Digital Equipment Corporation (DEC) PDP-1 Computer System Information

Source Documents

The following records are a very incomplete listing selected from the available correspondence, letters, memoranda and reports published by the Computer History Museum. Refer to their Collections and contents for the complete records. The references below may only contain the title or a brief summary of a Memoranda's contents. A subset will contain an extract of a memoranda, and rarely, some will have nearly most of a record. Do not rely on any of the information below, always refer to the source material published by the Computer History Museum. The listings below are only a fraction of the material published by the Computer History Museum. Any copyright of the Computer History Museum remains with the Computer Museum.

DEC records are not organised as a single stream organised into a simple timeline but consists of multiple batches of records in multiple, ovelapping, forward or reversed timelines. There can be multiple copies of a record in different batches. A single batch can appear in different subsets of the DEC Collection. The following was selected to trace the history and produce a timeline for the PDP-1. Other records are included purely out of curiosity, or a whim. You have been warned.

1960

Mr. Charlton Walter of Cambridge Research Center telephoned today to follow up earlier discussion with some people who apparently work for him at CRC. They are interested in obtaining a new computer for systems simulation work. I promised that we would send him a copy of the PDP-3 specification as soon as it is available in mimeograph form and that I would then call him for a visit at our place.

His address is Mr, Charlton Walter, Cambridge Research Center, CRRBI Hanscom Field, Bedford Massachusetts.

Date February 3, 1960

From R. L. Best

Supercedes Memo Dated October 28, 1959
* New Numbers Added.

Date March 18, 1960

To Ben Gurley/Harlan Anderson

From Ken Olsen

Our first advertising of PDP-1 appeals to the people already in the computer business and they tend to mislead us as to where the market possibilities are for this machine. Maybe we shouldn't listen to hot programmers but should go out aggressively after the fields which we know are there and not being touched by other people. The examples which have come up without looking include machine tool positioning tape production, mailing list maintenance, format control, and I am sure there are a lot of others.

We tend to lose interest in these because they always need something special but when you look at what Bolt, Beranek & Newman want, it's not going to be an off-the-shelf item either. When you consider what they're asking for, for the money, it might be an awful lot more work than what we do per dollar on memory testers.

Date March 18, 1960

To Ben Gurley/Harlan Anderson

From Ken Olsen

<DARC

I called Dan Geisler at M.I.T. to get his opinions as to what an average response computer should be. He feels it is desirable to take 10,000 samples when one is working with human beings because the background signals are often 40 to 60 microvolts and the desired signals are often as low as 2 microvolts. When people are working with animals they can slip off the skulls and get signals that will look good after a 100 samples because they can get much closer to the cortex.

He felt that 256 points on the curve is very nice, but 128 might do and he suggested that one might even get by with 64 points. They use an Offner differential input amplifier with a gain of one to ten million. He uses a frequency response of 8 to 600 cycles but some people like to go lower and a few even like to go down to DC. As one gets closer to the source of the signals, the pulses are narrower and get down to about a millisecond in width which would need an ampolifier response of about 5000 cycles.

On May 10 and 12 there is going to be a conference at M.I.T. on computers and biophysics. This is only for invited people and we would not normally be invited but Dan will see if he can get us on the list. We definitely should have some propaganda on our equipment then and it would be nice if we had a brochure written on DARC. It would be nice to have the computer installed nearby where people could come visit it. This show conflicts with the shows in California and we can't have the computer in both places. If the computer is tied up with work at Maynard, we may not want to ship it to California anyway.

Date April 21, 1960

To Ben Gurley/Harlan Anderson/Stan Olsen

From Ken Olsen

Ted Johnson says that JPL is very much interested in a computer like PDP-1 for format control. The competition is the small CDC computer. They are expert programmers and not at all disturbed by the fact that we are limited in our programs. They simply want to do a simple fixed program. They are going to be at the I.R.E. Show and so we should keep an eye out for them.

Date April 21, 1960

To Ben Gurley/Harlan Anderson/Stan Olsen

From Ken Olsen

This is the second in a series of programs for the testing and use of the PDP-1. It comes as the next logical step, in preparing tapes in the binary read in mode tape format, after the Test Word to Binary Tape Converter Program (Memorandum 1061).*

The purpose of this program is to convert a flexowriter octal tape into a tape in the binary read in mode tape format. This program makes it possible for the computer to recognize the octal characters punched on a tape in the alpha numeric mode and to punch out a tape in the binary read in mode.

Memorandum 1060, Preliminary PDP-l Order Code, expresses all machine instructions in terms of octal numbers, therefore, this program only has to recognize flexowriter characters 0 through 7. For special use this program also recognizes the letters D and J and converts them into the DIO and JMP instructions on the binary tape.

In order to get this program on a tape in the binary read in mode, the following sequence of steps should be followed. First read the binary tape of the "Test Word to Binary Tape Converter Program" into the computer. Then, using the operator instructions contained in Memorandum 1061, place the "Flexowriter Octal to Binary Tape Converter Program", which follows in the computer. Two things will have been accomplished at this point, the "Flexowriter Octal to Binary Tape Converter Program" will be in the computer and this program will also have been punched out on a tape in the binary read in mode tape format.

By utilizing this program all subsequent pregram can be converted to flexowriter tapes in the binary read in mode, by typing them out on the flexowriter** and then feeding this alpha numeric tape into the computer. The following operator instructions are needed to use this program.

_________________________________

OPERATOR INSTRUCTIONS

After the "Flexowriter Octal to Binary Tape Converter Program" has been placed in the machine, follow these instructions to convert flexo-writer octal tapes to tapes in the binary read in mode tape format.

1. Set Test Address to 0040.
2. Push start button on the console. (This will cause 6 lines of tape to be punched for each instruction or word on the flexowriter octal tape. The first three will be DIO-N and the next three the word or instruction).
3. The machine will halt when the last instruction or word has been punched, completing the tape in the read in mode tape format.

FLEXOWRITER OCTAL TO BINARY TAPE CONVERTER PROGRAM

Appendix

Read in Binary Mode Tape Format

Read in binary mode tape format refers to the layout a flexowriter tape must have in order to be read into the computer using the "Tape Reader Binary Mode."

Flexowriter tape has eight channels in which holes may be punched. A hole puhched in a tape refers to a binary "one", and no hole refers to a binary "zero", when the tape is in the binary mode. In the PDP-1, one computer word is 18 bits long and appears as three ccusecutive lines on a tape punched in the binary mode format. A binary tape has the eighth channel punched in each line to show that it is in the binary mode.

A piece of tape, with the first few instructions of the Test Word to Binary Tape Converter Program, is drawn to show the binary mode tape format. 1*s represent holes in the tape and 0's the blank spaces.

A fellow by the name of Jim Fitzgerald, of Lincoln Lab, Group 22, called teday and they are looking for a cathode ray tube system to tie into their 7090 computer and wondered if we had anything for them. They probably want something like what was on 709 including a large scope for viewing and a small scope for camera work.

I told him that you would call and tell him more about what we are doing in this area. However, my own feeling is we should not waste much time on this since it would probably be stopped by Lincoln Lab Purchasing Department anyway since it is in the same category as the tape buffer unit that we talked about. Give him a call and invite him out and tell him what we are doing but other than that I wouldn't waste any time on it.

Date April 20, 1960

To Loren Prentice/Ben Gurley

From Ken Olsen

I iike the package we developed for the new photoelectric tape reader, and I think we should follow the same general design for our final packaging. I think the final package should contain photoelectric tape reader, paper tape punch, a reeler, and a tape feeder. Underneath the 7 inch pane] that holds the tape reader, I think there should be another 7 or 5&frac4; inch panel that holds the reeler in tape feeder. This should be the same general design as used om the Bendix computer but we can probably make a neater looking one. This panel could also hold the switches. The reeler might be a low inertia reel with a low inertia torque motor or it might be servo controlled like the reeler on the Ferranti. The tape feeder should be like the one we used in TX-0, TX-2 and Whirlwind. The tape punch, which should be on the side of the cabinet, could be the separate Ferranti punch.

I think the cabinet should be welded back just like the console computer is with just a cutout on the side for the punch. Rubber feet should be ahout ¾ inch or one inch high and mounted in the bottom so that air is blown out of ome side of the bottom and sucked in the other.

Date April 25, 1960

To Dick Best

Power Supply for PDP-1 Display

From Ken Olsen

Here's a suggestion as to how to we should make the final power supply for the PDP-1 display. Instead of two 749 panels we can have a single power supply box with four standard 110 velt solar transformers each with their own bridge rectifier and filter. With these in series we get almost 600 wolts. If we make all the transformers or one of them the one amp size, we can put an isolation transformer across one of them with a separate bridge rectifier that would give us the -150 volts we need for the display. We can stack electrolytic capacitors until we get enough filtering.

Date May 16, 1960

To Stan Olsen

ITT Visit to DEC

From Jack Brown

Harvey Selz from I.T.T. Labs will arrive tomorrow morning (May 17) around 9 er 10 o'clock to discuss his applications for our equipment. He will contaet you on arrival. I shall be baek from my plant visit around 1 o'clock to join you.

Date May 24, 1960

To Harlan Anderson/Ben Gurley

ITT Visit to DEC

From Ken Olsen

Mr. Kurkjian from I.T.&T. ealled and would like to visit us next Wednesday, June 1, at 10 o'clock to discuss our computer. They are interested in using it, and he's talking about other means of cooperation. He used to work on the SAGE eystem and was one of the very pleasant people at IBM. He left IBM a year and a half ago and has been working on various I.T.&T. divisions, and is now baok at I.T.&T. proper for about the last six weeks. Randy DeHart will be with Kurkjian.

Date May 24, 1960

To H. Anderson; Ben Gurley

From Ted Johnson

Date June 30, 1960

From Ken Olsen

I made an appointment for July 6 at Bolt, Beranwk & Newman with Dan Geisler and Ed Fredkin to discuss the electroencepalographic experiments we'd like to do this summer. This might be a good chance for Andy and myself to go down. It might be necessary for Ben to go down to show the method for hooking in the analog-to-digital converter.

Date July 28, 1960

Subject Elextroencephalogicaal Experiment with DEC Analog-to-Digital Converter at Bolt, Beranek & Newman

To Harlan Anderson/Benjamin Gurley

From Ken Olsen

Dan Geisler called today to say that he is all set up with his experiments at Bolt, Beranek & Newman on averaging Elextroencephalogicaal responses. Next Tuesday evening, which would be August 2, he plans to record responses from his subject. We promised that we would set up our analog-digital converter for them before they needed it. I told them that on Tuesday one of our people, probably Ed Harwood, would go down and show them how to use it. When we go, or if we don't go, we should call Ed Fredkin because he is the one who will be running the machine. Dan Geisler will be working in the recording laboratory and won't get involved in the computer during the day.

If we are unable to make this trip, please let me know.

I think it would be good to bring the land camera along so that Dan Geisler can photograph the subject he is getting responses from. This is important for the brochure we hope to of his work.

Date August 2, 1960

Subject Expanded PDP-3 Design Request

To Harlan Anderson/Benjamin Gurley

From Ken Olsen

Ed Fredkin called today and asked us to bid on a very large memory. This is for a company comfidential project they are working on which would give them one of the the largest computing facilities in the world. I think they have read the Stretch literature and think they can make one that is better for a fraction of the money,

They'll need a computer like our PDP-3 but with extra features which might involve complete redesign. They'll need two limit registers which will involve complete redesign. They'll need two limit registers which will trap the program if they try to use memory outside memory outside the limits. They also need a relocation register which adds the number to all addresses so the program can be moved around easily. They want floating point arithmetic, parity check, real in-out registers, and a few instructions. They'd also like to tie into phone lines.

The biggest part of the macchine is the core memory. They would like to have four quotes from us and it should be in the form of a letter before August 8. They would like bids on 130 and 260, 36 bit words of memory and the most efficient size modules. They would like the same size memories bid in 4000 word modules which are completely independent with their own timing circuits and power supplies. They would be willing to take 5 microsecond memories but would just love 2 microsecond memories. Ben Gurley thinks we could approach 2 microseconds if we don't saturate our transistors and if we used somewhat faster transistors. If we outlawed using the same word of memory in succession we probably would not be bothered with several problems.

He feels quite sure that they will order this system from somebody and they will contact Ceramics, RCA, CDC, and Telemeter Magnetics.

They expect to get the order within two or three months and would like to have hardware in about a year. I mentioned that we couldn't rent a machine like this and he felt they probably could buy it. Of course, an ordinary lease would be the same as buying as far as we're concerned.

Mr. Norman Raver of I.E.C. called today, August 9, at 1000 A.M. to ask us to visit them to discuss the possibility of supplying a computer with display capability for a project they are bidding to the Defense Communication Agency.

They want the information in a day or two because they have to turn the proposal in in four weeks, spend the next week and a half in writing the proposal, and two weeks for management approval and art work.

I told him I would call him back and let him know when I could go down to see him. They are in Paramus, New Jersey, which is about fifteen miles north of Newark. He suggests we take Route 4 from the George Washington Bridge to Route 17, where you turn north. I.E.C. is about a mile from the intersection, but on the wrong side of the parkway so one has to continue to the next intersection and make a U turn and backtrack about a half mile.

His number is COfax 2-6800, extension 519 or 227. Dick Palmer, who is formerly of IBM and Lincoln Laboratory, is working with him.

They would like an 8 ft. by 15 ft. display which shows the status of all communication lines. Geography doesn't change very often but it has to be possible to modify it. Numerical displays may be all that they need.

Date August 15, 1961

To Bolt, Beranek and Newman

From Ken Olsen

in a conversation yesterday with Deve Mittelman, I was told ahout several pending things with their machine in Cambridge and also their forthcoming machine in Los Angeles. Some of these are in writing at the moment; others are likely to be put in soon.

Cambridgs Machine

They have sent us« letter authorizing us to proceed with multiply and divide, simple tape control and one tape unit, extra core module with jump fletd. This letter arrived todey, and we should start work immediately. They did not include at this time the 18M typewriters that they hed often spoken ahout.

Los Angeles Machine

As I understand it from Dave Mittelman, they plan to remove from the optional equipment the on or offline typewriter punch station that is now a part of the order. This is not yet confirmed in writing, but will be shortly.

The lease terms that we sent down to them to cover the Los Angeles machine are basically acceptable with minor modifications. They intend to retun a copy to us with their suggested corrections on it, but it was not in with the letter this morning, even though their letter mentioned it. Dick Mills will follow through on this.

The optional items thet they heve ordered for the Cambridge machine they went to have on a lease also. This would be the same type lease as covers the Los Angeles machine, and Dick Mills will take of this also. Ken Olsen can sign the lease, assuming that the changes are satisfectory, next week.

August 16, 1960

Cambridge Research Laboratories
Astro Surveillance Science Laboratory
United States Air Force L. G. Hascom Field
Bedford, Massachusetts

Gentlemen:

Enclosed is an unsolicited proposal from Digital Equipment Corporation to develop a Dynamic Simulator for the Air Force. We would appreciate your evaluation of this proposal and be pleased to discuss it with you at your conveniance.

In connection with submitting this proposal, DEC has filled an ARDC Form #91 entitled "Policy Agreemeent for the Evaluation of Articles or Disclosures".

Sincerely

Harlan E. Anderson
Vice President

Date August 26, 1960

Subject PDP Nomenclature

To Harlan Anderson/Ken Olsen

From T. Johnson

Might be worth while to note that there is a PDP (Peripheral Data Processing) Division of American Electronics, Inc., in Brooklyn, New York. pdp/Data Intergrators, for example, is the way they would advertise their data intergrators.

Why not select a catchy computer trade-name now? Maybe DECOMP-1, etc.

September 8, 1960

Mr. Floyd Cook
Air Force Cambridge Ressearch Center
Electronic Research Directorate (CRRB)
Laurence G. Hascom Field
Bedford, Massachusetts

Dear Mr. Cook:

Enclosed is a copy of the programming manual for the PDP-1 computer that I mentioned to you on the telephone yesterday. Should you wish additionall information, please feel free to contact me.

Sincerely,

Harlan E. Anderson

September 14, 1960

Mr. D. L. Mittelman, Comptroller
Bolt Beranek and Newman, Inc.
50 Moulton Street
Cambridge 38, Massachusetts

Dear Dave:

I am enclosing with this letter one fully executed copy of the lease agreement for your files. The Programmed Data Processor - 1 covered by this lease is in the final testing phase, and we look forward to delivering it to you in the near future. At that time, we will go through an acceptance testing procedure with whomever you designate to represent Bolt Beranek & Newman. When these acceptance tests have been completed to the satisfaction of both parties, we would like a letter to that effect and you may start using the machine immediately thereafter and the lease will immediately become effective.

Please let us know if there is anything else that we can do to be of service to you.

Sincerely

Harlan E. Anderson
Vice President

Mr. Floyd Cook, of the Computer Lab, telephoned today and indicated that they are becoming more interested in our PDP-3 computer and that they would like to buy one if their funding for such a project comes through. They are optimistic about the funding and expect that it might be hers within a month or so. They would like to use it in a compatible way with their existing equipment which means that they would like to use 7-hole paper tape going into the machine. They would also like to have three photoelectric tape readers reading into the computer and probably three of the Soroban high speed punches. Their use is totally independent of Charlton Walter of Cambridge Research Center. They are interested in the computer art, experimenting with in-out devices, etc. Charlton Walter, on the other hand, is interested in using the computer as a tool in carrying out research work.

There is nothing to be done on this particular project at the moment.

Mr. Robert Dunn from ITT Laboratories, Department 052, 492 River Road, Nutley, New Jersey called yesterday, September 26, 1960. He wants a letter from us stating firm price and how long this price will be in effect for a PDP-1 with 4096 words of memory. I think this is in connection with the Ascension Island system which probably would be the one which would have the decision in November, contract in January, and their system must be done in October. This is true of General Electric in Syracuse, the Hughes in Fullerton, California, and the ITT Laboratories.

I spoke with Charlton Walter this morning and he told me that the proposal that we submitted to CRC has apparently been approved by all the local people and has been forwarded for higher headquarters approval at Andrews Air Force Base outside Washington, D. C. The only slightly negative reaction so far has been that the justification for doing this on a sole source should be a little more complete than what is now in the document that Cambridge Research Center is submitting. On the whole, his optimism continues to be very high on the anticipated outcome.

I spoke with Charlton Walter this morning and he told me that the proposal that we submitted to CRC has apparently been approved by all the local people and has been forwarded for higher headquarters approval at Andrews Air Force Base outside Washington, D. C. The only slightly negative reaction so far has been that the justification for doing this on a sole source should be a little more complete than what is now in the document that Cambridge Research Center is submitting. On the whole, his optimism continues to be very high on the anticipated outcome.

I spoke with Charlton Walter this morning and he told me that the proposal that we submitted to CRC has apparently been approved by all the local people and has been forwarded for higher headquarters approval at Andrews Air Force Base outside Washington, D. C. The only slightly negative reaction so far has been that the justification for doing this on a sole source should be a little more complete than what is now in the document that Cambridge Research Center is submitting. On the whole, his optimism continues to be very high on the anticipated outcome.

While in Washington on Monday, Sept. 26, 1960, I visited Mr. James C. Clarken at the U.S. Navy Bureau of Weapons. His address is Constitution Avenue at 19th St., N.W., Room 2W60, and his telephone number is OXford 6-5536. Mr. Clarken is handling the Bureau of Weapons approval for the Naval Ordnance Test Station computer. As far as he is concerned, there is only one more signature required within the Bureau of Weapons, and then the procurement will be turned over to the purchasing group. He expressed the opinion that the justification of sole source procurement was adequate as submitted by the USNOTS, and it was his conclusion that the procurement would be handled out of the Navy Purchasing Office from Los Angeles. This is contrary to what Maudlin had been told since Clarken was originally of the opinion that DEC had a GSA schedule. This is a General Service Administration schedule for people who sell large numbers of the same item to the Government. Since we do not have such a schedule, there is no need for the purchasing to be done through Washington.

As soon as this Bureau of Weapons approval is obtained (and this is expected any day) he will notify Maudlin and tell him to submit an up-to-date requisition to the Navy Purchasing Office in L.A. Any technical changes that we want to make in this requisition should be done at this time.

On Friday, September 30, I learned from Ted Johnson that the internal paper work for the USNOTS computer (PDP-3) had been approved by the Bureau of Weapons in Washington, D. C. This means that arriving at a contract is now a local matter for the Naval Ordnance Test Station in Pasadena in conjunction with the Navy Purchasing Office in Los Angeles. The next thing that is needed at this stage is a firm definition of what each item of equipment will do. The existing document on the standard machine is probably adequate for the standard machine. However, a more accurate description of some of the optional items must be arrived at before the Navy Purchasing Office can issue a contract.

I telephoned Lloyd Maudlin in Pasadena on Friday and Monday to set up a meeting for next week to try and pin down many of these items. Ben Gurley will be going out Tuesday, October 11, to be in Los Angeles Tuesday afternoon. I probably will be going Wednesday. Ben's objectives will be to tie down with a document and a schedule all of the technical matters. My objective will be to start negotiations with the Navy Purchasing Office including discussion of a first draft of the lease and also a Letter of Intent and any other business matters that might be required.

Lloyd Maudlin indicated that the Navy Purchasing people including Commander Hicks were very interested in meeting with us soon and the timing of next week appears just right. USNOTS appear to have a good working relationship with their local purchaing office, and this should be helpful. The estimate from the Purchasing Office is that it will take six weeks to a firm contract. A formal approval by the Secretary of the Navy may have to be obtained, but this is regarded as a formality which will probably be done by telephone.

DEC should prepare the first draft of the lease. We have previously had a copy of the lease which I think we showed to the Navy about six months ago and the present draft should be patterned after that.

The Navy wants us to install the analog-to-digital and digital-to-analog converter as part of the system, and they want the Packard-Bell machine which does the conversion at the rate of four microseconds per bit, and has an accuracy of 11 bits plus sign. It appears that we would be released from any technical responsibility if we merely included the unit that they recommend, If we wish to do anything different than this, we must formulate our recommendations prior to making the trip to California.

Probably the area where we have the least amount of information available at the present time but yet is very important to them is the programs that will be availshie and when they will be available. My recommendation here is that we should prepare a technical description of the DECAL progam for the PDP-1 that Bennett is writing, and then say that the PDP-3 program will be essentially the same. A copy of the memorandum dealing with magnetic tape instructions should be included also. In addition to this, we should give them dates by which they can expect to have a specification for the compiling program and any other pertinent dates.

We should also discuss while in California the combination programmer maintenance who will be handled under a separate contract from the main computer contract.

Maudlin hopes to get a letter of intent processed very shortly, which would cary a maximunm commitment of $15,000 to $20,000 and presumably would cover any work that we might do prior to the signing of a formal contract. He indicated on Monday that their Purchasing Office processes these letters of intent just as if they were a regular purchase order, and therefore, it does take one or two weeks to complete the process. This conceiveably could be done by the time go to California. This would allow us to get a head start of approximately 4 to 5 weeks which is worth while from their standpoint.

At the present time, they feel that they will not ask us to put card machines on the computer, but instead will rely on off-line equipment to make card-to-tape and tape-to-card conversions. However, they do still want to have a line printer directly connected to the computer, since they've been led to believe by Anelex and others that an off-line line printer becomes a significant expenditure of money.

This project has reached the status where it requires some concentrated effort on the part of Ben Gurley and myself in order to make it turn into a fruitful contract.

Ed Fredkin from BBN called to tell us that they are ready to go ahead with the drum for their PDP-1. They would like as many fields as possible and would like to be able to interchange memory and drum as quickly as possible. I told him that we requested the loan of the drum from Bryant so that we will start work on the drum immediately. He is comming out to visit us next Thursday morning, October 13, to discuss the logical details of the problem.

The 3600 r.p.m. drum has a cycle time of 4 microseconds, which seems too fast for our memory; and the 1800 r.p.m. drum has a cycle time of 8 microseconds, which slower than necessary. Our tentative plans are to buy an 1900 r.p.m. drum, but to be sure that we have the ability to drive it from an 85 cycle ioscillator which would give us about a 5½ microsecond time.

October 12, 1960

Mr. C. E. Jones
ITT Laboratories
Department 02312
492 River Road
Nutley, New Jersey

Dear Mr. Jones:

This will confirm a telephone conversation between Mr. Dave Perry and myself regarding certain optional units for use with the PDP-1 computer. An extra 4,096 word module of magnetic core memory costs $40,000 each. The Sequence Break System not including buffers associated with the in-out devices costs $12,000. This utilizes the the Ampex FR-400 tape in a compatible mode with IBM Type 729I tape units. In addition, a tape control is capable of working with up to 64 tape units.

I will enclose with this letter additional copies of the PDP-1 Programming Manual as well as technical information describing the mode of operation with magnetic tapes.

Sincerely,

Harlan E. Anderson

Date October 17, 1960

Subject Bolt, Beranek &Newman Visit 10/13

To Harlan Anderson/Benjamin Gurley/Gordon Bell

From Ken Olsen

Ed Fredkin, Licklider, and a couple other people from Bolt, Beranek &Newman came out Thursday, October 13, to discuss the addition of the drum to their PDP-1. They first looked over the computer and were very critical. Some thought the table should be higher, some thought it should be lower. They didn't like the way the lights were located and eavh one was sure that things were different than what he understood or what he thought they should have been. This upset me so much I almost told them to take their purchase order and go home. but I think I maintained a civil tongue even though my color might have changred during this session.

When we talked about adding a drum to this system, I told them it would have to be on much more formal arrangements so that there would be complete understanding as to what we were delivering.

They would like this drum as soon as possible and, therefore, would not like to stretch the art. They are mostly interested in proving the feasibility of multiple inputs to the drum.

We concluded that it would be necessary to have an 1800 r.p.m. drum. This gives about 8 microseconds per cycle which appears slow, but we may need it in order to synchronize comfortably with the memory timing. We can always drive the motors from a higher frequenct source. It seems that it would be best to have 4036 words around, which would be in their ten inch drum. The largest 10 inch drum has 420 tracks which would give 22 fields at 19 bits, or 20 or 21 if we would like to have a large number of spare tracks.

They would like to have parity on the drum and the only way this seems possible is to add parity to memory and assume any drum failures would be caught as the information goes in and out to memory.

The sequence they are interested in would have an executive field which would take part in every interchange. This would simplify the cost of our problem because the executive field would be completely isolated from the rest. When a new customer wants to cal;l on the computer, the executive field is read in in the executive program, decides how and what the next customer will take over, then this executive field is swapperd with the new customer's field.

According to Bryant literature, an 1800 r.p.m. drum should have an average bearing life of nine and one-half years. We should get about 30 millivolts signal out from the head on the return to saturation recording. Optimum write current for half the coil appears to be between 100 and 150 ma.

I can think of four types of recording. The early IBM method was return to ZERO where a positive flux pulse was a ONE, and a negative flux pulse was a ZERO. The output voltage for a ONE looked like a sinewave going negative first. If these signals were integrated, a ONE would be a positive pulse and ZERO would be a negative pulse,

The return to bias method is somewhat the same, but one does not have to worry about finding a ZERO flux level to return to. In this method the material is always saturated in one state as long as there is a ZERO and the flux is switched all the way to the positive state when there is a ONE. This means that there is a signal every time there is a ONE and no signal when there is a ZERO.

The nonreturn to ZERO method approximately doubles the density because it uses half the flux changes In this case, every flux change is considered a ONE and no flux change is considered a ZERO. this is a little more complecatedbecause one has to know the previous history in order to know whether to record positively or negatively for ONE.

The Manchester system has a number of advantages because it puts out a positive signal for ONE and a negative signal for a ZERO. The difficulty is that you won't have to set up the state which allows the positive or negative flux change half a cycle before the change is made. This might be particularly difficult in a system like BBN wants because we will not know the state which we want to set up until we want to do the actual recording.

Date October 25, 1960

Subject Time Tickets

To Harlan Anderson/Benjamin Gurley

From Loren Prentice

In reply to your memo of October 18, 1960. Starting the week of October 3, 1960, Yalton Greene and John Culpon are turning their the time tickets in to Nancy Dawes. She is going to keep a record of the time expended on various jobs, with two objects in mind. One, to find out what the overall cost of such items as PDP-1 consoles and display units, and two, to find out if excessive times are spent on individual items.

For the two weeks ending October 14, 1960, approximately 54 hours and 45 minutes were spent on PDP-1 display. Some of the times of various items which seem to be in excess of the amount necessary are the front bezel for the display, 20&frac4; hours; focus coil and yoke mounting plated, 10 hours; side mounting blocks and tube retainer clamps. These are very small pieces, very simple to make, and the time on these was 10&frac4; hours. This is outside of the items just mentioned. This is outside of the items just mentioned. This is for the display island not included in the 54 hours and 45 minutes. Twelve switch bats, 7&frac4; hours. Also, it should be kept in mind that these parts have all been made by the shop previously. That is, this is not the first time they were run in any instance. Four hours should be sufficient to make the switch bats, eight hours should be sufficient to make the front bezels for the display, and certainly making mounting blocks and tube retainer clamps in 10&frac4; hours is certainly excessive.

Several steps should be taken. One, we should do what we can to reduce the time required to make these parts by brining it to the attention of the people involved. Secondly, we should attempt to place outside, some types of work which we have very good drawings or at least get prices on them to find out whether or not they are competitive. In the past we have found it very difficult to get the machinists to do work without prints. They seem to be able to mis-interpret instructions to the point that the parts are very expensive or not well enough made or they take time themselves to make extensive sketches which is just as costly to have then make sketches as it is to have them made in the drafting room. We have in the immediate future, four, three hole dies and punches which could be used as a trial run on outside machine shop work. Also, two mounting assemblies for Telemeter Magnetics memory stacks. Drawings for the stacks exist and are being checked, and the other prints will be available in the near future.

Thursday, November 3, a request for quotation came in from Convair, in Fort Worth, Texas, on the Programmed Data Processor. We want to be sure that this gets our intense attention so that we can get the quote out on time.

November 28, 1960

Major Wiliam Harris, CCSET-2
Command Control and Development Division
United States Air Force
Bedford, Massachusetts

Dear Major Harris:

I am pleased to furnish you with price estimates for the Programmed Data Processor - 3 (PDP-3) for your evaluation. Prices for Item 1 to 6 are quite firm as of this date. Prices for Items 1 to 6 are quite firm as of this date. Prices for Items 7 to 9 would be somewhat dependent on your detail technical requirements such as message length, type of entry into the PDP-3, etc.

The machine delivery is, of course, a function of when an order is placed, but would currently run from 10 to 12 months after receipt of order. The machine would be delivered with programming aids consisting principally of: a. FRAP-3 (Symbolic Address Program) b. DECAL-3 (Expandable Compiler)

The PDP-1 prototype version of this machine has been successfully in operation for approximately one year, and the first production delivery was made just recently. The magnetic tape units available with the PDP-3 have IBM compatible formats which would be of considerable usefulness in your program.

If you wish further information concerning the PDP-3, please feel free to call upon me.

Sincerely,

Harlan E. Anderson

Date December 5, 1960

To Itek Corporation
Information Technology Laboratories
Waltham, Massachusetts

Attention: Mr. Norman N. Taylor, Vice President

Gentlemen:

Digital Equipment Corporation is pleased to propose the purchase of a DEC Programmed Data Processor - 1 by Itek Corporation. The PDP-1 will be as specified in the attached manual.

Equipment List and Prices -

Terms -

This quotation shall remain in effect for 30 days if not accepted or extended by that time. Prices shown are net and full payment is due 30 days after acceptance of the machine by Itek. Prices do not include applicable taxes of local, state or federal governments. All shipments are made f.o.b. Maynard, Massachusetts.

Warranty -

The PDP-1 System and its component parts, as delivered, are guaranteed to be free from design and manufacturing defects for a period of one year following installation of the machine. Any component which fails during this period will be repaired, or at DEC option, replaced.

The below information is of the nature of performance objectives and shall only be used as such.

Proper operation of the machine as used above is defined as an average of at least 10 days between failures. A sequence of past failures directly attributable to a single repaired component are to be considered as a single failure when computing average times between failures. If magnetic tape storage is included in this system a relatively low level of random errors in reading are not to be considered failures as used asbove.

The unscheduled repair time will be, at least, 10 hours per month.

This warranty does not cover components which have been modified without DEC approval or have been subjected to unusual physical or electrical stress. Failures due to such components or due to equipment connected to the computer without DEC approval are not to be included when computing average time between failures for the purpose of this warranty.

The maximum liability of DEC under this warranty is limited to necessary service calls and repair or replacement of defective parts.

Delivery -

PDP-1 wwill be delivered 4*months from receipt of a firm order. This means that if the Itek delivery date of April 15, 1961 is to be met, a firm order must be placed no later than 15 December 1960. Between January 1, 1961 and April 15, 1961 computer time will be available on a PDP-1 computer at Digital Equipment Corporation plant in Maynard, Massachusatts for program checkout and preparation by Itek personnel. A maximum of 10 hours per week are available at a rate of $50 per hour. Cost of this rental time may be applied toward the purchase prioce of the PDP-1 computer proposed up to a maximum of #10,000.

Maintenance -

There will be no charge for service of a period of one year following installation. This service shall include that necessary to insure proper operation of the machine, repair of any failures that ovvur, and routine checks during this period. Service contracts beyond the initial period are available either at a flat rate or on a per call basis.

* If magnetic tape option is selected, this part of the system will be delivered wiyhin 6 months after receipt of a firm order.

Installation -

All equipment except magnetic tape will operate from 10 amps of 115 volts single phase 60 cycle power. Each magnetic tape requires an additional 15 ampares. No direct connection of air conditioning to PDP-1 is required. General air conditioning of the building is recommended. but not required. Floor space of the computer is as follows:

We trust you will finf this proposal acceptable andd we look forward to an opportunity to be of service to you. SWhould you wish addiutional information, please feel free to contact me.

Sincerely yours,

DIGITAL EQUIPMENT CORPORATION

Harlan E. Anderson
Vice President

MEMORANDUM

____________________________________________________________

It is proposed that DEC purchase about 200 hours consulting time of BBN personnel in order to accomplish the following:

It is believed that the proposed modifications (Nos. 1-4) will allow the PDP computers to function in a manner similat to the currently proposed PDP with sequence break, and with the high speed channel. The proposed modification will also allow all input-output devices to operate without buffer registers, counters, format control, etc. Thus, the magnetic tape control units could be eliminated. In addition the modifications would also be available to facilitate internal programming.

Time spent working on this project should be at the mutual convenience of the parties concerned, and at a rate that ensures completion of the outlined tasks within 60 daye after inception. John McCarthy's major contribution will be accomplished during tne M.I.T. Christmas vacation.

The final report will include a co-authored paper, suitable for publication or presentation at a professional conference, that describes the PDP-3 computer as modified. We invite participation by DEC in the authorship of the paper.

The major part of these services would be provided by John McCarthy and Edward Fredkin, whose charge-out rates are $25/hour and $20/hour, respectively. However, incidental time of other of our staff may be required during the course ef this work. This time will be chargeable in accordance with our standard consulting rate schedule, as follows:

Date December 5, 1960

Subject Suggested Production Schedule

To Harlan Anderson

From Dave Denniston

Digital Processor 2305 (NEL)
(EN - 2029)

It is intended that some mounting panels will go into and come out of wiring before the proposed deadline dates.

All wiring has been completed on the Digital Processor for the U.S. Navy. Required units have been completed and delivered; checkout can commence immediately.

Date December 22, 1960

To Ben Gurley/Harlan Anderson

From Ken Olsen

Bob Hughes, of IT&T Computer Group, is going to visit us next Wednesday, December 29, at about nine in the morning with John Acme, and maybe some others, to discuss the purchase of production quantities of our computers. They are interested in a prototype in about six months and, in a year, production quantities of about one a month. I gather that these machines are pretty much like what we are doing now.

His name is Robert W. Hughes, and according to the I.R.E. Directory his address is Paramus, New Jersey, and after his name it has the initials VP which I would guess means he is a vice president of IT&T.

Date December 16, 1960

To Jack Atwood

From Ken Olsen

Our oscilloscope and light pen deserve news releases that should make the front sections of most magazines because of their unusual interest. We may want to play them as separate items or together, but I am sure if we do it right we should make the front section.

Date December 29, 1960

To Harlan Anderson

From Ken Olsen

Mr. R. W. Hughes, John Ackley, and Leo Murphy visited us on Wednesday morning, December 29, to discuss the application of a PDP-1 to a message switching center,

After the meeting, I sent them te American Research to meet Billi Congleton and other people there. Bill Elfers is a very close friend of the president of , Harold Geneen. I suggested that Bill do a little name dropping because we may want to use any influence we have when Hughes and company make their pitch to their management for this project.

IT&T is planning to make a decision about Pebruary 1 as to the program. If it goes through, most likely they would want a prototype machine delivered in six months and production machines starting a year later at a rete of a duplex computer every month or two. I said we could deliver the preduction machines in about eight months after receipt of order.

They are particularly interested in the multiple sequence but would need about 256 inputs. John Ackley suggested that the four flip-flops which we have at each input should be tied to the external equipment supplied by IT&T.

They would want complete trouble shooting manuals and instructional information and would also like to have a man in residence for several months. Our Logic Bock, of course, is a big step in the direction ef educational material.

They are going to send a specification and proposal to us in the next day or two, and I said that within a few days after we receive we will send beck a price and quote. It may take a few telephone calls before we reach common agreement on this, but we should act on it wery quickly.

They consider thie information preprietary and would lkike us to keep it confidential so that other people in IT&T wi11 aot be disturbed by this project and so that their potential competitors will not know ahout it.

We promised to send them some figures of reliability - either measured or estimated. We also sald that included in the package would be maintenance programs, checkout programs, and utility programs.

We aise promised to send them a recommended list of spare partes for duplex center and a list ef necessary test equipment.