a fun one... If only ... that company that was directly responsible for the formation of the Free Software Foundation would let go of their stuff, you could build your own G-machine! if only ... Mike McDonald mikemac@mikemac.com Proud owner of an XL1201 From stevelisp@grape-krueger.com Tue Aug 31 02:14:45 2004 From: stevelisp@grape-krueger.com (Steve Krueger) Date: Mon, 30 Aug 2004 20:14:45 -0500 Subject: [LMH] Re: [LMH]bored? read pages 7,8,9 and explain In-Reply-To: <200408302038.i7UKcvlH013855@mwave.heeltoe.com> References: <200408302038.i7UKcvlH013855@mwave.heeltoe.com> Message-ID: <4133D105.5010903@sbcglobal.net> Brad Parker wrote: > Hi, > > If anyone (1) bored, (2) interested in CADR architecture and (3) > familiar with ldb and dpb, would you mind reading "A. I. memo 528" page > 7,8 & 9 > > and then talking with me about how the ldb and dpb parts of the byte > instruction should work? Hi Brad, I can probably answer any question you might have. I know how these instructions work and so find the description in AI Memo 528 to be completely clear (well mostly clear). I suggest you pose a question and I'll try to explain it. Maybe this programmers' summary will help: You use LDB to extract a bit field from a word, aligning it to the lsb end of the word. If the A-SRC is all zero, this produces a result that is zero extended. This is a very frequent operation. The rotation count is the bit number that the field begins at, and the length is the field width. |<- len ->|<--- rot ---->| +--------------------------------+ | | field | | M-SRC +--------------------------------+ |<- len ->| +--------------------------------+ |AAAAAAAAAAAAAAAAAAAAAA| field | RESULT +--------------------------------+ "A" are bits from the A-SRC that are not modified. Only the len bits at the right of the word are modified. DPB is the logical opposite of LDB, it takes an lsb-justified bit field and places it at the bit field described by the rotation and length. |<- len ->| +--------------------------------+ | | field | M-SRC +--------------------------------+ |<- len ->|<--- rot ---->| +--------------------------------+ |AAAAAAA| field |AAAAAAAAAAAAAA| RESULT +--------------------------------+ "A" represents bits from the A-SRC that are not modified. Selective deposite replaces the corresponding bits of the A-SRC with a bit field from the M-SRC described by its position and length. It does a DPB without actually rotating the M-SRC first. |<- len ->|<--- rot ---->| +--------------------------------+ | | field | | M-SRC +--------------------------------+ |<- len ->|<--- rot ---->| +--------------------------------+ |AAAAAAA| field |AAAAAAAAAAAAAA| RESULT +--------------------------------+ "A" represents bits of the A-SRC that are not modified. Note that the length is encoded in the microinstruction as length-1. You could read the Q-DATATYPE field of a Lisp word into an untyped (pure) number with: ((M-1) LDB M-T (BYTE 5. 25.) A-ZERO) This encodes ROT as 25 decimal (31 octal) and length-1 as 4. Note too that fields can be counted from the left instead of the right by doing a bit of arithmetic on the position and width. Does this help or are there more questions? BTW, I don't think I found the left mask and right mask description helpful when I read AI Memo 528 almost 25 years ago either. -Steve From nyef@softhome.net Fri Aug 27 22:53:30 2004 From: nyef@softhome.net (Nyef) Date: Fri, 27 Aug 2004 17:53:30 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> References: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> Message-ID: <20040827215330.GA8306@miyu.paradiesanalytics.com> On Fri, Aug 27, 2004 at 04:42:14PM -0400, Brad Parker wrote: > > (ps: i think if someone says lm-2 sources are unbuildable w/o a cadr I > don't think they have not tried very hard; but I may be wrong) Have a look at the function ASSEMBLE in sys/sys/cadrlp.lisp.134. Note the references to things in the SI and FS packages. Note the use of READ-METER. Those references won't build on a modern system. They might build on MACLISP for all I know, but they don't build on SBCL. Have a look at sys/sys2/comdefs.lisp.8, specifically the DEFUN for COMPILING-WHOLE-FILE-P. Note that it calls the stream as if it were a Flavors object. Is that going to work in MACLISP? Have a look at sys/sys/qcdefs.lisp.349, ADD-OPTIMIZER. I don't believe that MACLISP supports the use of "E in an arglist. I know that SBCL doesn't. I'm not saying that they're completely unbuildable, just that it'll take some hacking (possibly a lot of hacking) to get it working. --Alastair Bridgewater From stevelisp@grape-krueger.com Sat Aug 28 18:36:32 2004 From: stevelisp@grape-krueger.com (Steve Krueger) Date: Sat, 28 Aug 2004 12:36:32 -0500 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> References: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> Message-ID: <4130C2A0.3040405@sbcglobal.net> Brad Parker wrote:
I'm enjoying the notion of sucking the CADR netlist files into a modern
synthesis tool and using one of the cheap FPGA w/sdram eval boards to
fire up nuevo-cadr.  There's certainly enough ram in most modern parts
to hold all the microcode.  It's a sick idea but a fun one...
I'd be surprised if that would work at all.  The CADR has an interesting clocking scheme that is "mostly synchronous".  The main clock generation is by means of a pulse delay line.  The period is controlled by the type of the microinstruction, which is used to select a tap for recycling the pulse.  Some microinstructions (DISPATCH) run slower than others.  Furthermore, the machine goes asynchronous when performing a memory cycle.  The clock stops completely in the processor once a memory request has gone onto the bus.  If you follow the memory ACK through the schematics, it actually forms the pulse that goes back into the delay line to clock the next cycle.

All of that can be sorted out, but not be any automatic tool.

All in all, the CADR processor architecture is unlike modern designs in a lot of ways that are likely to result in poor automatic synthesis.

I think the more interesting project would be a CADR-compatible design in a modern style.  It would use more pipelining and maybe larger memory maps.  It would have to run CADR microcode unchanged.  It would need to have all of the compatible peripherals so that the microcode could access them.

This discussion reminds me, I had started to write a description of (some of the arcane aspects of) the CADR and how it was morphed into Explorer I and Explorer II.  I'll have to find that.  Oh, and _that_ reminds me that I promised to let the the ex-AI Lab and LMI guys I ran into this summer in Cambridge take a look at it for accurracy and expansion.  I guess I'd better get busy.

    -Steve


 From brad@heeltoe.com Sun Aug 29 22:14:20 2004 From: brad@heeltoe.com (Brad Parker) Date: Sun, 29 Aug 2004 17:14:20 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: Message from Steve Krueger of "Sat, 28 Aug 2004 12:36:32 CDT." <4130C2A0.3040405@sbcglobal.net> Message-ID: <200408292114.i7TLEKvs028301@mwave.heeltoe.com> [if you can, please don't send email in html] Steve Krueger wrote: .... >The CADR has an interesting clocking scheme that is >"mostly synchronous" ... Thanks - that's very interesting. (and I agree what what you are saying; still some of the logic might be nice to have, at least for a rev 0.0) >This discussion reminds me, I had started to write a description of >(some of the arcane aspects of) the CADR and how it was morphed into >Explorer I and Explorer II.  I'll have to find that.  Oh, and _that_ >reminds me that I promised to let the the ex-AI Lab and LMI guys I ran >into this summer in Cambridge take a look at it for accurracy and >expansion.  I guess I'd better get busy. such a document would be interesting. a quick glance looks like some of the microcode decode is very similar... -brad From brad@heeltoe.com Sat Aug 28 00:16:43 2004 From: brad@heeltoe.com (Brad Parker) Date: Fri, 27 Aug 2004 19:16:43 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: Message from Robert Swindells of "Sat, 28 Aug 2004 00:08:52 BST." <20040827230852.450275A7@fdy2.demon.co.uk> Message-ID: <200408272316.i7RNGhK2008482@mwave.heeltoe.com> Robert Swindells wrote: > >It should be pretty close, particularly if you load lmhacks.l. > >What have you needed to change ? lots of little things to "port" to linux ldb, dpb, byte, locf missing some things with arrays too, if I recall. I hacked the reader to, to do some odd escapes. I should diff it all and see. -brad From aek@spies.com Thu Aug 5 14:53:01 2004 From: aek@spies.com (Al Kossow) Date: Thu Aug 5 13:53:01 2004 Subject: [LMH]reinventing the LISPM list Message-ID: <20040805215244.1A1573C7A@spies.com> http://groups.yahoo.com/group/lispmachines Appears to have been created on Aug 3 Someone should let them know this list is still (semi) active. From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 11:58:01 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri Aug 27 10:58:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: <20040805215244.1A1573C7A@spies.com> References: <20040805215244.1A1573C7A@spies.com> Message-ID: Does anyone know if anyone out there in net-land has CADR software? From nyef@softhome.net Fri Aug 27 12:04:00 2004 From: nyef@softhome.net (Nyef) Date: Fri Aug 27 11:04:00 2004 Subject: [LMH]CADR software/images? In-Reply-To: References: <20040805215244.1A1573C7A@spies.com> Message-ID: <20040827184926.GA7996@miyu.paradiesanalytics.com> On Fri, Aug 27, 2004 at 01:58:25PM -0500, Daniel Seagraves wrote: > Does anyone know if anyone out there in net-land has CADR software? It is my understanding that a few people have copies of an LM-2 source tape image, but that the source is unbuildable due to there not being a working LM-2 or CADR system to build them from. You're thinking of building a CADR microcode emulator, aren't you? --Alastair Bridgewater From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 12:46:01 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri Aug 27 11:46:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: <20040827184926.GA7996@miyu.paradiesanalytics.com> References: <20040805215244.1A1573C7A@spies.com> <20040827184926.GA7996@miyu.paradiesanalytics.com> Message-ID: On Fri, 27 Aug 2004, Nyef wrote: > It is my understanding that a few people have copies of an LM-2 > source tape image, but that the source is unbuildable due to there > not being a working LM-2 or CADR system to build them from. > > You're thinking of building a CADR microcode emulator, aren't you? Considering it, but it's kinda pointless if the software doesn't exist. Let me ask a different question - What machines DO have software obtainable? So far just the Explorer 1? From nyef@softhome.net Fri Aug 27 13:25:01 2004 From: nyef@softhome.net (Nyef) Date: Fri Aug 27 12:25:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: References: <20040805215244.1A1573C7A@spies.com> <20040827184926.GA7996@miyu.paradiesanalytics.com> Message-ID: <20040827201022.GA8108@miyu.paradiesanalytics.com> On Fri, Aug 27, 2004 at 02:45:59PM -0500, Daniel Seagraves wrote: > On Fri, 27 Aug 2004, Nyef wrote: > > > It is my understanding that a few people have copies of an LM-2 > > source tape image, but that the source is unbuildable due to there > > not being a working LM-2 or CADR system to build them from. > > > > You're thinking of building a CADR microcode emulator, aren't you? > > Considering it, but it's kinda pointless if the software doesn't exist. > > Let me ask a different question - What machines DO have software > obtainable? So far just the Explorer 1? We have supposedly-bootable images for the Exp1 and the uExp. We have partial source sets for the entire Exp series (looks like the same code supports all three systems, but is missing the ucode, uassembler, and cold load tools). I'm not familiar with the current situation in regards to the Symbolics 3600 and later machines, but given that you can still buy the systems used it should be possible to get a bootable disk image. And there's this LM-2 source tape, which is apparently complete but unbuildable. Now, Al Kossow has a CADR, but the disks are too old to spin up, so that's not where we get an image from. We basically have three chances to get a working CADR image: First, someone finds a dumpable tape or disk with the compiled system on it. Second, someone manages to compile the LM-2 sources from MACLISP running on ITS in an emulator (may not work, the code might have drifted too far from being able to cross-build that way). Third, it might be possible to hack up the compiler and build tools to run on a modern lisp system, use that to create fasls and then a full drive image. --Alastair Bridgewater. From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 13:37:01 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri Aug 27 12:37:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: <20040827201022.GA8108@miyu.paradiesanalytics.com> References: <20040805215244.1A1573C7A@spies.com> <20040827184926.GA7996@miyu.paradiesanalytics.com> <20040827201022.GA8108@miyu.paradiesanalytics.com> Message-ID: On Fri, 27 Aug 2004, Nyef wrote: > We have supposedly-bootable images for the Exp1 and the uExp. We have > partial source sets for the entire Exp series (looks like the same code > supports all three systems, but is missing the ucode, uassembler, and > cold load tools). I'm not familiar with the current situation in regards > to the Symbolics 3600 and later machines, but given that you can still > buy the systems used it should be possible to get a bootable disk image. > And there's this LM-2 source tape, which is apparently complete but > unbuildable. From what I understand, emulating Symbolics machines is off-limits as Symbolics still exists and still does business. > Now, Al Kossow has a CADR, but the disks are too old to spin up, so > that's not where we get an image from. We basically have three chances > to get a working CADR image: First, someone finds a dumpable tape or > disk with the compiled system on it. Second, someone manages to compile > the LM-2 sources from MACLISP running on ITS in an emulator (may not > work, the code might have drifted too far from being able to cross-build > that way). Third, it might be possible to hack up the compiler and build > tools to run on a modern lisp system, use that to create fasls and then > a full drive image. I have a KS10 but I don't have any disks, so that probably doesn't help. I had two RM03s but I crashed them because I am a dumbass. (Moved them from New York to Illinois without setting the head locks, then crashed all my disk packs figuring out how I'd screwed up, which included scratching a real ITS disk pack. I felt like shooting myself afterward.) What's the current problem with E1 emulation? From brad@heeltoe.com Fri Aug 27 13:43:01 2004 From: brad@heeltoe.com (Brad Parker) Date: Fri Aug 27 12:43:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: Message from Daniel Seagraves of "Fri, 27 Aug 2004 14:45:59 CDT." Message-ID: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> Daniel Seagraves wrote: >On Fri, 27 Aug 2004, Nyef wrote: > >> It is my understanding that a few people have copies of an LM-2 >> source tape image, but that the source is unbuildable due to there >> not being a working LM-2 or CADR system to build them from. >> >> You're thinking of building a CADR microcode emulator, aren't you? Many of us want to build a CADR microcode emulator. And some continue to work on getting what we need to do that... (ok, so, *where* is the stuff? :-) Many, many, many old ITS backup tapes are currently being read and after some "cleansing" (removing personal info) will be available, I'm told, on cd-rom. You can assume I'll be like a 4-year old kid camping on the door step of the people who are doing that :-) It's quite likely that all the CADR software will be found there. There are also some old tapes in my area (near Cambridge, MA) which claim to maybe possibly have CADR sources on them. I'm in the process coaxing people to let me read them. I have a working 9-track with a vax on ethernet. There's also a few CADRs around, in various states, and some disk drives which might still work, but the tapes seem to be a more viable option since people seem (rightly so) afraid to load 30 year old disk packs. It may be possible to do some work before that by using the CADR schematics and a dump of CADR boot roms. I think those are available now. The holy grail would be a CADR microcode and world band. I've yet to get a hint of one of those on a tape, but who knows. I've harassed many people about this issue and so far everyone has been very kind and very helpful. I've got a lot of leads but no pay dirt yet. Still, I think I'm close. At some point I'll organize my notes into some sort of web page. Some of the stories are hilarious and some are downright scarey. While waiting, I've been working on a emulator for the Symbolics l-machines, since I have a few and I also world files. I've gotten about 35,000 instructions in the the boot so far. I'm hopeful it will some day (heh, sure) boot Genera. We'll see. If nothing more it's been fun to learn the 3600 architecture. In my day job I've been working on some very low level JVM code and binary recompilation, so I have some thoughts ways to emulate stack machines. (and I run qemu, which is a fine example of what can be done). Emulating a CADR is nicer because it would be unencumbered. The MIT folks are happy to release it's sources. Symbolics is not (yet) interested in releasing even old CADR work, but I'll continue to ask them and pray for a miracle. (ps: i think if someone says lm-2 sources are unbuildable w/o a cadr I don't think they have not tried very hard; but I may be wrong) -brad From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 13:49:00 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri Aug 27 12:49:00 2004 Subject: [LMH]CADR software/images? In-Reply-To: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> References: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> Message-ID: On Fri, 27 Aug 2004, Brad Parker wrote: > Many, many, many old ITS backup tapes are currently being read and after > some "cleansing" (removing personal info) will be available, I'm told, > on cd-rom. You can assume I'll be like a 4-year old kid camping on the > door step of the people who are doing that :-) Sorry to disappoint, but I have the fabled full dumps of the AI and MC KS10s, including "personal" data, and the LISPM source was not present in the dumps. I checked a year ago, at least. There is already a "free" ITS distribution out on the internet, but I forget where. From brad@heeltoe.com Fri Aug 27 14:08:01 2004 From: brad@heeltoe.com (Brad Parker) Date: Fri Aug 27 13:08:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: Message from Daniel Seagraves of "Fri, 27 Aug 2004 15:49:08 CDT." Message-ID: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> Daniel Seagraves wrote: >On Fri, 27 Aug 2004, Brad Parker wrote: > >> Many, many, many old ITS backup tapes are currently being read and after >> some "cleansing" (removing personal info) will be available, I'm told, >> on cd-rom. You can assume I'll be like a 4-year old kid camping on the >> door step of the people who are doing that :-) > >Sorry to disappoint, but I have the fabled full dumps of the AI and MC >KS10s, including "personal" data, and the LISPM source was not present in >the dumps. I checked a year ago, at least. There is already a "free" ITS >distribution out on the internet, but I forget where. I was told (by the person who made the dumps originally) that there was something like a full pallet of tapes to be read. What I found on the net was, well, maybe 3-4 tapes worth. It looked more like a partial snap shot of what was on those machines at one point in time (i.e. one day). But not everything. I think these backup tapes span a large amount of time, like serveral years. Does that sounds more plausible? -brad From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 14:17:00 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri Aug 27 13:17:00 2004 Subject: [LMH]CADR software/images? In-Reply-To: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> References: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> Message-ID: On Fri, 27 Aug 2004, Brad Parker wrote: > I was told (by the person who made the dumps originally) that there was > something like a full pallet of tapes to be read. Possible. What I have are the hard-disk images of what was on the machines when they were turned off. > What I found on the net was, well, maybe 3-4 tapes worth. It looked > more like a partial snap shot of what was on those machines at one > point in time (i.e. one day). But not everything. I have the list of everything that was supposed to exist. I'll append it to the message. > I think these backup tapes span a large amount of time, like serveral years. GFR was a nightly (?) job, those tapes are probably GFR tapes. Maybe a year or two would cover a pallet. ** AI:LMFONT;-READ- -THIS- ** -*-Text-*- This directory is used by RG, MOON, BEE, ALAN, RMS, TK, H, MMCM, HIC, CENT, CWH, and JLK in connection with the Lisp Machine project. Relevant directories are: AI: LISPM; compiler, simulator, microcode, microassembler, etc. and nuclear system source and qfasl. LISPM1; cold loads and other miscellaneous files. LISPM2; overflow directory for LISPM ZWEI; Editor ZMAIL; Mail reading program LMCONS; console programs. LMDOC; documentation, etc., other than the manual itself. LMFILE; file computer programs LMFONT; Fonts (ast, kst, and qfasl files) LMIO; random I/O code LMIO1; randomer I/O code LMWIN; window system LMWIND; window system documentation LMLREC; LREC files. LMMAN; Lisp Machine manual. LMDEMO; Copies of qfasl files on MC:LMDEMO; LCADR; Software specific to CADR. (ucode, etc) LMPAT; Patch files for Lisp Machine minor system changes. LMPICT; Lisp Machine PICTure stuff LCHESS; Lisp Machine / Chess Machine chess program WOODS; The Woods Lunar rock program CONS; hardware (processor) CONS1; hardware (memory, backplane, etc.) CADRIO; hardware (random i/o devices, IOB) CADRDC; hardware (disk control) CONSWD; hardware (mem interface etc.) CADR; hardware (second) CADRWD; hardware (second machine wirelist) CADRPL; hardware (plot files) CADR1; hardware (second machine misc, bus interface, & c.) CADRM; hardware (memories) CADRMW; hardware (multiwire memory) CADRTV; hardware (tv controllers) CADRPT; hardware (part lists and purchase orders) CHAOS; hardware (chaosnet part of IOB) MC: LMFONT; Copies of fonts on AI:LMFONT; LMDEMO; Sources for demo files LMMAX; The primary MACSYMA directory for the LISPM version LMMAXQ; QFASL files for MACSYMA Relevant TAGS files are: LMDOC; LISPM TAGS - nuclear system LMMAN; .LMMAN TAGS - the manual ZWEI; ZWEI TAGS - new editor KWC; ENDGME TAGS - KWC's program LCHESS; CHESS TAGS - RG's program LMPICT; LT TAGS - vision program MC: MINIMA; MINIMA TAGS - Macsyma WOODS; WOODS TAGS - Lunar ** END ** From brad@heeltoe.com Fri Aug 27 14:29:01 2004 From: brad@heeltoe.com (Brad Parker) Date: Fri Aug 27 13:29:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: Message from Daniel Seagraves of "Fri, 27 Aug 2004 16:17:29 CDT." Message-ID: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> Daniel Seagraves wrote: > >I have the list of everything that was supposed to exist. I'll append it >to the message. That's a beautiful list. But the snapshot I found on the internet (as you also discovered) only had a few of those directories. Certainly no lispm or lispm* directories. And no cons* or cadr* dirs. I would have wet my pants if I had found those ;-) I'm hoping for the tapes... Which reminds me, time to ping some folks :-) I know many people are interested in a new lisp-os or other forward looking things. I am not (yet) interested in that. I'm still far too interested in the old machines and running them. I love my lispms :-) I'm enjoying the notion of sucking the CADR netlist files into a modern synthesis tool and using one of the cheap FPGA w/sdram eval boards to fire up nuevo-cadr. There's certainly enough ram in most modern parts to hold all the microcode. It's a sick idea but a fun one... -brad From rjs@fdy2.demon.co.uk Fri Aug 27 14:39:00 2004 From: rjs@fdy2.demon.co.uk (Robert Swindells) Date: Fri Aug 27 13:39:00 2004 Subject: [LMH]CADR software/images? In-Reply-To: (message from Daniel Seagraves on Fri, 27 Aug 2004 15:37:22 -0500 (CDT)) Message-ID: <20040827213727.B02B5574@fdy2.demon.co.uk> Daniel Seagraves wrote >On Fri, 27 Aug 2004, Nyef wrote: >> We have supposedly-bootable images for the Exp1 and the uExp. We have >> partial source sets for the entire Exp series (looks like the same code >> supports all three systems, but is missing the ucode, uassembler, and >> cold load tools). I'm not familiar with the current situation in regards >> to the Symbolics 3600 and later machines, but given that you can still >> buy the systems used it should be possible to get a bootable disk image. >> And there's this LM-2 source tape, which is apparently complete but >> unbuildable. There is also the LMI source tree. >From what I understand, emulating Symbolics machines is off-limits as >Symbolics still exists and still does business. I made a suggestion a while back that someone or small group of people could make a business proposal to new Symbolics to port OpenGenera to AMD64 and/or G5 in exchange for X number of licences to use the result. >> Now, Al Kossow has a CADR, but the disks are too old to spin up, so >> that's not where we get an image from. We basically have three chances >> to get a working CADR image: First, someone finds a dumpable tape or >> disk with the compiled system on it. Second, someone manages to compile >> the LM-2 sources from MACLISP running on ITS in an emulator (may not >> work, the code might have drifted too far from being able to cross-build >> that way). Third, it might be possible to hack up the compiler and build >> tools to run on a modern lisp system, use that to create fasls and then >> a full drive image. >I have a KS10 but I don't have any disks, so that probably doesn't help. >I had two RM03s but I crashed them because I am a dumbass. (Moved them >from New York to Illinois without setting the head locks, then crashed >all my disk packs figuring out how I'd screwed up, which included >scratching a real ITS disk pack. I felt like shooting myself afterward.) I have booted ITS on the simh emulator, lack of hardware shouldn't be the sticking point. >What's the current problem with E1 emulation? Which emulator ? Exploiter is stuck at trying to work out what one of the options to switch a stack group needs to do. Nevermore is probably only suffering from too much real life. I have been a bit too busy with learning how to program FPGAs and with some family stuff to do anything for the last few months. Robert Swindells From brad@heeltoe.com Fri Aug 27 14:51:01 2004 From: brad@heeltoe.com (Brad Parker) Date: Fri Aug 27 13:51:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: Message from Robert Swindells of "Fri, 27 Aug 2004 22:37:27 BST." <20040827213727.B02B5574@fdy2.demon.co.uk> Message-ID: <200408272149.i7RLnimT007782@mwave.heeltoe.com> Robert Swindells wrote: > >There is also the LMI source tree. Yup - I have some tapes to read. More are welcome if you have them! >I made a suggestion a while back that someone or small group of people >could make a business proposal to new Symbolics to port OpenGenera to >AMD64 and/or G5 in exchange for X number of licences to use the result. An excellent suggestion. I've made it also. This may not make sense, but I would say the business model currently being persued by the person who owns Symbolics does not allow for any such entrepreneurial projects. I am hoping this may change at some point, but it's hard to say if it will. I find it sad, as the world would benefit from more exposure to Genera. -brad From nyef@softhome.net Fri Aug 27 14:53:00 2004 From: nyef@softhome.net (Nyef) Date: Fri Aug 27 13:53:00 2004 Subject: [LMH]CADR software/images? In-Reply-To: <20040827213727.B02B5574@fdy2.demon.co.uk> References: <20040827213727.B02B5574@fdy2.demon.co.uk> Message-ID: <20040827213809.GA8257@miyu.paradiesanalytics.com> On Fri, Aug 27, 2004 at 10:37:27PM +0100, Robert Swindells wrote: > > Daniel Seagraves wrote > > >What's the current problem with E1 emulation? > > Which emulator ? > > Exploiter is stuck at trying to work out what one of the options to > switch a stack group needs to do. > > Nevermore is probably only suffering from too much real life. Nevermore is stuck at trying to figure out why one of the standalone CPU diagnostic tests is failing. And not too much real life, competition from other projects. --Alastair Bridgewater From nyef@softhome.net Fri Aug 27 15:08:01 2004 From: nyef@softhome.net (Nyef) Date: Fri Aug 27 14:08:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> References: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> Message-ID: <20040827215330.GA8306@miyu.paradiesanalytics.com> On Fri, Aug 27, 2004 at 04:42:14PM -0400, Brad Parker wrote: > > (ps: i think if someone says lm-2 sources are unbuildable w/o a cadr I > don't think they have not tried very hard; but I may be wrong) Have a look at the function ASSEMBLE in sys/sys/cadrlp.lisp.134. Note the references to things in the SI and FS packages. Note the use of READ-METER. Those references won't build on a modern system. They might build on MACLISP for all I know, but they don't build on SBCL. Have a look at sys/sys2/comdefs.lisp.8, specifically the DEFUN for COMPILING-WHOLE-FILE-P. Note that it calls the stream as if it were a Flavors object. Is that going to work in MACLISP? Have a look at sys/sys/qcdefs.lisp.349, ADD-OPTIMIZER. I don't believe that MACLISP supports the use of "E in an arglist. I know that SBCL doesn't. I'm not saying that they're completely unbuildable, just that it'll take some hacking (possibly a lot of hacking) to get it working. --Alastair Bridgewater From aek@spies.com Fri Aug 27 15:20:01 2004 From: aek@spies.com (Al Kossow) Date: Fri Aug 27 14:20:01 2004 Subject: [LMH]Re: LispM-Hackers digest, Vol 1 #282 - 11 msgs In-Reply-To: <20040827220901.13148.qmail@kappa.unlambda.com> Message-ID: <20040827221949.6D5943C74@spies.com> I was told (by the person who made the dumps originally) that there was something like a full pallet of tapes to be read. -- Every tape AI still had is being read (roughly 15 THOUSAND) It will be over a terabyte of data. The pallets of tapes filled a 65' tractor trailer. I'm hopeful the LISPM dirs will turn up on GFR tapes from the early 80s. I dumped the boot proms from my CADR a few weeks ago. Problem is the prom count doesn't match what I have on the schematics. I'll have to look again to see if I may have missed a few. From rjs@fdy2.demon.co.uk Fri Aug 27 15:22:01 2004 From: rjs@fdy2.demon.co.uk (Robert Swindells) Date: Fri Aug 27 14:22:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> (message from Brad Parker on Fri, 27 Aug 2004 17:28:25 -0400) Message-ID: <20040827222001.70D685A7@fdy2.demon.co.uk> Brad Parker wrote: >I know many people are interested in a new lisp-os or other forward >looking things. I am not (yet) interested in that. I'm still far too >interested in the old machines and running them. I love my lispms :-) My feeling is that starting from scratch is hard. I like playing with older stuff too. >I'm enjoying the notion of sucking the CADR netlist files into a >modern synthesis tool and using one of the cheap FPGA w/sdram eval >boards to fire up nuevo-cadr. There's certainly enough ram in most >modern parts to hold all the microcode. It's a sick idea but a fun >one... I would be amazed if you could do anything automatically. Creating a CPU in a HDL seems to be a fairly common student project at the moment, I only know of one microcoded one though. You use up the ram in a FPGA fairly quickly if your microcode is 64 bits wide, I have been doing the sums. Robert Swindells From rjs@fdy2.demon.co.uk Fri Aug 27 15:41:01 2004 From: rjs@fdy2.demon.co.uk (Robert Swindells) Date: Fri Aug 27 14:41:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: <20040827215330.GA8306@miyu.paradiesanalytics.com> (message from Nyef on Fri, 27 Aug 2004 17:53:30 -0400) Message-ID: <20040827223927.457455A7@fdy2.demon.co.uk> Alastair Bridgewater wrote: >On Fri, Aug 27, 2004 at 04:42:14PM -0400, Brad Parker wrote: >> >> (ps: i think if someone says lm-2 sources are unbuildable w/o a cadr I >> don't think they have not tried very hard; but I may be wrong) [snip] >I'm not saying that they're completely unbuildable, just that it'll >take some hacking (possibly a lot of hacking) to get it working. One other possible host would be Franz Lisp. It can be run on modern systems, but is probably easier to hack than MACLISP. Robert Swindells From aek@spies.com Fri Aug 27 16:13:00 2004 From: aek@spies.com (Al Kossow) Date: Fri Aug 27 15:13:00 2004 Subject: [LMH]cadr boot proms Message-ID: <20040827231205.5CFB73C27@spies.com> www.bitsavers.org/bits/MIT/cadr There are 6 512x8's, so that may be the right number. Unfortunately, I don't know the order in the microword that the bits go in (hopefully they didn't scramble them within a prom) From mikemac@mikemac.com Fri Aug 27 16:18:01 2004 From: mikemac@mikemac.com (Mike McDonald) Date: Fri Aug 27 15:18:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: Your message of "Fri, 27 Aug 2004 17:28:25 EDT." <200408272128.i7RLSPvV007461@mwave.heeltoe.com> Message-ID: <200408272319.i7RNJwY29863@saturn.mikemac.com> >To: Daniel Seagraves >Date: Fri, 27 Aug 2004 17:28:25 -0400 >From: Brad Parker >I'm enjoying the notion of sucking the CADR netlist files into a modern >synthesis tool and using one of the cheap FPGA w/sdram eval boards to >fire up nuevo-cadr. There's certainly enough ram in most modern parts >to hold all the microcode. It's a sick idea but a fun one... If only ... that company that was directly responsible for the formation of the Free Software Foundation would let go of their stuff, you could build your own G-machine! if only ... Mike McDonald mikemac@mikemac.com Proud owner of an XL1201 From stevelisp@grape-krueger.com Sat Aug 28 10:37:00 2004 From: stevelisp@grape-krueger.com (Steve Krueger) Date: Sat Aug 28 09:37:00 2004 Subject: [LMH]CADR software/images? In-Reply-To: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> References: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> Message-ID: <4130C2A0.3040405@sbcglobal.net> Brad Parker wrote:
I'm enjoying the notion of sucking the CADR netlist files into a modern
synthesis tool and using one of the cheap FPGA w/sdram eval boards to
fire up nuevo-cadr.  There's certainly enough ram in most modern parts
to hold all the microcode.  It's a sick idea but a fun one...
I'd be surprised if that would work at all.  The CADR has an interesting clocking scheme that is "mostly synchronous".  The main clock generation is by means of a pulse delay line.  The period is controlled by the type of the microinstruction, which is used to select a tap for recycling the pulse.  Some microinstructions (DISPATCH) run slower than others.  Furthermore, the machine goes asynchronous when performing a memory cycle.  The clock stops completely in the processor once a memory request has gone onto the bus.  If you follow the memory ACK through the schematics, it actually forms the pulse that goes back into the delay line to clock the next cycle.

All of that can be sorted out, but not be any automatic tool.

All in all, the CADR processor architecture is unlike modern designs in a lot of ways that are likely to result in poor automatic synthesis.

I think the more interesting project would be a CADR-compatible design in a modern style.  It would use more pipelining and maybe larger memory maps.  It would have to run CADR microcode unchanged.  It would need to have all of the compatible peripherals so that the microcode could access them.

This discussion reminds me, I had started to write a description of (some of the arcane aspects of) the CADR and how it was morphed into Explorer I and Explorer II.  I'll have to find that.  Oh, and _that_ reminds me that I promised to let the the ex-AI Lab and LMI guys I ran into this summer in Cambridge take a look at it for accurracy and expansion.  I guess I'd better get busy.

    -Steve


 From nyef@softhome.net Sat Aug 28 12:28:00 2004 From: nyef@softhome.net (Nyef) Date: Sat Aug 28 11:28:00 2004 Subject: [LMH]Explorer I EXPT load sources? Message-ID: <20040828191300.GA9737@miyu.paradiesanalytics.com> Hello all. This question is primarily for Steve Krueger, but if anyone else can help out that would be appreciated: Does anyone have the sources for the EXPT diagnostic microload? If so, would they be willing to answer some questions about it? Thanks in advance. --Alastair Bridgewater From brad@heeltoe.com Sun Aug 29 14:15:01 2004 From: brad@heeltoe.com (Brad Parker) Date: Sun Aug 29 13:15:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: Message from Steve Krueger of "Sat, 28 Aug 2004 12:36:32 CDT." <4130C2A0.3040405@sbcglobal.net> Message-ID: <200408292114.i7TLEKvs028301@mwave.heeltoe.com> [if you can, please don't send email in html] Steve Krueger wrote: .... >The CADR has an interesting clocking scheme that is >"mostly synchronous" ... Thanks - that's very interesting. (and I agree what what you are saying; still some of the logic might be nice to have, at least for a rev 0.0) >This discussion reminds me, I had started to write a description of >(some of the arcane aspects of) the CADR and how it was morphed into >Explorer I and Explorer II.  I'll have to find that.  Oh, and _that_ >reminds me that I promised to let the the ex-AI Lab and LMI guys I ran >into this summer in Cambridge take a look at it for accurracy and >expansion.  I guess I'd better get busy. such a document would be interesting. a quick glance looks like some of the microcode decode is very similar... -brad From cgd@theworld.com Mon Aug 30 07:55:01 2004 From: cgd@theworld.com (Chris Daniels) Date: Mon Aug 30 06:55:01 2004 Subject: [LMH]CADR software/images? In-Reply-To: <200408272149.i7RLnimT007782@mwave.heeltoe.com> References: Message from Robert Swindells of "Fri, 27 Aug 2004 22:37:27 BST." <20040827213727.B02B5574@fdy2.demon.co.uk> <200408272149.i7RLnimT007782@mwave.heeltoe.com> Message-ID: <1194.24.131.150.67.1093877033.squirrel@www.TheWorld.com> Robert Swindells wrote: > "This may not make sense, but I would say the business model currently > being persued by the person who owns Symbolics does not allow for any > such entrepreneurial projects." My ex-wife was one of the last employees of Symbolics, and I beleive that when she was let go, the ownership of Symbolics (technically) was by the Boston Law firm Ropes and Grey. The "person in charge" had some ambitions for the software, but I think it's safe to say they don't have a business model to speak of. Simultaneously they remain in a strong position to protect their intellectual properties. I hope this might clarify the nature of the enormous barriers to releasing any of the software. (including the Macsyma software) From rjs@fdy2.demon.co.uk Mon Aug 30 08:32:00 2004 From: rjs@fdy2.demon.co.uk (Robert Swindells) Date: Mon Aug 30 07:32:00 2004 Subject: [LMH]CADR software/images? In-Reply-To: <1194.24.131.150.67.1093877033.squirrel@www.TheWorld.com> (cgd@theworld.com) Message-ID: <20040830153025.66C987BD@fdy2.demon.co.uk> Chris Daniels wrote: >Robert Swindells wrote: >> "This may not make sense, but I would say the business model currently >> being persued by the person who owns Symbolics does not allow for any >> such entrepreneurial projects." I didn't write this, Brad Parker did. >My ex-wife was one of the last employees of Symbolics, and I beleive that >when she was let go, the ownership of Symbolics (technically) was by the >Boston Law firm Ropes and Grey. The "person in charge" had some ambitions >for the software, but I think it's safe to say they don't have a business >model to speak of. Simultaneously they remain in a strong position to >protect their intellectual properties. >I hope this might clarify the nature of the enormous barriers to releasing >any of the software. (including the Macsyma software) I wasn't suggesting that Symbolics release anything. What I was suggesting was that Symbolics are (presumably) short of software developers, while a small group of people have been interested in working on lispm type environments. Some of those interested could form a company or not-for-profit and contract to port OpenGenera to other platforms for Symbolics, with payment being in licences to use the finished product. Symbolics would then be in a position to sell OpenGenera on platforms other than Alpha. Robert Swindells From brad@heeltoe.com Mon Aug 30 13:40:02 2004 From: brad@heeltoe.com (Brad Parker) Date: Mon Aug 30 12:40:02 2004 Subject: [LMH]bored? read pages 7,8,9 and explain Message-ID: <200408302038.i7UKcvlH013855@mwave.heeltoe.com> Hi, If anyone (1) bored, (2) interested in CADR architecture and (3) familiar with ldb and dpb, would you mind reading "A. I. memo 528" page 7,8 & 9 http://www.unlambda.com/lispm/memo528.pdf and then talking with me about how the ldb and dpb parts of the byte instruction should work? I think I get it but I have a head cold and I'm not completely sure. Another set of eyes would help. -brad From stevelisp@grape-krueger.com Mon Aug 30 18:15:02 2004 From: stevelisp@grape-krueger.com (Steve Krueger) Date: Mon Aug 30 17:15:02 2004 Subject: [LMH]bored? read pages 7,8,9 and explain In-Reply-To: <200408302038.i7UKcvlH013855@mwave.heeltoe.com> References: <200408302038.i7UKcvlH013855@mwave.heeltoe.com> Message-ID: <4133D105.5010903@sbcglobal.net> Brad Parker wrote: > Hi, > > If anyone (1) bored, (2) interested in CADR architecture and (3) > familiar with ldb and dpb, would you mind reading "A. I. memo 528" page > 7,8 & 9 > > and then talking with me about how the ldb and dpb parts of the byte > instruction should work? Hi Brad, I can probably answer any question you might have. I know how these instructions work and so find the description in AI Memo 528 to be completely clear (well mostly clear). I suggest you pose a question and I'll try to explain it. Maybe this programmers' summary will help: You use LDB to extract a bit field from a word, aligning it to the lsb end of the word. If the A-SRC is all zero, this produces a result that is zero extended. This is a very frequent operation. The rotation count is the bit number that the field begins at, and the length is the field width. |<- len ->|<--- rot ---->| +--------------------------------+ | | field | | M-SRC +--------------------------------+ |<- len ->| +--------------------------------+ |AAAAAAAAAAAAAAAAAAAAAA| field | RESULT +--------------------------------+ "A" are bits from the A-SRC that are not modified. Only the len bits at the right of the word are modified. DPB is the logical opposite of LDB, it takes an lsb-justified bit field and places it at the bit field described by the rotation and length. |<- len ->| +--------------------------------+ | | field | M-SRC +--------------------------------+ |<- len ->|<--- rot ---->| +--------------------------------+ |AAAAAAA| field |AAAAAAAAAAAAAA| RESULT +--------------------------------+ "A" represents bits from the A-SRC that are not modified. Selective deposite replaces the corresponding bits of the A-SRC with a bit field from the M-SRC described by its position and length. It does a DPB without actually rotating the M-SRC first. |<- len ->|<--- rot ---->| +--------------------------------+ | | field | | M-SRC +--------------------------------+ |<- len ->|<--- rot ---->| +--------------------------------+ |AAAAAAA| field |AAAAAAAAAAAAAA| RESULT +--------------------------------+ "A" represents bits of the A-SRC that are not modified. Note that the length is encoded in the microinstruction as length-1. You could read the Q-DATATYPE field of a Lisp word into an untyped (pure) number with: ((M-1) LDB M-T (BYTE 5. 25.) A-ZERO) This encodes ROT as 25 decimal (31 octal) and length-1 as 4. Note too that fields can be counted from the left instead of the right by doing a bit of arithmetic on the position and width. Does this help or are there more questions? BTW, I don't think I found the left mask and right mask description helpful when I read AI Memo 528 almost 25 years ago either. -Steve From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 21:49:08 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri, 27 Aug 2004 15:49:08 -0500 (CDT) Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> References: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> Message-ID: On Fri, 27 Aug 2004, Brad Parker wrote: > Many, many, many old ITS backup tapes are currently being read and after > some "cleansing" (removing personal info) will be available, I'm told, > on cd-rom. You can assume I'll be like a 4-year old kid camping on the > door step of the people who are doing that :-) Sorry to disappoint, but I have the fabled full dumps of the AI and MC KS10s, including "personal" data, and the LISPM source was not present in the dumps. I checked a year ago, at least. There is already a "free" ITS distribution out on the internet, but I forget where. From brad@heeltoe.com Fri Aug 27 22:06:34 2004 From: brad@heeltoe.com (Brad Parker) Date: Fri, 27 Aug 2004 17:06:34 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: Message from Daniel Seagraves of "Fri, 27 Aug 2004 15:49:08 CDT." Message-ID: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> Daniel Seagraves wrote: >On Fri, 27 Aug 2004, Brad Parker wrote: > >> Many, many, many old ITS backup tapes are currently being read and after >> some "cleansing" (removing personal info) will be available, I'm told, >> on cd-rom. You can assume I'll be like a 4-year old kid camping on the >> door step of the people who are doing that :-) > >Sorry to disappoint, but I have the fabled full dumps of the AI and MC >KS10s, including "personal" data, and the LISPM source was not present in >the dumps. I checked a year ago, at least. There is already a "free" ITS >distribution out on the internet, but I forget where. I was told (by the person who made the dumps originally) that there was something like a full pallet of tapes to be read. What I found on the net was, well, maybe 3-4 tapes worth. It looked more like a partial snap shot of what was on those machines at one point in time (i.e. one day). But not everything. I think these backup tapes span a large amount of time, like serveral years. Does that sounds more plausible? -brad From dseagrav@sakura.lunar-tokyo.net Fri Aug 27 22:17:29 2004 From: dseagrav@sakura.lunar-tokyo.net (Daniel Seagraves) Date: Fri, 27 Aug 2004 16:17:29 -0500 (CDT) Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> References: <200408272106.i7RL6Ybh006142@mwave.heeltoe.com> Message-ID: On Fri, 27 Aug 2004, Brad Parker wrote: > I was told (by the person who made the dumps originally) that there was > something like a full pallet of tapes to be read. Possible. What I have are the hard-disk images of what was on the machines when they were turned off. > What I found on the net was, well, maybe 3-4 tapes worth. It looked > more like a partial snap shot of what was on those machines at one > point in time (i.e. one day). But not everything. I have the list of everything that was supposed to exist. I'll append it to the message. > I think these backup tapes span a large amount of time, like serveral years. GFR was a nightly (?) job, those tapes are probably GFR tapes. Maybe a year or two would cover a pallet. ** AI:LMFONT;-READ- -THIS- ** -*-Text-*- This directory is used by RG, MOON, BEE, ALAN, RMS, TK, H, MMCM, HIC, CENT, CWH, and JLK in connection with the Lisp Machine project. Relevant directories are: AI: LISPM; compiler, simulator, microcode, microassembler, etc. and nuclear system source and qfasl. LISPM1; cold loads and other miscellaneous files. LISPM2; overflow directory for LISPM ZWEI; Editor ZMAIL; Mail reading program LMCONS; console programs. LMDOC; documentation, etc., other than the manual itself. LMFILE; file computer programs LMFONT; Fonts (ast, kst, and qfasl files) LMIO; random I/O code LMIO1; randomer I/O code LMWIN; window system LMWIND; window system documentation LMLREC; LREC files. LMMAN; Lisp Machine manual. LMDEMO; Copies of qfasl files on MC:LMDEMO; LCADR; Software specific to CADR. (ucode, etc) LMPAT; Patch files for Lisp Machine minor system changes. LMPICT; Lisp Machine PICTure stuff LCHESS; Lisp Machine / Chess Machine chess program WOODS; The Woods Lunar rock program CONS; hardware (processor) CONS1; hardware (memory, backplane, etc.) CADRIO; hardware (random i/o devices, IOB) CADRDC; hardware (disk control) CONSWD; hardware (mem interface etc.) CADR; hardware (second) CADRWD; hardware (second machine wirelist) CADRPL; hardware (plot files) CADR1; hardware (second machine misc, bus interface, & c.) CADRM; hardware (memories) CADRMW; hardware (multiwire memory) CADRTV; hardware (tv controllers) CADRPT; hardware (part lists and purchase orders) CHAOS; hardware (chaosnet part of IOB) MC: LMFONT; Copies of fonts on AI:LMFONT; LMDEMO; Sources for demo files LMMAX; The primary MACSYMA directory for the LISPM version LMMAXQ; QFASL files for MACSYMA Relevant TAGS files are: LMDOC; LISPM TAGS - nuclear system LMMAN; .LMMAN TAGS - the manual ZWEI; ZWEI TAGS - new editor KWC; ENDGME TAGS - KWC's program LCHESS; CHESS TAGS - RG's program LMPICT; LT TAGS - vision program MC: MINIMA; MINIMA TAGS - Macsyma WOODS; WOODS TAGS - Lunar ** END ** From rjs@fdy2.demon.co.uk Fri Aug 27 23:20:01 2004 From: rjs@fdy2.demon.co.uk (Robert Swindells) Date: Fri, 27 Aug 2004 23:20:01 +0100 (BST) Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> (message from Brad Parker on Fri, 27 Aug 2004 17:28:25 -0400) Message-ID: <20040827222001.70D685A7@fdy2.demon.co.uk> Brad Parker wrote: >I know many people are interested in a new lisp-os or other forward >looking things. I am not (yet) interested in that. I'm still far too >interested in the old machines and running them. I love my lispms :-) My feeling is that starting from scratch is hard. I like playing with older stuff too. >I'm enjoying the notion of sucking the CADR netlist files into a >modern synthesis tool and using one of the cheap FPGA w/sdram eval >boards to fire up nuevo-cadr. There's certainly enough ram in most >modern parts to hold all the microcode. It's a sick idea but a fun >one... I would be amazed if you could do anything automatically. Creating a CPU in a HDL seems to be a fairly common student project at the moment, I only know of one microcoded one though. You use up the ram in a FPGA fairly quickly if your microcode is 64 bits wide, I have been doing the sums. Robert Swindells From rjs@fdy2.demon.co.uk Sat Aug 28 00:08:52 2004 From: rjs@fdy2.demon.co.uk (Robert Swindells) Date: Sat, 28 Aug 2004 00:08:52 +0100 (BST) Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272303.i7RN3hHx008299@mwave.heeltoe.com> (message from Brad Parker on Fri, 27 Aug 2004 19:03:42 -0400) Message-ID: <20040827230852.450275A7@fdy2.demon.co.uk> Brad Parker wrote: >Robert Swindells wrote: >> >>One other possible host would be Franz Lisp. >> >>It can be run on modern systems, but is probably easier to hack than >>MACLISP. >it's what I've been using on linux, trying to hack it back into a more >maclisp-ish style. It should be pretty close, particularly if you load lmhacks.l. What have you needed to change ? Robert From mikemac@mikemac.com Sat Aug 28 00:19:58 2004 From: mikemac@mikemac.com (Mike McDonald) Date: Fri, 27 Aug 2004 19:19:58 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: Your message of "Fri, 27 Aug 2004 17:28:25 EDT." <200408272128.i7RLSPvV007461@mwave.heeltoe.com> Message-ID: <200408272319.i7RNJwY29863@saturn.mikemac.com> >To: Daniel Seagraves >Date: Fri, 27 Aug 2004 17:28:25 -0400 >From: Brad Parker >I'm enjoying the notion of sucking the CADR netlist files into a modern >synthesis tool and using one of the cheap FPGA w/sdram eval boards to >fire up nuevo-cadr. There's certainly enough ram in most modern parts >to hold all the microcode. It's a sick idea but a fun one... If only ... that company that was directly responsible for the formation of the Free Software Foundation would let go of their stuff, you could build your own G-machine! if only ... Mike McDonald mikemac@mikemac.com Proud owner of an XL1201 From stevelisp@grape-krueger.com Tue Aug 31 02:14:45 2004 From: stevelisp@grape-krueger.com (Steve Krueger) Date: Mon, 30 Aug 2004 20:14:45 -0500 Subject: [LMH] Re: [LMH]bored? read pages 7,8,9 and explain In-Reply-To: <200408302038.i7UKcvlH013855@mwave.heeltoe.com> References: <200408302038.i7UKcvlH013855@mwave.heeltoe.com> Message-ID: <4133D105.5010903@sbcglobal.net> Brad Parker wrote: > Hi, > > If anyone (1) bored, (2) interested in CADR architecture and (3) > familiar with ldb and dpb, would you mind reading "A. I. memo 528" page > 7,8 & 9 > > and then talking with me about how the ldb and dpb parts of the byte > instruction should work? Hi Brad, I can probably answer any question you might have. I know how these instructions work and so find the description in AI Memo 528 to be completely clear (well mostly clear). I suggest you pose a question and I'll try to explain it. Maybe this programmers' summary will help: You use LDB to extract a bit field from a word, aligning it to the lsb end of the word. If the A-SRC is all zero, this produces a result that is zero extended. This is a very frequent operation. The rotation count is the bit number that the field begins at, and the length is the field width. |<- len ->|<--- rot ---->| +--------------------------------+ | | field | | M-SRC +--------------------------------+ |<- len ->| +--------------------------------+ |AAAAAAAAAAAAAAAAAAAAAA| field | RESULT +--------------------------------+ "A" are bits from the A-SRC that are not modified. Only the len bits at the right of the word are modified. DPB is the logical opposite of LDB, it takes an lsb-justified bit field and places it at the bit field described by the rotation and length. |<- len ->| +--------------------------------+ | | field | M-SRC +--------------------------------+ |<- len ->|<--- rot ---->| +--------------------------------+ |AAAAAAA| field |AAAAAAAAAAAAAA| RESULT +--------------------------------+ "A" represents bits from the A-SRC that are not modified. Selective deposite replaces the corresponding bits of the A-SRC with a bit field from the M-SRC described by its position and length. It does a DPB without actually rotating the M-SRC first. |<- len ->|<--- rot ---->| +--------------------------------+ | | field | | M-SRC +--------------------------------+ |<- len ->|<--- rot ---->| +--------------------------------+ |AAAAAAA| field |AAAAAAAAAAAAAA| RESULT +--------------------------------+ "A" represents bits of the A-SRC that are not modified. Note that the length is encoded in the microinstruction as length-1. You could read the Q-DATATYPE field of a Lisp word into an untyped (pure) number with: ((M-1) LDB M-T (BYTE 5. 25.) A-ZERO) This encodes ROT as 25 decimal (31 octal) and length-1 as 4. Note too that fields can be counted from the left instead of the right by doing a bit of arithmetic on the position and width. Does this help or are there more questions? BTW, I don't think I found the left mask and right mask description helpful when I read AI Memo 528 almost 25 years ago either. -Steve From nyef@softhome.net Fri Aug 27 22:53:30 2004 From: nyef@softhome.net (Nyef) Date: Fri, 27 Aug 2004 17:53:30 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> References: <200408272042.i7RKgG4G005716@mwave.heeltoe.com> Message-ID: <20040827215330.GA8306@miyu.paradiesanalytics.com> On Fri, Aug 27, 2004 at 04:42:14PM -0400, Brad Parker wrote: > > (ps: i think if someone says lm-2 sources are unbuildable w/o a cadr I > don't think they have not tried very hard; but I may be wrong) Have a look at the function ASSEMBLE in sys/sys/cadrlp.lisp.134. Note the references to things in the SI and FS packages. Note the use of READ-METER. Those references won't build on a modern system. They might build on MACLISP for all I know, but they don't build on SBCL. Have a look at sys/sys2/comdefs.lisp.8, specifically the DEFUN for COMPILING-WHOLE-FILE-P. Note that it calls the stream as if it were a Flavors object. Is that going to work in MACLISP? Have a look at sys/sys/qcdefs.lisp.349, ADD-OPTIMIZER. I don't believe that MACLISP supports the use of "E in an arglist. I know that SBCL doesn't. I'm not saying that they're completely unbuildable, just that it'll take some hacking (possibly a lot of hacking) to get it working. --Alastair Bridgewater From stevelisp@grape-krueger.com Sat Aug 28 18:36:32 2004 From: stevelisp@grape-krueger.com (Steve Krueger) Date: Sat, 28 Aug 2004 12:36:32 -0500 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> References: <200408272128.i7RLSPvV007461@mwave.heeltoe.com> Message-ID: <4130C2A0.3040405@sbcglobal.net> Brad Parker wrote:
I'm enjoying the notion of sucking the CADR netlist files into a modern
synthesis tool and using one of the cheap FPGA w/sdram eval boards to
fire up nuevo-cadr.  There's certainly enough ram in most modern parts
to hold all the microcode.  It's a sick idea but a fun one...
I'd be surprised if that would work at all.  The CADR has an interesting clocking scheme that is "mostly synchronous".  The main clock generation is by means of a pulse delay line.  The period is controlled by the type of the microinstruction, which is used to select a tap for recycling the pulse.  Some microinstructions (DISPATCH) run slower than others.  Furthermore, the machine goes asynchronous when performing a memory cycle.  The clock stops completely in the processor once a memory request has gone onto the bus.  If you follow the memory ACK through the schematics, it actually forms the pulse that goes back into the delay line to clock the next cycle.

All of that can be sorted out, but not be any automatic tool.

All in all, the CADR processor architecture is unlike modern designs in a lot of ways that are likely to result in poor automatic synthesis.

I think the more interesting project would be a CADR-compatible design in a modern style.  It would use more pipelining and maybe larger memory maps.  It would have to run CADR microcode unchanged.  It would need to have all of the compatible peripherals so that the microcode could access them.

This discussion reminds me, I had started to write a description of (some of the arcane aspects of) the CADR and how it was morphed into Explorer I and Explorer II.  I'll have to find that.  Oh, and _that_ reminds me that I promised to let the the ex-AI Lab and LMI guys I ran into this summer in Cambridge take a look at it for accurracy and expansion.  I guess I'd better get busy.

    -Steve


 From brad@heeltoe.com Sun Aug 29 22:14:20 2004 From: brad@heeltoe.com (Brad Parker) Date: Sun, 29 Aug 2004 17:14:20 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: Message from Steve Krueger of "Sat, 28 Aug 2004 12:36:32 CDT." <4130C2A0.3040405@sbcglobal.net> Message-ID: <200408292114.i7TLEKvs028301@mwave.heeltoe.com> [if you can, please don't send email in html] Steve Krueger wrote: .... >The CADR has an interesting clocking scheme that is >"mostly synchronous" ... Thanks - that's very interesting. (and I agree what what you are saying; still some of the logic might be nice to have, at least for a rev 0.0) >This discussion reminds me, I had started to write a description of >(some of the arcane aspects of) the CADR and how it was morphed into >Explorer I and Explorer II.  I'll have to find that.  Oh, and _that_ >reminds me that I promised to let the the ex-AI Lab and LMI guys I ran >into this summer in Cambridge take a look at it for accurracy and >expansion.  I guess I'd better get busy. such a document would be interesting. a quick glance looks like some of the microcode decode is very similar... -brad From brad@heeltoe.com Sat Aug 28 00:16:43 2004 From: brad@heeltoe.com (Brad Parker) Date: Fri, 27 Aug 2004 19:16:43 -0400 Subject: [LMH] Re: [LMH]CADR software/images? In-Reply-To: Message from Robert Swindells of "Sat, 28 Aug 2004 00:08:52 BST." <20040827230852.450275A7@fdy2.demon.co.uk> Message-ID: <200408272316.i7RNGhK2008482@mwave.heeltoe.com> Robert Swindells wrote: > >It should be pretty close, particularly if you load lmhacks.l. > >What have you needed to change ? lots of little things to "port" to linux ldb, dpb, byte, locf missing some things with arrays too, if I recall. I hacked the reader to, to do some odd escapes. I should diff it all and see. -brad