localroger said...
Yeah Phil, but you werent' selling people shares in your eventual success on the Netflix thing.
Oh, geez. I'm so sorry. You didn't get a prospectus? The contest may have just ended, but it's not too late to get in on the, uh ... top floor. Yeah, that's it: top floor! That's the ticket!
Rsadeika said...
As far as I can tell we have had a lot of posturing, but nobody has shown me any real proof, there own personal theories, and beliefs, but no proof.
Ray,
A very neat summation. We have had a lot of posturing from Mark (in these forums) and Dr Jim (in his videos), but no real proof. We have heard their own personal theories and beliefs - but again, no proof.
To quote Carl Sagan (amongst others) "Extraordinary claims require extraordinary evidence". Mark and Dr Jim are the ones making the extraordinary claims here.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
Thanks for doing this research - the articles you have pointed us to contain more real information than anything Mark or Dr Jim have yet given us.
James O. Gouge, Sally B. Gouge, and Peter Krueger said...
In the junctives technology, the first step is the extraction and encoding of image pixel-pattern components to form determinants that are then fed into a synaptic generator (see Fig. 1). In a manner similar to the way the human brain forms synapses or junctions during learning, the machine also forms synapses as it is taught to recognize features and objects. This formation is displayed on the computer screen as a report of the number of new synapses formed based on the training on the region-of-interest window in the current image. These synapses become a part of the synaptic network that makes up the basal interconnective paths to the knowledge elements or "knixels." The knixels become the basic element learned by the computer and are taken from the "knixel pool" and assigned via the synapses generated as a basis for their interconnection. Knixels can be associated together to form themes. These can be linked to external databases.
On the basis of this, and also of the patent extracts you quoted, one would have to conclude that Dr Jim's solution for machine intelligence is based on a fairly traditional neural network model - but renamed to appear sufficently novel to justify a patent application.
To those who have played with neural networks in the past, it will come as no surprise to find yet another person trying to make the leap from visual processing to cognitive processing - it's a well trodden path. All attempts to do this have so far been unsuccessful (at least to my knowledge - I'll admit to being a few decades out of date now, but then this approach was already a few decades old when Dr Jim published his article).
While Dr Jim is not the only one who still believes that this is a valid approach (I do myself) it simply doesn't work on the scale Dr Jim is trying to achieve it - a couple of Propeller chips and 16Mb or RAM is MANY orders of magnitude too small for the task.
Ross.
P.S. I agree about "knixels" - what a neat buzzword!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
Given that there are ~100 billion neurons in the brain, you'd probably only need a million or so of those GPUs to be able to simulate it - if you could figure out how to connect them all together (there are 100 TRILLION synaptic connections between these 100 billion neurons).
But one single GPU might be enough to simulate an insect brain, given that they have only about 100,000 neurons.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
--> in approx 20 years (if current trends continue) a GPU may be able to simulate a human brain
RossH said...
@Rayman,
Nice link. NOW we're talking!
Given that there are ~100 billion neurons in the brain, you'd probably only need a million or so of those GPUs to be able to simulate it - if you could figure out how to connect them all together (there are 100 TRILLION synaptic connections between these 100 billion neurons).
But one single GPU might be enough to simulate an insect brain, given that they have only about 100,000 neurons.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full Morpheus & Mem+ Advanced dual Propeller SBC with XMM and 256 Color VGA - PCB, kit, A&T available! www.mikronauts.com - my site 6.250MHz custom Crystals for running Propellers at 100MHz Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
A few days ago I reposted some specifications that mallread had previously posted. I had already commented on this in a previous thread. No response to either. They are not doing anything special with the hardware, especially when they can only achieve half the throughput from hub to cog that a number of us are already achieving.
Mainframes: Mallread said there were only a few (Univac, IBM & CDC) but in fact there were many... including ICL,·Sperry, Unisys (I think from a merger of Univac and ?),·Fujitsu. It was certainly my understanding that the DEC VAX was represented as a mainframe even though the mini I worked on sold against it. I also believe Burroughs made a mainframe.
And just to correct localrogers' interpretation about Windows NT, 2000 etc. They do not have their roots in mainframe operating systems. We are still waiting for Windows to remove the missunderstanding of mainframe operating systems used by·the original writers of Windows - and it's IBM predecessor variant - was it called OS/2???
Just to offer an olive branch... In the 90's it was said that based on current computer increases in performance, it would take >100 years to map the human genome. However, in 5-10 years it was mapped using a new algorithm. So indeed, mallred and Dr Jim could be onto something.
However, given their refusal to answer anything technical, I remain skeptical and believe they are "gouging" (pun intended)·the market with their overpriced product.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
However, given their refusal to answer anything technical, I remain skeptical and believe they are "gouging" (pun intended) the market with their overpriced product.
Beanie2K - yes forgot about Honeywell. I did't know Xerox had mainframes but it doesn't surprise me given their other work in Xeorx Park. My wife and I were trying to recall the other brands of the 70's & 80's. My wife worked on ICL mainframes.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
Xerox actually bought Sigma and renamed it as Xerox Data Systems. They paid far more than the company was worth, not that it mattered as they were making vast profits from their photocopier business. I was working for Xerox-Research (UK) at the time, and used an XDS Sigma-7 in the USA via a terminal. It was a very good machine with some unusual software on it, like SNOBOL.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Amdahl - yes I recall the name. I know there were others.
Another question... How old is Dr Jim ??? I figure the Apollo Telemetry code must have been written at least by 1967 and surely NASA would have required a university qualification plus some years experience. So my estimate is that Dr Jim must be 65++
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
If you watch this series about the LEM on YouTube, you'll get some insights about "software engineering" at that time.
< (part 1 of 5, all worth watching!)
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.: YADRO
Nick: If you really want to get into software engineering circa 1968 you can get all the original Apollo command module software from the Virtual AGC project web site www.ibiblio.org/apollo/download.html#Source_Highlighting complete with an AGC assembler simulator and a lot more.
For example here is the download telemetry program as designed by Dr Jim I guess:
I like the comment about totally rewriting it to save 150 bytes of memory !
Now to get that AGC simulator compiled with Catalina and running on a Prop....
# Copyright: Public domain.
# Filename: DOWN_TELEMETRY_PROGRAM.agc
# Purpose: Part of the source code for Colossus, build 249.
# It is part of the source code for the Command Module's (CM)
# Apollo Guidance Computer (AGC), possibly for Apollo 8 and 9.
# Assembler: yaYUL
# Reference: Starts on p. 1075 of 1701.pdf.
# Contact: Ron Burkey <info@sandroid.org>.
# Website: www.ibiblio.org/apollo.
# Mod history: 08/28/04 RSB. Adapted from corresponding Luminary 131 file.
#
# The contents of the "Colossus249" files, in general, are transcribed
# from a scanned document obtained from MIT's website,
# http://hrst.mit.edu/hrs/apollo/public/archive/1701.pdf. Notations on this
# document read, in part:
#
# Assemble revision 249 of AGC program Colossus by NASA
# 2021111-041. October 28, 1968.
#
# This AGC program shall also be referred to as
# Colossus 1A
#
# Prepared by
# Massachussets Institute of Technology
# 75 Cambridge Parkway
# Cambridge, Massachusetts
# under NASA contract NAS 9-4065.
#
# Refer directly to the online document mentioned above for further information.
# Please report any errors (relative to 1701.pdf) to info@sandroid.org.
#
# In some cases, where the source code for Luminary 131 overlaps that of
# Colossus 249, this code is instead copied from the corresponding Luminary 131
# source file, and then is proofed to incorporate any changes.
# Page 1075
# PROGRAM NAME -- DOWN TELEMETRY PROGRAM
# MOD NO. -- 0 TO COMPLETELY REWRITE THE DOWN TELEMETRY PROGRAM AND DOWNLINK ERASABLE DUMP PROGRAM FOR THE
# PURPOSE OF SAVING APPROXIMATELY 150 WORDS OF CORE STORAGE.
# THIS CHANGE REQUIRES AN ENTIRELY NEW METHOD OF SPECIFYING DOWNLINK LISTS. REFER TO DOWNLINK
# LISTS LOG SECTION FOR MORE DETAILS. HOWEVER THIS CHANGES WILL NOT AFFECT THE GROUND PROCESSING
# OF DOWN TELEMETRY DATA.
# MOD BY -- KILROY, SMITH, DEWITT
# DATE -- 02 OCT 67
# AUTHORS -- KILROY, SMITH, DWWITT, DEWOLF, FAGIN
# LOG SECTION -- DOWN-TELEMETRY PROGRAM
#
# FUNCTIONAL DESCRIPTION -- THIS ROUTINE IS INITIATED BY TELEMETRY END
# PULSE FROM THE DOWNLINK TELEMETRY CONVERTER. THIS PULSE OCCURS
# AT 50 TIMES PER SEC (EVERY 20 MS) THEREFORE DODOWNTM IS
# EXECUTED AT THESE RATES. THIS ROUTINE SELECTS THE APPROPRIATE
# AGC DATA TO BE TRANSMITTED DOWNLINK AND LOADS IT INTO OUTPUT
# CHANNELS 34 AND 35. THE INFORMATION IS THEN GATED OUT FROM THE
# LGC IN SERIAL FASHION.
#
# THIS PROGRAM IS CODED FOR A 2 SECOND DOWNLIST. SINCE DOWNRUPTS
# OCCUR EVERY 20 MS AND 2 AGC COMPUTER WORDS CAN BE PLACED IN
# CHANNELS 34 AND 35 DURING EACH DOWNRUPT THE PROGRAM IS CAPABLE
# OF SENDING 200 AGC WORDS EVERY 2 SECONDS.
#
# CALLING SEQUENCE -- NONE
# PROGRAM IS ENTERED VIA TCF DODOWNTM WHICH IS EXECUTED AS A
# RESULT OF A DOWNRUPT. CONTROL IS RETURNED VIA TCF RESUME WHICH
# IN EFFECT IS A RESUME.
#
# SUBROUTINES CALLED -- NONE
#
# NORMAL EXIT MODE -- TCF RESUME
#
# ALARM OR ABORT EXIT MODE -- NONE
#
# RESTART PROTECTION:
# ON A FRESH START AND RESTART THE `STARTSUB' SUBROUTINE WILL INITIALIZE THE DOWNLIST POINTER (ACTUALLY
# DNTMGOTO) TO THE BEGINNING OF THE CURRENT DOWNLIST (I.E., CURRENT CONTENTS OF DNLSTADR). THIS HAS THE
# EFFECT OF IGNORING THE REMAINDER OF THE DOWNLIST WHICH THE DOWN-TELEMETRY PROGRAM WAS WORKING ON WHEN
# THE RESTART (OR FRESH START) OCCURRED AND RESUME DOWN TELEMETRY FROM THE BEGINNING OF THE CURRENT
# DOWNLIST.
#
# ALSO OF INTEREST IS THE FACT THAT ON A RESTART THE AGC WILL ZERO DOWNLINK CHANNELS 13, 34 AND 35.
#
# DOWNLINK LIST SELECTION:
# THE APPROPRIATE DOWNLINK LISTS ARE SELECTED BY THE FOLLOWING:
# 1. FRESH START
# 2. V37EXXE WHERE XX = THE MAJOR MODE BEING SELECTED.
# 3. UPDATE PROGRAM (P27)
# 4. NON-V37 SELECTABLE TYPE PROGRAMS (E.G., AGS INITIALIZATION (SUNDANCE, LUMINARY) AND P61-P62
# TRANSITIONS (COLOSSUS) ETC.).
#
# DOWNLINK LIST RULES AND LIMITATIONS:
# READ SECTION(S) WHICH FOLLOW `DEBRIS' WRITEUP.
#
# OUTPUT -- EVERY 2 SECONDS 100 DOUBLE PRECISION WORDS (I.E., 200 LGC
# COMPUTER WORDS) ARE TRANSMITTED VIA DOWNLINK.
#
# ERASABLE INITIALIZATION REQUIRED -- NONE
# `DNTMGOTO' AND `DNLSTADR' ARE INITIALIZED BY THE FRESH START PROGRAM.
#
# DEBRIS (ERASABLE LOCATIONS DESTROYED BY THIS PROGRAM) --
# LDATALST, DNTMBUFF TO DNTMBUFF +21D, TMINDEX, DNQ.
# Page 1076
# (No source on this page of the original assembly listing.)
# Page 1077
# DODOWNTM IS ENTERED EVERY 20 MS BY AN INTERRUPT TRIGGERED BY THE
# RECEIPT OF AN ENDPULSE FROM THE SPACECRAFT TELEMETRY PROGRAMMER.
#
# NOTES REGARDING DOWNLINK LISTS ASSOCIATED WITH THIS PROGRAM:
# 1. DOWNLISTS. DOWNLISTS MUST BE COMPILED IN THE SAME BANK AS THE
# DOWN-TELEMETRY PROGRAM. THIS IS DONE FOR EASE OF CODING, FASTER
# EXECUTION.
# 2. EACH DOWNLINK LIST CONSISTES OF A CONTROL LIST AND A NUMBER OF
# SUBLISTS.
# 3. A SUBLIST REFERS TO A SNAPSHOT OR DATA COMMON TO THE SAME OR OTHER
# DOWNLINK LISTS. ANY SUBLIST CONTAINING COMMON DATA NEEDS TO BE
# CODED ONLY ONCE FOR THE APPLICABLE DOWNLINK LISTS.
# 4. SNAPSHOT SUBLISTS REFER SPECIFICALLY TO HOMOGENEOUS DATA WHICH MUST BE
# SAVED IN A BUFFER DURING ONE DOWNRUPT.
# 5. THE 1DNADR FOR THE 1ST WORD OF SNAPSHOT DATA IS FOUND AT THE END
# OF EACH SNAPSHOT SUBLIST, SINCE THE PROGRAM CODING SENDS THIS DP WORD
# IMMEDIATELY AFTER STORING THE OTHERS IN THE SNAPSHOT BUFFER.
# 6. ALL LISTS ARE COMBINATIONS OF CODED ERASABLE ADDRESS CONSTANTS
# CREATED FOR THE DOWNLIST PROGRAM.
# A. 1DNADR 1-WORD DOWNLIST ADDRESS.
# SAME AS ECADR, BUT USED WHEN THE WORD ADDRESSED IS THE LEFT
# HALF OF A DOUBLE-PRECISION WORD FOR DOWN TELEMETRY.
# B. 2DNADR - 6DNADR N-WORD DOWNLIST ADDRESS, N = 2 - 6.
# SAME AS 1DNADR, BUT WTIH THE 4 UNUSED BITS OF THE ECADR FORMAT
# FILLED IN WITH 0001-0101. USED TO POINT TO A LIST OF N DOUBLE-
# PRECISION WORDS, STORED CONSECUTIVELY, FOR DOWN TELEMETRY.
# C. DNCHAN DOWNLIST CHANNEL ADDRESS.
# SAME AS 1DNADR, BUT WITH PREFIX BITS 0111. USED TO POINT TO
# A PAIR OF CHANNELS FOR DOWN TELEMETRY.
# D. DNPTR DOWN-TELEMETRY SUBLIST POINTER.
# SAME AS CAF BUT TAGGES AS A CONSTANT. USED IN CONTROL LIST TO POINT TO A SUBLIST.
# CAUTION --- A DNPTR CANNOT BE USED IN A SUBLIST.
# 7. THE WORD ORDER CODE IS SET TO ZERO AT THE BEGINNING OF EACH DOWNLIST (I.E., CONTROL LIST) AND WHEN
# A `1DNADR TIME2' IS DETECTED IN THE CONTROL LIST (ONLY).
# 8. IN THE SNAPSHOT SUBLIST ONLY, THE DNADR'S CANNOT POINT TO THE FIRST WORD OF ANY EBANK.
#
# DOWNLIST LIST RESTRICTIONS:
# (THE FOLLOWING POINTS MAY BE LISTED ELSEWHERE BUT ARE LISTED HERE SO IT IS CLEAR THAT THESE THINGS CANNOT BE
# DONE)
# 1. SNAPSHOT DOWNLIST:
# (A) CANNOT CONTAIN THE FOLLOWING ECADRS (I.E., 1DNADR'S): Q, 400, 1000, 1400, 2000, 2400, 3000, 3400.
# (B) CAN CONTAIN ONLY 1DNADR'S
# 2. ALL DOWNLINKED DATA (EXCEPT CHANNELS) IS PICKED UP BY A DCA SO DOWNLINK LISTS CANNOT CONTAIN THE
# EQUIVALENT OF THE FOLLOWING ECADRS (I.E., IDNADRS): 377, 777, 1377, 1777, 2377, 2777, 3377, 3777.
# (NOTE: TE TERM `EQUIVALENT' MEANT THAT THE IDNADR TO 6DNADR WILL BE PROCESSED LIKE 1 TO 6 ECADRS)
# 3. CONTROL LISTS AND SUBLISTS CANNOT HAVE ENTRIES = OCTAL 00000 OR OCTAL 77777
# Page 1078
# 4. THE `1DNADR TIME2' WHICH WILL CAUSE THE DOWNLINT PROGRAM TO SET THE WORDER CODE TO 3 MUST APPEAR IN THE
# CONTROL SECTION OF THE DOWNLIST.
# 5. `DNCHAN 0' CANNOT BE USED.
# 6. `DNPTR 0' CANNOT BE USED.
# 7. DNPTR CANNOT APPEAR IN A SUBLIST.
#
# EBANK SETTINGS
# IN THE PROCESS OF SETTING THE EBANK (WHEN PICKING UP DOWNLINK DATA) THE DOWN TELEMETRY PROGRAM PUTS
# `GARBAGE' INTO BITS15-12 OF EBANK. HUGH BLAIR-SMITH WARNS US THAT BITS15-12 OF EBANK MAY BECOME
# SIGNIFICANT SOMEDAY IN THE FUTURE. IF/WHEN THAT HAPPENS, THE PROGRAM SHOULD INSURE (BY MASKING ETC.)
# THAT BITS 15-12 OF EBANK ARE ZERO.
#
# INITIALIZATION REQUIRED -- TO INTERRUPT CURRENT LIST AND START A NEW ONE.
# 1. ADRES OF DOWNLINK LIST INTO DNLSTADR
# 2. NEGONE INTO SUBLIST
# 3. NEGONE INTO DNECADR
BANK 22
SETLOC DOWNTELM
BANK
EBANK= DNTMBUFF
COUNT 05/DPROG
DODOWNTM TS BANKRUPT
EXTEND
QXCH QRUPT # SAVE Q
CA BIT7 # SET WORD ORDER CODE TO 1. EXCEPTION: AT
EXTEND # THE BEGINNING OF EACH LIST THE WORD
WOR CHAN13 # CODE WILL BE SET BACK TO 0.
TC DNTMGOTO # GOTO APPROPRIATE PHASE OF PROGRAM
DNPHASE1 CA NEGONE # INITIALIZE ALL CONTROL WORDS
TS SUBLIST # WORDS TO MINUS ONE
TS DNECADR
CA LDNPHAS2 # SET DNTMGOTO = 0 ALL SUSEQUENT DOWRUPTS
TS DNTMGOTO # GO TO DNPHASE2
TCF NEWLIST
DNPHASE2 CCS DNECADR # SENDING OF DATA IN PROGRESS
DODNADR TC FETCH2WD # YES -- THEN FETCH THE NEXT 2 SP WORDS
MINTIME2 -1DNADR TIME2 # NEGATIVE OF TIME2 1DNADR
TCF +1 # (ECADR OF 3776 + 74001 = 77777)
CCS SUBLIST # IS THE SUBLIST IN CONTROL
# Page 1079
TCF NEXTINSL # YES
DNADRDCR OCT 74001 # DNADR COUNT AND ECADR DECREMENTER
CHKLIST CA CTLIST
EXTEND
BZMF NEWLIST # IT WILL BE NEGATIVE AT END OF LIST
TCF NEXTINCL
NEWLIST INDEX DNLSTCOD
CA DNTABLE # INITIALIZE CTLIST WITH
TS CTLIST # STARTING ADDRESS OF NEW LIST
CS DNLSTCOD
TCF SENDID +3
NEXTINCL INDEX CTLIST
CA 0
CCS A
INCR CTLIST # SET POINTER TO PICK UP NEXT CTLIST WORD
TCF +4 # ON NEXT ENTRY TO PROG. (A SHOULD NOT =0)
XCH CTLIST # SET CTLIST TO NEGATIVE AND PLACE(CODING)
COM # UNCOMPLEMENTED DNADR INTO A. (FOR LA)
XCH CTLIST # (ST IN )
+4 INCR A # (CTLIST)
TS DNECADR # SAVE DNADR
AD MINTIME2 # TEST FOR TIME2 (NEG. OF ECADR)
CCS A
TCF SETWO +1 # DON'T SET WORD ORDER CODE
MINB1314 OCT 47777 # MINUS BIT 13 AND 14 (CAN'T GET HERE)
TCF SETWO +1 # DON'T SET WORD ORDER CODE
SETWO TC WOZERO # GO SET WORD ORDER CODE TO ZERO.
+1 CA DNECADR # RELOAD A WITH THE DNADR.
+2 AD MINB1314 # IS THIS A REGULAR DNADR?
EXTEND
BZMF FETCH2WD # YES. (A MUST NEVER BE ZERO)
AD MINB12 # NO. IS IT A POINTER (DNPTR) OR A
EXTEND # CHANNEL(DNCHAN)
BZMF DODNPTR # IT'S A POINTER. (A MUST NEVER BE ZERO)
DODNCHAN TC 6 # (EXECUTED AS EXTEND) IT'S A CHANNEL
INDEX DNECADR
INDEX 0 -4000 # (EXECUTED AS READ)
TS L
TC 6 # (EXECUTED AS EXTEND)
INDEX DNECADR
INDEX 0 -4001 # (EXECUTED AS READ)
TS DNECADR # SET DNECADR
CA NEGONE # TO MINUS
XCH DNECADR # WHILE PRESERVING A.
TCF DNTMEXIT # GO SEND CHANNELS
WOZERO CS BIT7
EXTEND
# Page 1080
WAND CHAN13 # SET WORD ORDER CODE TO ZERO
TC Q # RETURN TO CALLER
DODNPTR INDEX DNECADR # DNECADR CONTAINS ADRES OF SUBLIST
0 0 # CLEAR AND ADD LIST ENTRY INTO A.
CCS A # IS THIS A SNAPSHOT SUBLIST
CA DNECADR # NO, IT IS A REGULAR SUBLIST.
TCF DOSUBLST # A MUST NOT BE ZERO.
XCH DNECADR # YES. IT IS A SNAPSHOT SUBLIST.
TS SUBLIST # C(DNECADR) INTO SUBLIST
CAF ZERO # A INTO A
XCH TMINDEX # (NOTE: TMINDEX = DNECADR)
# THE FOLLOWING CODING (FROM SNAPLOOP TO SNAPEND) IS FOR THE PURPOSE OF TAKING A SNAPSHOT OF 12 DP REGISTERS.
# THIS IS DONE BY SAVING 11 DP REGISTERS IN DNTMBUFF AND SENDING THE FIRST DP WORD IMMEDIATELY.
# THE SNAPSHOT PROCESSING IS THE MOST TIME CONSUMING AND THEREFORE THE CODING AND LIST STRUCTURE WERE DESIGNED
# TO MINIMIZE TIME. THE TIME OPTIMIZATION RESULTS IN RULES UNIQUE TO THE SNAPSHOT PORTION OF THE DOWNLIST.
# THESE RULES ARE ......
# 1. ONLY 1DNADR'S CAN APPEAR IN THE SNAPSHOT SUBLIST
# 2. THE 1DNADR'S CANNOT REFER TO THE FIRST LOCATION IN ANY BANK.
SNAPLOOP TS EBANK # SET EBANK
MASK LOW8 # ISOLATE RELATIVE ADDRESS
EXTEND
INDEX A
EBANK= 1401
DCA 1401 # PICK UP 2 SNAPSHOT WORDS.
EBANK= DNTMBUFF
INDEX TMINDEX
DXCH DNTMBUFF # STORE 2 SNAPSHOT WORDS IN BUFFER
INCR TMINDEX # SET BUFFER INDEX FOR NEXT 2 WORDS.
INCR TMINDEX
SNAPAGN INCR SUBLIST # SET POINTER TO NEXT 2 WORDS OF SNAPSHOT
INDEX SUBLIST
0 0 # = CA SSSS (SSSS = NEXT ENTRY IN SUBLIST)
CCS A # TEST FOR LAST TWO WORDS OF SNAPSHOT.
TCF SNAPLOOP # NOT LAST TWO.
LDNPHAS2 GENADR DNPHASE2
TS SUBLIST # YES, LAST. SAVE A.
CA NEGONE # SET DNECADR AND
TS DNECADR # SUBLIST POINTERS
XCH SUBLIST # TO NEGATIVE VALUES
TS EBANK
MASK LOW8
EXTEND
INDEX A
EBANK= 1401
# Page 1081
DCA 1401 # PICK UP FIRST 2 WORDS OF SNAPSHOT.
EBANK= DNTMBUFF
SNAPEND TCF DNTMEXIT # NOW TO SEND THEM.
FETCH2WD CA DNECADR
TS EBANK # SET EBANK
MASK LOW8 # ISOLATE RELATIVE ADDRESS
TS L
CA DNADRDCR # DECREMENT COUNT AND ECADR
ADS DNECADR
EXTEND
INDEX L
EBANK= 1400
DCA 1400 # PICK UP 2 DATA WORDS
EBANK= DNTMBUFF
TCF DNTMEXIT # NOW GO SEND THEM.
DOSUBLST TS SUBLIST # SET SUBLIST POINTER
NEXTINSL INDEX SUBLIST
0 0 # = CA SSSS (SSSS = NEXT ENTRY IN SUBLIST)
CCS A # IS IT THE END OF THE SUBLIST
INCR SUBLIST # NO --
TCF +4
TS SUBLIST # SAVE A.
CA NEGONE # SET SUBLIST TO MINUS
XCH SUBLIST # RETRIEVE A.
+4 INCR A
TS DNECADR # SAVE DNADR
TCF SETWO +2 # GO USE COMMON CODING (PROLEMS WOULD
# OCCUR IF THE PROGRAM ENCOUNTERED A
# DNPTR NOW)
DNTMEXIT EXTEND # DOWN-TELEMETRY EXIT
WRITE DNTM1 # TO SEND A + L TO CHANNELS 34 + 35
CA L # RESPECTIVELY
TMEXITL EXTEND
WRITE DNTM2
TMRESUME TCF RESUME # EXIT TELEMTRY PROGRAM VIA RESUME.
MINB12 EQUALS -1/8
DNECADR EQUALS TMINDEX
CTLIST EQUALS LDATALST
SUBLIST EQUALS DNQ
# Page 1082
# SUBROUTINE NAME -- DNDUMP
#
# FUNCTIONAL DESCRIPTION -- TO SEND (DUMP) ALL 8 BANKS OF ERASABLE STORAGE TWICE. BANKS ARE SENT ONE AT A TIME
# EACH BANK IS PRECEDED BY AN ID WORD, SYNCH BITS, ECADR AND TIME1 FOLLOWED BY THE 256D WORDS OF EACH
# EBANK. EBANKS ARE DUMPED IN ORDER (I.E., EBANK 0 FIRST, THEN EBANK1 ETC.)
#
# CALLING SEQUENCE -- THE GROUND OR ASTRONAUT BY KEYING V74E CAN INITIALIZE THE DUMP.
# AFTER KEYING IN V74E THE CURRENT DOWNLIST WILL BE IMMEDIATELY TERMINATED AND THE DOWNLINK ERASABLE DUMP
# WILL BEGIN.
#
# ONCE INITITIATED THE DOWNLINK ERASABLE DUMP CAN BE TERMINATED (AND INTERRUPTED DOWNLIST REINSTATED) ONLY
# BY THE FOLLOWING:
#
# 1. A FRESH START
# 2. COMPLETION OF ALL DOWNLINK DUMPS REQUESTED (ACCORDING TO BITS SET IN DUMPCNT). NOTE THAT DUMPCNT
# CAN BE ALTERED BY A V21N01.
# 3. AND INVOLUNTARILY BY A RESTART.
#
# NORMAL EXIT MODE -- TCF DNPHASE1
#
# ALARM OR ABORT MODE -- NONE
#
# *SUBROUTINES CALLED -- NONE
#
# ERASABLE INITIALIZATION REQUIRED --
# DUMPCNT OCT 20000 IF 4 COMPLETE ERASABLE DUMPS ARE DESIRED
# DUMPCNT OCT 10000 IF 2 COMPLETE ERASABLE DUMPS ARE DESIRED
# DUMPCNT OCT 04000 IF 1 COMPLETE ERASABLE DUMP IS DESIRED
#
# DEBRIS -- DUMPLOC, DUMPSW, DNTMGOTO, EBANK, AND CENTRAL REGISTERS
#
# TIMING -- TIME (IN SECS) = ((NO.DUMPS)*(NO.EBANKS)*(WDSPEREBANK + NO.IDWDS)) / NO.WDSPERSEC
# TIME (IN SECS) = ( 4 )*( 8 )*( 256 + 4 ) / 100
# THUS TIME (IN SECS TO SEND DUMP OF ERASABLE 4 TIMES VIA DOWNLINK) = 83.2 SECONDS
#
# STRUCTURE OF ONE EBANK AS IT IS SENT BY DOWNLINK PROGRAM --
# (REMINDER -- THIS ONLY DESCRIBES ONE OF THE 8 EBANKS X 4 (DUMPS) = 32 EBANKS WHICH WILL BE SENT BY DNDUMP)
#
# DOWNLIST W
# WORD TAKEN FROM CONTENTS OF EXAMPLE O COMMENTS
# 1 ERASID 0177X 0 DOWNLIST I.D. FOR DOWNLINK ERASABLE DUMP (X=7 CSM, 6 LM)
# 2 LOWIDCOD 77340 1 DOWNLINK SYNCH BITS. (SAME ONE USED IN ALL OTHER DOWNLISTS)
# 3 DUMPLOC 13400 1 (SEE NOTES ON DUMPLOC) 1 = 3RD ERAS DUMP, 3400=ECADR OF 5TH WD
# 4 TIME1 14120 1 TIME IN CENTISECONDS
# 5 FIRST WORD OF EBANK X 03400 1 IN THIS EXAMPLE THIS WORD = CONTENTS OF E7,1400 (ECADR 3400)
# 6 2ND WORD OF EBANK X 00142 1 IN THIS EXAMPLE THIS WORD = CONTENTS OF E7,1401 (ECADR 3401)
# 7. 3RD WORD OF EBANK X 00142 1 IN THIS EXAMPLE THIS WORD = CONTENTS OF E7,1402 (ECADR 3402)
# .
# .
# .
# 260D 256TH WORD OF EBANK X 03777 1 IN THIS EXAMPLE THIS WORD = CONTENTS OF E7,1777 (ECADR 3777)
#
# NOTE -- DUMPLOC CONTAINS THE COUNTER AND ECADR FOR EACH WORD BEING SENT.
# THE BIT STRUCTURE OF DUMPLOC IS FOLLOW --
# X = NOT USED
# X ABC EEE RRRRRRRR ABC = ERASABLE DUMP COUNTER (I.E. ABC = 0,1,2, OR 3 WHICH MEANS THAT
# COMPLETE ERASABLE DUMP NUMBER 1,2,3, OR 4 RESPECTIVELY IS IN PROGRESS)
# EEE = EBANK BITS
# RRRRRRRR = RELATIVE ADDRESS WITHIN AN EBANK
# Page 1083
DNDUMPI CA ZERO # INITIALIZE DOWNLINK
TS DUMPLOC # ERASABLE DUMP
+2 TC SENDID # GO SEND ID AND SYNCH BITS
CA LDNDUMP1 # SET DNTMGOTO
TS DNTMGOTO # TO LOCATION FOR NEXT PASS
CA TIME1 # PLACE TIME1
XCH L # INTO L
CA DUMPLOC # AND ECADR OF THIS EBANK INTO A
TCF DNTMEXIT # SEND DUMPLOC AND TIME1
LDNDUMP ADRES DNDUMP
LDNDUMP1 ADRES DNDUMP1
DNDUMP CA TWO # INCREMENT ECADR IN DUMPLOC
ADS DUMPLOC # TO NEXT DP WORD TO BE
MASK LOW8 # DUMPED AND SAVE IT.
CCS A # IS THIS THE BEGINNING OF A NEW EBANK
TCF DNDUMP2 # NO -- THEN CONTINUE DUMPING
CA DUMPLOC # YES -- IS THIS THE END OF THE
MASK DUMPCNT # N-TH(N = 1 TO 4) COMPLETE ERASABLE
MASK PRIO34 # DUMP(BIT14 FOR 4, BIT13 FOR 2 OR BIT12
CCS A # FOR 1 COMPLETE ERASABLE DUMP(S)).
TCF DNPHASE1 # YES -- SEND DOWNLIST AGAIN
# AGAIN
TCF DNDUMPI +2 # NO -- GO BACK AND INITIALZE NEXT BANK
DNDUMP1 CA LDNDUMP # SET DNTMGOTO
TS DNTMGOTO # FOR WORDS 3 TO 256D OF CURRENT EBANK
DNDUMP2 CA DUMPLOC
TS EBANK # SET EBANK
MASK LOW8 # ISOLATE RELATIVE ADDRESS.
TS Q # (NOTE: MASK INSTRUCTION IS USED TO PICK
CA NEG0 # UP ERASABLE REGISTERS TO THAT EDITING
TS L # REGISTERS 20-23 WILL NOT BE ALTERED.)
INDEX Q
EBANK= 1400 # PICK UP LOW ORDER REGISTER OF PAIR
MASK 1401 # OF ERASABLE REGISTERS.
XCH L
INDEX Q # PICK UP HIGH ORDER REGISTER OF PAIR
MASK 1400 # OF ERASABLE REGISTERS.
EBANK= DNTMBUFF
TCF DNTMEXIT # GO SEND THEM
SENDID EXTEND # ** ENTRANCE USED BY ERASABLE DUMP PROG. **
QXCH DNTMGOTO # SET DNTMGOTO SO NEXT TIME PROG WILL GO
CAF ERASID # TO LOCATION FOLLOWING `TC SENDID'
TS L # ** ENTRANCE USED BY REGULAR DOWNLINK PG **
# Page 1084
TC WOZERO # GO SET WORD ORDER CODE TO ZERO
CAF LOWIDCOD # PLACE SPECIAL ID CODE INTO L
XCH L # AND ID BACK INTO A
TCF DNTMEXIT # SEND DOWNLIST ID CODE(S).
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
A little is an understatement. I am almost double your age and they are before my time.
My first computer was a 286 at age of 5 had a 20meg hard drive.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
mctrivia said...
A little is an understatement. I am almost double your age and they are before my time.
C'mon. The CD player is before his time!
For the record my first computer was a cheap Asian knock off of an Apple ][noparse][[/noparse]+. I still have the photocopy of the ][noparse][[/noparse]+ manual that came with it, including fold out schematic. I learned all I know about logic by analysing that schematic. (well, that and the schematic from the Disk ][noparse][[/noparse] and Z80 soft card)
Woz rates as one of my all time heroes.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
lt's not particularly silly, is it?
I was lucky. That was a huge hard drive and super fast computer back then. My mom is an accountant and worked for herself.
.by grade 4 I had 9600 baud internet connection which I was allowed to use 1 day a week. Was shared amongst all the staff at my moms work.
Much improvement over bbs
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
The university where I did some of my undergraduate work had a working Univac I (working as in generating income doing data processing). They had a transistorized Univac 1107 as well for students and faculty to use. One of the stories that illustrates the people savvy of the engineers there goes as follows:
The Univac 1107 had a "job" status monitor made with one of those really big display tubes (a "charactron") that produced a shaped beam of electrons for each character to be displayed (by running the beam through one of a bunch of shaped holes in a mask). Anyway, the display didn't have a buffered controller, so it had to be refreshed by the operating system from a buffer in main memory and that used about 10% of the CPU throughput when it was active. The display was useful, but expensive to use, so the engineers drilled a hole on the front panel and installed a small pushbutton with a stiff spring set up so you had to use a finger to push it. As long as you could hold down the pushbutton, the operating system would refresh the display. When you let go of the button, the operating system would stop refreshing. Most people could hold down the button for maybe 10 seconds before it became painful to do so, long enough to find what they wanted on the screen, but not enough to significantly impact the system throughput.
LEO (Lyons Electronic Office) built the world's first commercial computer, the LEO I. I spent a few months working at the LEO factory just as they were starting manufacture of the LEO III, which used transistors instead of valves. I designed an audible monitor circuit for it, which enabled an engineer to listen to the noise made by a CPU address line. It was basically a diode pump circuit and UJT made from a PNP and NPN transistor which divided the pulses by 100. I couldn't use an actual UJT as they had just been invented and we couldn't buy any. I also made a logic probe by modifying a "staticiser" board (I think it had two flip-flops on it with neons on the outputs). I then spent several weeks testing microcode units - lots of ferrite cores wired as pulse transformers and performing logic functions. I was supposed to reject them if they were faulty but I fixed any problems myself as I couldn't face all the paperwork involved with returning them to the factory where they were made.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
When I was in college, "Happy Hour' was 3 hours at night restricted to short programs (I think 1 minute) with fairly rapid turn around. On a good night with a little luck you could get 4 runs of your program. You loaded your program in a card reader and waited for one the the graduate students who worked as an operator at night to bring your printout to a set of shelves. A friend of mine was working on a nude line art program using a Texas Transportation Institute account. The night operators were waiting for him to finish so they could rerun the program to get their copies. The computer went down. But all jobs remained in the queue. The next day his program ran and offended the woman who was the day shift operator. He almost lost his part time job with Texas Transportation Institute. All of this was on an IBM 7090 or 7094. At the same time the university decided to sell its IBM 650. The only bidder was a scrap metal dealer.
My first computer (yes it was mine) was an 18 month old Singer (later ICL) System Ten mini-computer. I bought it in 1977. It was housed in my lined air-conditioned garage - and it was the length of the garage!!! 110KB core memory (maximum) 2.2uS cycle time, and 3 x 10MB Disk Drives (washing machine size) - maximum 100MB, 45mS average access time. Absolute minimum read time for the whole 10MB was 105 seconds using interlaced reading. I wrote this code to verify all sectors were readable - no processing was possible. Also had a 9 track tape drive and a VDU (video terminal). No printer as they cost a cheap $15,000 at the time.
I guess the cat's out of the bag - at least I have·hair but unfortunately it's grey.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
I once knew someone who programmed those Singer machines. He mentioned that the assembly language was so complex that programs were usually put through a SNOBOL program acting as a macro-processor.
leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
No Leon. True, it was assembler, but it also had an extremely powerful macro assembler. In fact there was no high level language really ever used. At one stage there was a Cobol release but it was about 1 cobol to 2.2 assembler instructions and didn't really work as the System Ten was an online transaction machine. In fact, there were only 16 instructions until the 1981 release.
In fact, it was very much like a prop.
It was a RISC type computer with A & B operands and no registers (except 3 index registers per partition). An instruction was 60 bits (10 characters of 6 bits which was a subset of ASCII). For instance, the multiply instruction was multiply A x B and place the result in B for a total length of A + B (no overflow possible) where A and B were each 1-10 digits. The divide was effectively the reverse of a multiply. The move character instruction could move from 1-100 bytes. The branch instructions were just like the prop, where the call variant stored the return address in memory just like the prop (and was often itself a jmp instruction). The computer was decimal, and so was memory addressing. It had up to 20 partitions (like cogs) each with their own memory, and a common memory shared between all partitions (like hub memory). The operating system lived in common memory and was accessed with a call instruction. The partitions each ran their own programs and were hardware time sliced. Each partition could have up to 10 peripherals (video terminals, teleprinter type terminals, printers, etc). It stood the test of time, being first released in 1969 as the Friden System Ten, then Singer, then ICL in 1976. There was one major redesign release in 1981, and it was manufactured until 1993 and maintained by ICL until 2000 (later in some countries). Marks & Spence had them connected to their POS terminals in the UK and BBC had them in AUstralia.
Now you can see why I like the prop
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
SNOBOL is a string-processing language that was often used as a macro-processor. It's quite a small interpreted language. I'm fairly sure that it was used to process the assembler macros for the Singer machines.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Comments
-Phil
A very neat summation. We have had a lot of posturing from Mark (in these forums) and Dr Jim (in his videos), but no real proof. We have heard their own personal theories and beliefs - but again, no proof.
To quote Carl Sagan (amongst others) "Extraordinary claims require extraordinary evidence". Mark and Dr Jim are the ones making the extraordinary claims here.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
Thanks for doing this research - the articles you have pointed us to contain more real information than anything Mark or Dr Jim have yet given us.
On the basis of this, and also of the patent extracts you quoted, one would have to conclude that Dr Jim's solution for machine intelligence is based on a fairly traditional neural network model - but renamed to appear sufficently novel to justify a patent application.
To those who have played with neural networks in the past, it will come as no surprise to find yet another person trying to make the leap from visual processing to cognitive processing - it's a well trodden path. All attempts to do this have so far been unsuccessful (at least to my knowledge - I'll admit to being a few decades out of date now, but then this approach was already a few decades old when Dr Jim published his article).
While Dr Jim is not the only one who still believes that this is a valid approach (I do myself) it simply doesn't work on the scale Dr Jim is trying to achieve it - a couple of Propeller chips and 16Mb or RAM is MANY orders of magnitude too small for the task.
Ross.
P.S. I agree about "knixels" - what a neat buzzword!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
If I were to look into AI (and who knows), I think I'd use my video card!
My video card has >100 cores, runs at about 1 GHz, and has tons of memory!
I think GPU with CUDA is the easy way to go for AI...
But, looks like someone's already doing something like this:
http://code.google.com/p/lissom/
http://www.nvidia.com/object/cuda_app_tesla.html
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
My Prop Info&Apps: ·http://www.rayslogic.com/propeller/propeller.htm
Nice link. NOW we're talking!
Given that there are ~100 billion neurons in the brain, you'd probably only need a million or so of those GPUs to be able to simulate it - if you could figure out how to connect them all together (there are 100 TRILLION synaptic connections between these 100 billion neurons).
But one single GPU might be enough to simulate an insect brain, given that they have only about 100,000 neurons.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
~100 billion = 25 * 2^32, a bit less than 2^37
1M gpu ~= 2^20 GPU's of current generation
--> in approx 20 years (if current trends continue) a GPU may be able to simulate a human brain
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full
Morpheus & Mem+ Advanced dual Propeller SBC with XMM and 256 Color VGA - PCB, kit, A&T available!
www.mikronauts.com - my site 6.250MHz custom Crystals for running Propellers at 100MHz
Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
And we'll know we're there when you try to install Windows 2029 on your computer, and it responds with "Not on your life, meatball!"
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
Post Edited (RossH) : 8/2/2009 9:25:41 AM GMT
Mainframes: Mallread said there were only a few (Univac, IBM & CDC) but in fact there were many... including ICL,·Sperry, Unisys (I think from a merger of Univac and ?),·Fujitsu. It was certainly my understanding that the DEC VAX was represented as a mainframe even though the mini I worked on sold against it. I also believe Burroughs made a mainframe.
And just to correct localrogers' interpretation about Windows NT, 2000 etc. They do not have their roots in mainframe operating systems. We are still waiting for Windows to remove the missunderstanding of mainframe operating systems used by·the original writers of Windows - and it's IBM predecessor variant - was it called OS/2???
Just to offer an olive branch... In the 90's it was said that based on current computer increases in performance, it would take >100 years to map the human genome. However, in 5-10 years it was mapped using a new algorithm. So indeed, mallred and Dr Jim could be onto something.
However, given their refusal to answer anything technical, I remain skeptical and believe they are "gouging" (pun intended)·the market with their overpriced product.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBladeProp, RamBlade, TwinBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index) ZiCog (Z80), MoCog (6809)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
I second what you posted above. Especially
best regards
Stefan
Post Edited (StefanL38) : 8/2/2009 3:07:21 PM GMT
You left out the Xerox Sigma 7 and the Honeywell CP-V. The 2 mainframes at the University I went to. (Yes it was THAT long ago
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBladeProp, RamBlade, TwinBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index) ZiCog (Z80), MoCog (6809)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Oddly enough, my first IBM-compatible PC was a Sperry. At 8MHz it was smokin'!
-Phil
Post Edited (Phil Pilgrim (PhiPi)) : 8/3/2009 4:47:46 AM GMT
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Another question... How old is Dr Jim ??? I figure the Apollo Telemetry code must have been written at least by 1967 and surely NASA would have required a university qualification plus some years experience. So my estimate is that Dr Jim must be 65++
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBladeProp, RamBlade, TwinBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index) ZiCog (Z80), MoCog (6809)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
<
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO
For example here is the download telemetry program as designed by Dr Jim I guess:
I like the comment about totally rewriting it to save 150 bytes of memory !
Now to get that AGC simulator compiled with Catalina and running on a Prop....
# Copyright: Public domain. # Filename: DOWN_TELEMETRY_PROGRAM.agc # Purpose: Part of the source code for Colossus, build 249. # It is part of the source code for the Command Module's (CM) # Apollo Guidance Computer (AGC), possibly for Apollo 8 and 9. # Assembler: yaYUL # Reference: Starts on p. 1075 of 1701.pdf. # Contact: Ron Burkey <info@sandroid.org>. # Website: www.ibiblio.org/apollo. # Mod history: 08/28/04 RSB. Adapted from corresponding Luminary 131 file. # # The contents of the "Colossus249" files, in general, are transcribed # from a scanned document obtained from MIT's website, # http://hrst.mit.edu/hrs/apollo/public/archive/1701.pdf. Notations on this # document read, in part: # # Assemble revision 249 of AGC program Colossus by NASA # 2021111-041. October 28, 1968. # # This AGC program shall also be referred to as # Colossus 1A # # Prepared by # Massachussets Institute of Technology # 75 Cambridge Parkway # Cambridge, Massachusetts # under NASA contract NAS 9-4065. # # Refer directly to the online document mentioned above for further information. # Please report any errors (relative to 1701.pdf) to info@sandroid.org. # # In some cases, where the source code for Luminary 131 overlaps that of # Colossus 249, this code is instead copied from the corresponding Luminary 131 # source file, and then is proofed to incorporate any changes. # Page 1075 # PROGRAM NAME -- DOWN TELEMETRY PROGRAM # MOD NO. -- 0 TO COMPLETELY REWRITE THE DOWN TELEMETRY PROGRAM AND DOWNLINK ERASABLE DUMP PROGRAM FOR THE # PURPOSE OF SAVING APPROXIMATELY 150 WORDS OF CORE STORAGE. # THIS CHANGE REQUIRES AN ENTIRELY NEW METHOD OF SPECIFYING DOWNLINK LISTS. REFER TO DOWNLINK # LISTS LOG SECTION FOR MORE DETAILS. HOWEVER THIS CHANGES WILL NOT AFFECT THE GROUND PROCESSING # OF DOWN TELEMETRY DATA. # MOD BY -- KILROY, SMITH, DEWITT # DATE -- 02 OCT 67 # AUTHORS -- KILROY, SMITH, DWWITT, DEWOLF, FAGIN # LOG SECTION -- DOWN-TELEMETRY PROGRAM # # FUNCTIONAL DESCRIPTION -- THIS ROUTINE IS INITIATED BY TELEMETRY END # PULSE FROM THE DOWNLINK TELEMETRY CONVERTER. THIS PULSE OCCURS # AT 50 TIMES PER SEC (EVERY 20 MS) THEREFORE DODOWNTM IS # EXECUTED AT THESE RATES. THIS ROUTINE SELECTS THE APPROPRIATE # AGC DATA TO BE TRANSMITTED DOWNLINK AND LOADS IT INTO OUTPUT # CHANNELS 34 AND 35. THE INFORMATION IS THEN GATED OUT FROM THE # LGC IN SERIAL FASHION. # # THIS PROGRAM IS CODED FOR A 2 SECOND DOWNLIST. SINCE DOWNRUPTS # OCCUR EVERY 20 MS AND 2 AGC COMPUTER WORDS CAN BE PLACED IN # CHANNELS 34 AND 35 DURING EACH DOWNRUPT THE PROGRAM IS CAPABLE # OF SENDING 200 AGC WORDS EVERY 2 SECONDS. # # CALLING SEQUENCE -- NONE # PROGRAM IS ENTERED VIA TCF DODOWNTM WHICH IS EXECUTED AS A # RESULT OF A DOWNRUPT. CONTROL IS RETURNED VIA TCF RESUME WHICH # IN EFFECT IS A RESUME. # # SUBROUTINES CALLED -- NONE # # NORMAL EXIT MODE -- TCF RESUME # # ALARM OR ABORT EXIT MODE -- NONE # # RESTART PROTECTION: # ON A FRESH START AND RESTART THE `STARTSUB' SUBROUTINE WILL INITIALIZE THE DOWNLIST POINTER (ACTUALLY # DNTMGOTO) TO THE BEGINNING OF THE CURRENT DOWNLIST (I.E., CURRENT CONTENTS OF DNLSTADR). THIS HAS THE # EFFECT OF IGNORING THE REMAINDER OF THE DOWNLIST WHICH THE DOWN-TELEMETRY PROGRAM WAS WORKING ON WHEN # THE RESTART (OR FRESH START) OCCURRED AND RESUME DOWN TELEMETRY FROM THE BEGINNING OF THE CURRENT # DOWNLIST. # # ALSO OF INTEREST IS THE FACT THAT ON A RESTART THE AGC WILL ZERO DOWNLINK CHANNELS 13, 34 AND 35. # # DOWNLINK LIST SELECTION: # THE APPROPRIATE DOWNLINK LISTS ARE SELECTED BY THE FOLLOWING: # 1. FRESH START # 2. V37EXXE WHERE XX = THE MAJOR MODE BEING SELECTED. # 3. UPDATE PROGRAM (P27) # 4. NON-V37 SELECTABLE TYPE PROGRAMS (E.G., AGS INITIALIZATION (SUNDANCE, LUMINARY) AND P61-P62 # TRANSITIONS (COLOSSUS) ETC.). # # DOWNLINK LIST RULES AND LIMITATIONS: # READ SECTION(S) WHICH FOLLOW `DEBRIS' WRITEUP. # # OUTPUT -- EVERY 2 SECONDS 100 DOUBLE PRECISION WORDS (I.E., 200 LGC # COMPUTER WORDS) ARE TRANSMITTED VIA DOWNLINK. # # ERASABLE INITIALIZATION REQUIRED -- NONE # `DNTMGOTO' AND `DNLSTADR' ARE INITIALIZED BY THE FRESH START PROGRAM. # # DEBRIS (ERASABLE LOCATIONS DESTROYED BY THIS PROGRAM) -- # LDATALST, DNTMBUFF TO DNTMBUFF +21D, TMINDEX, DNQ. # Page 1076 # (No source on this page of the original assembly listing.) # Page 1077 # DODOWNTM IS ENTERED EVERY 20 MS BY AN INTERRUPT TRIGGERED BY THE # RECEIPT OF AN ENDPULSE FROM THE SPACECRAFT TELEMETRY PROGRAMMER. # # NOTES REGARDING DOWNLINK LISTS ASSOCIATED WITH THIS PROGRAM: # 1. DOWNLISTS. DOWNLISTS MUST BE COMPILED IN THE SAME BANK AS THE # DOWN-TELEMETRY PROGRAM. THIS IS DONE FOR EASE OF CODING, FASTER # EXECUTION. # 2. EACH DOWNLINK LIST CONSISTES OF A CONTROL LIST AND A NUMBER OF # SUBLISTS. # 3. A SUBLIST REFERS TO A SNAPSHOT OR DATA COMMON TO THE SAME OR OTHER # DOWNLINK LISTS. ANY SUBLIST CONTAINING COMMON DATA NEEDS TO BE # CODED ONLY ONCE FOR THE APPLICABLE DOWNLINK LISTS. # 4. SNAPSHOT SUBLISTS REFER SPECIFICALLY TO HOMOGENEOUS DATA WHICH MUST BE # SAVED IN A BUFFER DURING ONE DOWNRUPT. # 5. THE 1DNADR FOR THE 1ST WORD OF SNAPSHOT DATA IS FOUND AT THE END # OF EACH SNAPSHOT SUBLIST, SINCE THE PROGRAM CODING SENDS THIS DP WORD # IMMEDIATELY AFTER STORING THE OTHERS IN THE SNAPSHOT BUFFER. # 6. ALL LISTS ARE COMBINATIONS OF CODED ERASABLE ADDRESS CONSTANTS # CREATED FOR THE DOWNLIST PROGRAM. # A. 1DNADR 1-WORD DOWNLIST ADDRESS. # SAME AS ECADR, BUT USED WHEN THE WORD ADDRESSED IS THE LEFT # HALF OF A DOUBLE-PRECISION WORD FOR DOWN TELEMETRY. # B. 2DNADR - 6DNADR N-WORD DOWNLIST ADDRESS, N = 2 - 6. # SAME AS 1DNADR, BUT WTIH THE 4 UNUSED BITS OF THE ECADR FORMAT # FILLED IN WITH 0001-0101. USED TO POINT TO A LIST OF N DOUBLE- # PRECISION WORDS, STORED CONSECUTIVELY, FOR DOWN TELEMETRY. # C. DNCHAN DOWNLIST CHANNEL ADDRESS. # SAME AS 1DNADR, BUT WITH PREFIX BITS 0111. USED TO POINT TO # A PAIR OF CHANNELS FOR DOWN TELEMETRY. # D. DNPTR DOWN-TELEMETRY SUBLIST POINTER. # SAME AS CAF BUT TAGGES AS A CONSTANT. USED IN CONTROL LIST TO POINT TO A SUBLIST. # CAUTION --- A DNPTR CANNOT BE USED IN A SUBLIST. # 7. THE WORD ORDER CODE IS SET TO ZERO AT THE BEGINNING OF EACH DOWNLIST (I.E., CONTROL LIST) AND WHEN # A `1DNADR TIME2' IS DETECTED IN THE CONTROL LIST (ONLY). # 8. IN THE SNAPSHOT SUBLIST ONLY, THE DNADR'S CANNOT POINT TO THE FIRST WORD OF ANY EBANK. # # DOWNLIST LIST RESTRICTIONS: # (THE FOLLOWING POINTS MAY BE LISTED ELSEWHERE BUT ARE LISTED HERE SO IT IS CLEAR THAT THESE THINGS CANNOT BE # DONE) # 1. SNAPSHOT DOWNLIST: # (A) CANNOT CONTAIN THE FOLLOWING ECADRS (I.E., 1DNADR'S): Q, 400, 1000, 1400, 2000, 2400, 3000, 3400. # (B) CAN CONTAIN ONLY 1DNADR'S # 2. ALL DOWNLINKED DATA (EXCEPT CHANNELS) IS PICKED UP BY A DCA SO DOWNLINK LISTS CANNOT CONTAIN THE # EQUIVALENT OF THE FOLLOWING ECADRS (I.E., IDNADRS): 377, 777, 1377, 1777, 2377, 2777, 3377, 3777. # (NOTE: TE TERM `EQUIVALENT' MEANT THAT THE IDNADR TO 6DNADR WILL BE PROCESSED LIKE 1 TO 6 ECADRS) # 3. CONTROL LISTS AND SUBLISTS CANNOT HAVE ENTRIES = OCTAL 00000 OR OCTAL 77777 # Page 1078 # 4. THE `1DNADR TIME2' WHICH WILL CAUSE THE DOWNLINT PROGRAM TO SET THE WORDER CODE TO 3 MUST APPEAR IN THE # CONTROL SECTION OF THE DOWNLIST. # 5. `DNCHAN 0' CANNOT BE USED. # 6. `DNPTR 0' CANNOT BE USED. # 7. DNPTR CANNOT APPEAR IN A SUBLIST. # # EBANK SETTINGS # IN THE PROCESS OF SETTING THE EBANK (WHEN PICKING UP DOWNLINK DATA) THE DOWN TELEMETRY PROGRAM PUTS # `GARBAGE' INTO BITS15-12 OF EBANK. HUGH BLAIR-SMITH WARNS US THAT BITS15-12 OF EBANK MAY BECOME # SIGNIFICANT SOMEDAY IN THE FUTURE. IF/WHEN THAT HAPPENS, THE PROGRAM SHOULD INSURE (BY MASKING ETC.) # THAT BITS 15-12 OF EBANK ARE ZERO. # # INITIALIZATION REQUIRED -- TO INTERRUPT CURRENT LIST AND START A NEW ONE. # 1. ADRES OF DOWNLINK LIST INTO DNLSTADR # 2. NEGONE INTO SUBLIST # 3. NEGONE INTO DNECADR BANK 22 SETLOC DOWNTELM BANK EBANK= DNTMBUFF COUNT 05/DPROG DODOWNTM TS BANKRUPT EXTEND QXCH QRUPT # SAVE Q CA BIT7 # SET WORD ORDER CODE TO 1. EXCEPTION: AT EXTEND # THE BEGINNING OF EACH LIST THE WORD WOR CHAN13 # CODE WILL BE SET BACK TO 0. TC DNTMGOTO # GOTO APPROPRIATE PHASE OF PROGRAM DNPHASE1 CA NEGONE # INITIALIZE ALL CONTROL WORDS TS SUBLIST # WORDS TO MINUS ONE TS DNECADR CA LDNPHAS2 # SET DNTMGOTO = 0 ALL SUSEQUENT DOWRUPTS TS DNTMGOTO # GO TO DNPHASE2 TCF NEWLIST DNPHASE2 CCS DNECADR # SENDING OF DATA IN PROGRESS DODNADR TC FETCH2WD # YES -- THEN FETCH THE NEXT 2 SP WORDS MINTIME2 -1DNADR TIME2 # NEGATIVE OF TIME2 1DNADR TCF +1 # (ECADR OF 3776 + 74001 = 77777) CCS SUBLIST # IS THE SUBLIST IN CONTROL # Page 1079 TCF NEXTINSL # YES DNADRDCR OCT 74001 # DNADR COUNT AND ECADR DECREMENTER CHKLIST CA CTLIST EXTEND BZMF NEWLIST # IT WILL BE NEGATIVE AT END OF LIST TCF NEXTINCL NEWLIST INDEX DNLSTCOD CA DNTABLE # INITIALIZE CTLIST WITH TS CTLIST # STARTING ADDRESS OF NEW LIST CS DNLSTCOD TCF SENDID +3 NEXTINCL INDEX CTLIST CA 0 CCS A INCR CTLIST # SET POINTER TO PICK UP NEXT CTLIST WORD TCF +4 # ON NEXT ENTRY TO PROG. (A SHOULD NOT =0) XCH CTLIST # SET CTLIST TO NEGATIVE AND PLACE(CODING) COM # UNCOMPLEMENTED DNADR INTO A. (FOR LA) XCH CTLIST # (ST IN ) +4 INCR A # (CTLIST) TS DNECADR # SAVE DNADR AD MINTIME2 # TEST FOR TIME2 (NEG. OF ECADR) CCS A TCF SETWO +1 # DON'T SET WORD ORDER CODE MINB1314 OCT 47777 # MINUS BIT 13 AND 14 (CAN'T GET HERE) TCF SETWO +1 # DON'T SET WORD ORDER CODE SETWO TC WOZERO # GO SET WORD ORDER CODE TO ZERO. +1 CA DNECADR # RELOAD A WITH THE DNADR. +2 AD MINB1314 # IS THIS A REGULAR DNADR? EXTEND BZMF FETCH2WD # YES. (A MUST NEVER BE ZERO) AD MINB12 # NO. IS IT A POINTER (DNPTR) OR A EXTEND # CHANNEL(DNCHAN) BZMF DODNPTR # IT'S A POINTER. (A MUST NEVER BE ZERO) DODNCHAN TC 6 # (EXECUTED AS EXTEND) IT'S A CHANNEL INDEX DNECADR INDEX 0 -4000 # (EXECUTED AS READ) TS L TC 6 # (EXECUTED AS EXTEND) INDEX DNECADR INDEX 0 -4001 # (EXECUTED AS READ) TS DNECADR # SET DNECADR CA NEGONE # TO MINUS XCH DNECADR # WHILE PRESERVING A. TCF DNTMEXIT # GO SEND CHANNELS WOZERO CS BIT7 EXTEND # Page 1080 WAND CHAN13 # SET WORD ORDER CODE TO ZERO TC Q # RETURN TO CALLER DODNPTR INDEX DNECADR # DNECADR CONTAINS ADRES OF SUBLIST 0 0 # CLEAR AND ADD LIST ENTRY INTO A. CCS A # IS THIS A SNAPSHOT SUBLIST CA DNECADR # NO, IT IS A REGULAR SUBLIST. TCF DOSUBLST # A MUST NOT BE ZERO. XCH DNECADR # YES. IT IS A SNAPSHOT SUBLIST. TS SUBLIST # C(DNECADR) INTO SUBLIST CAF ZERO # A INTO A XCH TMINDEX # (NOTE: TMINDEX = DNECADR) # THE FOLLOWING CODING (FROM SNAPLOOP TO SNAPEND) IS FOR THE PURPOSE OF TAKING A SNAPSHOT OF 12 DP REGISTERS. # THIS IS DONE BY SAVING 11 DP REGISTERS IN DNTMBUFF AND SENDING THE FIRST DP WORD IMMEDIATELY. # THE SNAPSHOT PROCESSING IS THE MOST TIME CONSUMING AND THEREFORE THE CODING AND LIST STRUCTURE WERE DESIGNED # TO MINIMIZE TIME. THE TIME OPTIMIZATION RESULTS IN RULES UNIQUE TO THE SNAPSHOT PORTION OF THE DOWNLIST. # THESE RULES ARE ...... # 1. ONLY 1DNADR'S CAN APPEAR IN THE SNAPSHOT SUBLIST # 2. THE 1DNADR'S CANNOT REFER TO THE FIRST LOCATION IN ANY BANK. SNAPLOOP TS EBANK # SET EBANK MASK LOW8 # ISOLATE RELATIVE ADDRESS EXTEND INDEX A EBANK= 1401 DCA 1401 # PICK UP 2 SNAPSHOT WORDS. EBANK= DNTMBUFF INDEX TMINDEX DXCH DNTMBUFF # STORE 2 SNAPSHOT WORDS IN BUFFER INCR TMINDEX # SET BUFFER INDEX FOR NEXT 2 WORDS. INCR TMINDEX SNAPAGN INCR SUBLIST # SET POINTER TO NEXT 2 WORDS OF SNAPSHOT INDEX SUBLIST 0 0 # = CA SSSS (SSSS = NEXT ENTRY IN SUBLIST) CCS A # TEST FOR LAST TWO WORDS OF SNAPSHOT. TCF SNAPLOOP # NOT LAST TWO. LDNPHAS2 GENADR DNPHASE2 TS SUBLIST # YES, LAST. SAVE A. CA NEGONE # SET DNECADR AND TS DNECADR # SUBLIST POINTERS XCH SUBLIST # TO NEGATIVE VALUES TS EBANK MASK LOW8 EXTEND INDEX A EBANK= 1401 # Page 1081 DCA 1401 # PICK UP FIRST 2 WORDS OF SNAPSHOT. EBANK= DNTMBUFF SNAPEND TCF DNTMEXIT # NOW TO SEND THEM. FETCH2WD CA DNECADR TS EBANK # SET EBANK MASK LOW8 # ISOLATE RELATIVE ADDRESS TS L CA DNADRDCR # DECREMENT COUNT AND ECADR ADS DNECADR EXTEND INDEX L EBANK= 1400 DCA 1400 # PICK UP 2 DATA WORDS EBANK= DNTMBUFF TCF DNTMEXIT # NOW GO SEND THEM. DOSUBLST TS SUBLIST # SET SUBLIST POINTER NEXTINSL INDEX SUBLIST 0 0 # = CA SSSS (SSSS = NEXT ENTRY IN SUBLIST) CCS A # IS IT THE END OF THE SUBLIST INCR SUBLIST # NO -- TCF +4 TS SUBLIST # SAVE A. CA NEGONE # SET SUBLIST TO MINUS XCH SUBLIST # RETRIEVE A. +4 INCR A TS DNECADR # SAVE DNADR TCF SETWO +2 # GO USE COMMON CODING (PROLEMS WOULD # OCCUR IF THE PROGRAM ENCOUNTERED A # DNPTR NOW) DNTMEXIT EXTEND # DOWN-TELEMETRY EXIT WRITE DNTM1 # TO SEND A + L TO CHANNELS 34 + 35 CA L # RESPECTIVELY TMEXITL EXTEND WRITE DNTM2 TMRESUME TCF RESUME # EXIT TELEMTRY PROGRAM VIA RESUME. MINB12 EQUALS -1/8 DNECADR EQUALS TMINDEX CTLIST EQUALS LDATALST SUBLIST EQUALS DNQ # Page 1082 # SUBROUTINE NAME -- DNDUMP # # FUNCTIONAL DESCRIPTION -- TO SEND (DUMP) ALL 8 BANKS OF ERASABLE STORAGE TWICE. BANKS ARE SENT ONE AT A TIME # EACH BANK IS PRECEDED BY AN ID WORD, SYNCH BITS, ECADR AND TIME1 FOLLOWED BY THE 256D WORDS OF EACH # EBANK. EBANKS ARE DUMPED IN ORDER (I.E., EBANK 0 FIRST, THEN EBANK1 ETC.) # # CALLING SEQUENCE -- THE GROUND OR ASTRONAUT BY KEYING V74E CAN INITIALIZE THE DUMP. # AFTER KEYING IN V74E THE CURRENT DOWNLIST WILL BE IMMEDIATELY TERMINATED AND THE DOWNLINK ERASABLE DUMP # WILL BEGIN. # # ONCE INITITIATED THE DOWNLINK ERASABLE DUMP CAN BE TERMINATED (AND INTERRUPTED DOWNLIST REINSTATED) ONLY # BY THE FOLLOWING: # # 1. A FRESH START # 2. COMPLETION OF ALL DOWNLINK DUMPS REQUESTED (ACCORDING TO BITS SET IN DUMPCNT). NOTE THAT DUMPCNT # CAN BE ALTERED BY A V21N01. # 3. AND INVOLUNTARILY BY A RESTART. # # NORMAL EXIT MODE -- TCF DNPHASE1 # # ALARM OR ABORT MODE -- NONE # # *SUBROUTINES CALLED -- NONE # # ERASABLE INITIALIZATION REQUIRED -- # DUMPCNT OCT 20000 IF 4 COMPLETE ERASABLE DUMPS ARE DESIRED # DUMPCNT OCT 10000 IF 2 COMPLETE ERASABLE DUMPS ARE DESIRED # DUMPCNT OCT 04000 IF 1 COMPLETE ERASABLE DUMP IS DESIRED # # DEBRIS -- DUMPLOC, DUMPSW, DNTMGOTO, EBANK, AND CENTRAL REGISTERS # # TIMING -- TIME (IN SECS) = ((NO.DUMPS)*(NO.EBANKS)*(WDSPEREBANK + NO.IDWDS)) / NO.WDSPERSEC # TIME (IN SECS) = ( 4 )*( 8 )*( 256 + 4 ) / 100 # THUS TIME (IN SECS TO SEND DUMP OF ERASABLE 4 TIMES VIA DOWNLINK) = 83.2 SECONDS # # STRUCTURE OF ONE EBANK AS IT IS SENT BY DOWNLINK PROGRAM -- # (REMINDER -- THIS ONLY DESCRIBES ONE OF THE 8 EBANKS X 4 (DUMPS) = 32 EBANKS WHICH WILL BE SENT BY DNDUMP) # # DOWNLIST W # WORD TAKEN FROM CONTENTS OF EXAMPLE O COMMENTS # 1 ERASID 0177X 0 DOWNLIST I.D. FOR DOWNLINK ERASABLE DUMP (X=7 CSM, 6 LM) # 2 LOWIDCOD 77340 1 DOWNLINK SYNCH BITS. (SAME ONE USED IN ALL OTHER DOWNLISTS) # 3 DUMPLOC 13400 1 (SEE NOTES ON DUMPLOC) 1 = 3RD ERAS DUMP, 3400=ECADR OF 5TH WD # 4 TIME1 14120 1 TIME IN CENTISECONDS # 5 FIRST WORD OF EBANK X 03400 1 IN THIS EXAMPLE THIS WORD = CONTENTS OF E7,1400 (ECADR 3400) # 6 2ND WORD OF EBANK X 00142 1 IN THIS EXAMPLE THIS WORD = CONTENTS OF E7,1401 (ECADR 3401) # 7. 3RD WORD OF EBANK X 00142 1 IN THIS EXAMPLE THIS WORD = CONTENTS OF E7,1402 (ECADR 3402) # . # . # . # 260D 256TH WORD OF EBANK X 03777 1 IN THIS EXAMPLE THIS WORD = CONTENTS OF E7,1777 (ECADR 3777) # # NOTE -- DUMPLOC CONTAINS THE COUNTER AND ECADR FOR EACH WORD BEING SENT. # THE BIT STRUCTURE OF DUMPLOC IS FOLLOW -- # X = NOT USED # X ABC EEE RRRRRRRR ABC = ERASABLE DUMP COUNTER (I.E. ABC = 0,1,2, OR 3 WHICH MEANS THAT # COMPLETE ERASABLE DUMP NUMBER 1,2,3, OR 4 RESPECTIVELY IS IN PROGRESS) # EEE = EBANK BITS # RRRRRRRR = RELATIVE ADDRESS WITHIN AN EBANK # Page 1083 DNDUMPI CA ZERO # INITIALIZE DOWNLINK TS DUMPLOC # ERASABLE DUMP +2 TC SENDID # GO SEND ID AND SYNCH BITS CA LDNDUMP1 # SET DNTMGOTO TS DNTMGOTO # TO LOCATION FOR NEXT PASS CA TIME1 # PLACE TIME1 XCH L # INTO L CA DUMPLOC # AND ECADR OF THIS EBANK INTO A TCF DNTMEXIT # SEND DUMPLOC AND TIME1 LDNDUMP ADRES DNDUMP LDNDUMP1 ADRES DNDUMP1 DNDUMP CA TWO # INCREMENT ECADR IN DUMPLOC ADS DUMPLOC # TO NEXT DP WORD TO BE MASK LOW8 # DUMPED AND SAVE IT. CCS A # IS THIS THE BEGINNING OF A NEW EBANK TCF DNDUMP2 # NO -- THEN CONTINUE DUMPING CA DUMPLOC # YES -- IS THIS THE END OF THE MASK DUMPCNT # N-TH(N = 1 TO 4) COMPLETE ERASABLE MASK PRIO34 # DUMP(BIT14 FOR 4, BIT13 FOR 2 OR BIT12 CCS A # FOR 1 COMPLETE ERASABLE DUMP(S)). TCF DNPHASE1 # YES -- SEND DOWNLIST AGAIN # AGAIN TCF DNDUMPI +2 # NO -- GO BACK AND INITIALZE NEXT BANK DNDUMP1 CA LDNDUMP # SET DNTMGOTO TS DNTMGOTO # FOR WORDS 3 TO 256D OF CURRENT EBANK DNDUMP2 CA DUMPLOC TS EBANK # SET EBANK MASK LOW8 # ISOLATE RELATIVE ADDRESS. TS Q # (NOTE: MASK INSTRUCTION IS USED TO PICK CA NEG0 # UP ERASABLE REGISTERS TO THAT EDITING TS L # REGISTERS 20-23 WILL NOT BE ALTERED.) INDEX Q EBANK= 1400 # PICK UP LOW ORDER REGISTER OF PAIR MASK 1401 # OF ERASABLE REGISTERS. XCH L INDEX Q # PICK UP HIGH ORDER REGISTER OF PAIR MASK 1400 # OF ERASABLE REGISTERS. EBANK= DNTMBUFF TCF DNTMEXIT # GO SEND THEM SENDID EXTEND # ** ENTRANCE USED BY ERASABLE DUMP PROG. ** QXCH DNTMGOTO # SET DNTMGOTO SO NEXT TIME PROG WILL GO CAF ERASID # TO LOCATION FOLLOWING `TC SENDID' TS L # ** ENTRANCE USED BY REGULAR DOWNLINK PG ** # Page 1084 TC WOZERO # GO SET WORD ORDER CODE TO ZERO CAF LOWIDCOD # PLACE SPECIAL ID CODE INTO L XCH L # AND ID BACK INTO A TCF DNTMEXIT # SEND DOWNLIST ID CODE(S).▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Toys are microcontroled.
Robots are microcontroled.
I am microcontroled.
If it's not Parallax then don't even bother. :-)
·
Mini-Din/PS2 connectors are for sale! 5 for $1! PM me if you wish to make an order.
Cheap·shipping unless specified!··········150 left!!··
My first computer was a 286 at age of 5 had a 20meg hard drive.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
C'mon. The CD player is before his time!
For the record my first computer was a cheap Asian knock off of an Apple ][noparse][[/noparse]+. I still have the photocopy of the ][noparse][[/noparse]+ manual that came with it, including fold out schematic. I learned all I know about logic by analysing that schematic. (well, that and the schematic from the Disk ][noparse][[/noparse] and Z80 soft card)
Woz rates as one of my all time heroes.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
lt's not particularly silly, is it?
.by grade 4 I had 9600 baud internet connection which I was allowed to use 1 day a week. Was shared amongst all the staff at my moms work.
Much improvement over bbs
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5
Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
BTW, Amdahl's campus stretched for a mile or more on either side of Central Expressway in Sunnyvale.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
--Steve
Propalyzer: Propeller PC Logic Analyzer
http://forums.parallax.com/showthread.php?p=788230
The Univac 1107 had a "job" status monitor made with one of those really big display tubes (a "charactron") that produced a shaped beam of electrons for each character to be displayed (by running the beam through one of a bunch of shaped holes in a mask). Anyway, the display didn't have a buffered controller, so it had to be refreshed by the operating system from a buffer in main memory and that used about 10% of the CPU throughput when it was active. The display was useful, but expensive to use, so the engineers drilled a hole on the front panel and installed a small pushbutton with a stiff spring set up so you had to use a finger to push it. As long as you could hold down the pushbutton, the operating system would refresh the display. When you let go of the button, the operating system would stop refreshing. Most people could hold down the button for maybe 10 seconds before it became painful to do so, long enough to find what they wanted on the screen, but not enough to significantly impact the system throughput.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/3/2009 4:07:22 PM GMT
John Abshier
I guess the cat's out of the bag - at least I have·hair but unfortunately it's grey.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBladeProp, RamBlade, TwinBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index) ZiCog (Z80), MoCog (6809)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
In fact, it was very much like a prop.
It was a RISC type computer with A & B operands and no registers (except 3 index registers per partition). An instruction was 60 bits (10 characters of 6 bits which was a subset of ASCII). For instance, the multiply instruction was multiply A x B and place the result in B for a total length of A + B (no overflow possible) where A and B were each 1-10 digits. The divide was effectively the reverse of a multiply. The move character instruction could move from 1-100 bytes. The branch instructions were just like the prop, where the call variant stored the return address in memory just like the prop (and was often itself a jmp instruction). The computer was decimal, and so was memory addressing. It had up to 20 partitions (like cogs) each with their own memory, and a common memory shared between all partitions (like hub memory). The operating system lived in common memory and was accessed with a call instruction. The partitions each ran their own programs and were hardware time sliced. Each partition could have up to 10 peripherals (video terminals, teleprinter type terminals, printers, etc). It stood the test of time, being first released in 1969 as the Friden System Ten, then Singer, then ICL in 1976. There was one major redesign release in 1981, and it was manufactured until 1993 and maintained by ICL until 2000 (later in some countries). Marks & Spence had them connected to their POS terminals in the UK and BBC had them in AUstralia.
Now you can see why I like the prop
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBladeProp, RamBlade, TwinBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index) ZiCog (Z80), MoCog (6809)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle