BillBohan

I agree with afarrell that lying is not advisable. Lying is a very bad thing. Convincing someone else to do a bad thing is even worse.

The method I used when I was Software Development Manager was to ask one of the programmers how long he thought it should take (or estimate it myself). Then I told management that it would take 3 times that long. We usually could make that deadline but it was often close and sometimes a little late. I have since learned that it will take PI times as long as you think.

Telling the programmers that they have half as long does no good. They will not write code any faster. It may even distress them and a distressed programmer produces less code causing it to take longer.

FYI

Consider carefully what I say; my first computer program compiled and ran in the summer of 1967. I have typed in a program while both Bill Gates and Paul Allen watched over my shoulder. I'm retired now but still work daily on a subgame for minetest.

I am more trusting of naturally occuring substances than I am of chemicals that do not occur in nature cooked up in labs of big pharmaceutical companies.

These are some interesting observations regarding music and machine learning. It has been my experience that the majority of the output of ML music generators falls into the category I would classify as noise.

I briefly experimented with procedural music generation many years ago and will relate my experience in the hope that some may find it interesting or take inspiration from it.

I had read the Byte magazine article called "A Travesty Generator for Micros." which works with text files and realized that Markov chains could be applied either to whole word or individual letters. Sufficiently long chains of letters almost always produce actual words. Sufficiently long chains of words generally produce complete (although nonsensical) sentences. Excessively long chains copy the input to the output. See [1] and [2]

At the time I was playing LOTRO [3][4] which uses ABC files [5] which are a text representation of music. I used the .abc files as input to the travesty program and got very interesting output. I used the rescan method which reads the input file for each note to output. It is slow but uses far less memory than the array method which reads the input once and generates a complete table of all transitions.

Running travesty on a single .abc file produces an output which is very similar to the input and only mildly interesting. Chaining together 2 or more input files is when it gets more interesting. It did not work well unless the input files had the same key signature.

I considered the possibility of transposing all input files to a common key signature but did not implement it. Nearly all music representation is an abstraction of the music. Music is generally quantized into notes of the even tempered 12 note scale. The tune is recognizable regardless of the instrument it is played on. I wondered whether there were further abstractions which could be used similar to the way that either letters or words could be used for text but am not sufficiently musical that I could discover them.

If you try this I think you will quickly get results which encourage you to continue.

[1] https://en.wikipedia.org/wiki/Parody_generator

[2] http://runme.org/project/+travesty/

[3] https://en.wikipedia.org/wiki/The_Lord_of_the_Rings_Online

[4] http://www.lotro.com/en?

[5] http://abcnotation.com/

You might want to try the free game Minetest [1]. It is very similar to Minecraft but the basic version, minetest_game, has no animals. You can add mods that give you animals if you want them but they are not necessary. There are hundreds of open source mods to choose from or you can write your own mods in Lua. I learned Lua by examining the mods that come with the game and changing them. You can play in a mode where you can't take any damage. Multiplayer games can enable or disable PvP.

[1] http://minetest.net

I had considered this as a possibility following much the same train of thought expressed by abakker but because it diverges from most common practice (and from mathematics) and because I was unsure whether anyone would actually want it, I had not really considered using it.

While reading the article and the comments I realized that it will work perfectly in my NISC processor [1]. I just need to update my specification.

To extend the simple first specification to include multiply is very simple. I just need to have a MUL register which, when written, multiplies the written value by the value in the accumulator with the low part of the result being left in the ACC and the high part in the MUL register.

Extending to include integer divide is a little more complicated. I will have to add a DIV register and a DIVEX register. Writing to DIV will divide the value in the ACC by the value written leaving the result in the ACC and the remainder in the DIV register. DIVEX will contain the address of the routine to call when division by zero is attempted which means that DIVEX must be loaded before division is used. If I specify that DIVEX is initialized to zero by the hardware and that DIVEX==0 means that divide by zero leaves zero in the ACC and the remainder (former contents of ACC) is left in the DIV register then 1/0 = 0 will be the default behavior with the ability to change that behavior simply by writing the address of the divide exception handler to the DIVEX register.

I leave it as an exercise for the reader to determine how to deal with this in high level languages. I can fully support it in machine code and assembly language.

[1] https://github.com/BillBohan/NISC

For the rest of the article start by reading https://en.wikipedia.org/wiki/Datapoint_2200 Datapoint allowed Intel to continue development with the same architecture but required all future processors to have a shuffled instruction set. The mov group on the 8008 and the 2200 was octal 3ds with d being the destination register and s being the source register. The first digit 3 got shuffled to 1. The registers were also shuffled:

8008 bits 8080

A 000 B

B 001 C

C 010 D

D 011 E

E 100 H

H 101 L

L 110 (HL)

(HL) 111 A

Datapoint did not want Intel's future processors to be able to execute Datapoint machine code. They had a lot invested in their software.

Datapoint also had a 5500 processor which was more powerful than the 2200. It was microprogrammed rather than the simpler logic decode of the 2200. It was compatible with the 2200 but had many more instructions (think Z80). Yes it was that instruction set. The engineers at Intel who worked on the 5500 design left to form Zilog.

I worked as an electronics technician at Datapoint and performed test and repair of many 2200 and 5500 processor boards including writing short machine code routines to exercise the hardware.

After I left Datapoint (1977) I worked for Tandy where I repaired thousands of TRS-80 Model 1 computer boards which used the Z80. When the Tandy engineers came into the repair area with the programmers who had written Level 2 BASIC for the TRS-80 with the first masked ROMs, they tried the ROMs on a board and it didn't work. I told them that all the boards in the area needed repair but I had one that I had just repaired. I was introduced to Bill Gates and Paul Allen and the ROMs were put on my repaired board. I typed in a one line program to count and print numbers on the screen and it worked. Bill Gates offered me a job at Microsoft which had about a dozen employees but I turned him down and stayed at Tandy. I wrote more test programs and transitioned into full-time programming.

I tried getting friends to play RPG's in the early 1980's but they didn't show much interest until I came across a supply of 6" (150 mm) square pieces of styrofoam (discarded packing material from work) which I painted with gray acrylic paint. I cut some in half and some in quarters and used them to depict the dungeons for use with metal miniatures. After that I had a group which showed up every month for 5 years.

I wrote a program for my computer which generated dice rolls, monsters, treasures, and random names to assist me in my role as Dungeon Master but the critical element was my ability to ignore or re-roll any result which seemed too unfavorable. Good players really help the game but a good Dungeon Master is essential for an enjoyable game. Computers alone cannot exhibit the empathy necessary for a great game.

Modular arithmetic is mentioned but base 6 is not well covered.

I started gforth and entered:

3 3 + base ! ( avoiding 6)

I was in a world where there was no 6. 5 1 + . gives 10 as a result. 6 1 - complains that 6 is an undefined word.

Decimal returns things to normal.

This is very interesting but does not appear to be helpful for the development of my TTA processor design. It appears that this project aims for more parallelism similar to VLIW than my simple MOVE architecture targets.

I would very much appreciate feedback and suggestions regarding my preliminary specifications which will be changing in the near future (before year's end) to include MUL, DIV, and floating point support. I really want to have a firm specification before proceeding with the VHDL implementation.

You can view what I have done at https://github.com/BillBohan/NISC .

I always log out. I cannot think of any advantage to remaining logged in. I can easily envision disadvantages to remaining logged in whether those reasons are real or imaginary. I also lock the door to my apartment when I leave. There are people who will take advantage of any opportunity they find.

I have a simple strategy which has allowed me to win every game of Nomic [1] which I have ever played.

My strategy is to propose rule changes which allow every player with a nonzero score to be declared a winner.

[1] https://en.wikipedia.org/wiki/Nomic

One of my favorite numbers is 355/113 but it's still hard to guess.

When I worked for a company that made label printers we had a potential customer who wanted us to print labels with human readable and barcode fields with 4 random letters and 4 random digits but did not want the letters to spell any obscene words. We asked for a list of words to ban but they declined to provide such a list. We did not get the contract.

Only two days ago I learned that was the way to get a line break in my README.md file on Github.

I am designing a Single Instruction Set Computer and working to take it beyond the 'that's interesting but useless' stage and make it very useful.

https://github.com/BillBohan/NISC

I see one of the advantages being that it is very easy to learn due to its simplicity. There is also no need for complex instruction decoding placed on the hardware designer.

For software there may be a slight increase in complexity for control flow because it is necessary to specify the addresses where execution continues both when the condition is met and when it fails versus CISC which specifies only the address to branch to when the condition is met and defaults to the next instruction inline when it fails but there is also a gain in flexibility by specifying both addresses because you could, for example, call one subroutine for Z and call another subroutine for NZ and both would return to the next instruction inline.

When I first read about Move machines it appeared that they were very inefficient and the simple ones described were. In the decades since then I have devised means to offset that inefficiency with more powerful hardware functions. Especially with an FPGA target it is possible to perform complex functions in hardware that only needs to be memory mapped. Depending on your application you could have vector math, complex number math, or SIMD instructions which run in a single machine cycle. It is possible to have a shift-and-add multiplier which takes several cycles that can be loaded then you could perform some other operations while waiting for completion. I have written shift-and-add multiply routines for microprocessors which needed the processor to do the shifts and to do the adds which kept it fully occupied. While I do not claim that the design I have documented is the optimal processor, I maintain that it should have reasonable performance.

One of my goals in releasing this design is to draw interest and receive suggestions and recommendations for improving it from the highly intelligent community that reads HN and uses Github. I would like to see it optimized.

OK

https://github.com/BillBohan/NISC

Thanks for this link. I will read it when I have time.

I currently have an accumulator at an address.

Moving to the next address ANDs with the accumulator.

Moving to the following address ORs with the accumulator.

Subsequent addresses XOR, ADD, ADC, SUB, SBB with the accumulator.

I have a location called Z and one called NZ which may be written. Reading from either location returns what was written to Z if the Z flag is set, otherwise both read what was written to NZ. Moving either to the PC effects a conditional JUMP. Moving either to the relative register adds it to the PC (relative conditional Jump). Moving either to the Indirect register pushes the PC on the stack and writes to the PC (conditional call).

I envision the capability of using the accumulator as a floating point register and having locations which perform floating point operations in a similar manner. It could also be considered as a vector and there could be locations which perform vector operations on it.

I am currently waiting for my MATRIX Voice to arrive.

https://www.indiegogo.com/projects/matrix-voice-open-source-...

I have a Spartan 3AN dev board, another Spartan6 board, and an Arty. I was using Xilinx ISE free version to develop for the other boards until I bought the Arty. It came with a one year license for Vivado. I did not know that activating the Vivado license locked me in to developing only for Arty. ISE will no longer synthesize for any other target. I strongly dislike the closed nature of their software licensing.

I am retired but the last 10 years I worked writing VHDL. I can kind of read Verilog and understand what it does but do not know it well enough to write in it. The systems I worked on were for oil well logging. My circuits went down a 16,800 ft. hole where it was 350° F and the pressure was over 7000 PSI. Production quantities were typically less than 100. We used no bigger an FPGA than was needed to keep power at a minimum as there was no way to dissipate heat. Also the circuit boards were quite small since they needed to fit into a housing less than 2" in diameter. Frequent design changes were needed but all ran on the same boards.

I am currently working on a processor design that I call NISC. The set of all opcodes is the null set. It's a single instruction machine that does a move instruction with two operands, Source Address and Destination Address. I have considered putting the specifications and design on the internet as open source but am not sure where I should put it. Would anybody be interested in seeing it and where do you think I should put it?

This explains some digital logic but does not actually explain how a CPU works.

It does not explain an XOR gate, but references them. It also does not explain flip-flops, but references them also.

There should probably be an explanation of state machines since they are usually used to control the cycles in a CPU.

The title is actually "How does a CPU works?" but other than these few things it was a good article.

I know that text messages seem more harsh than intended and do not wish to seem demeaning, but helpful.

I bought a book called "Thinking Machines" by Irving Adler in 1962 which explained the concepts in this article (except for the 74181 ALU) and the XOR and flip-flops but did not understand how a computer actually worked until many years later.

Interesting. It sounds good at the start but becomes cacophonous shortly thereafter.

Another war story here: I used to run a 16 track mixing console with a rock band. The guitar player played with 2 different bands. He got a gig in a club one night and both bands showed up. Everybody insisted on playing.

It was OK but the less accomplished drummer insisted on playing a drum solo. I ran his drums through a delay in the range 160 - 200 mSec and mixed it back in. Every drum strike was doubled and the audience liked it. I don't think they would have been as enthusiastic if they had heard what he was actually playing.

Just some personal observations here.

I have read that most software breakthroughs occur around 1:00 AM and have found this to be true in my personal experience.

In the military I worked in a shop which was manned 24/7. Normally the shifts were 8:00 AM - 4:00 PM, 4:00 PM - midnight, and midnight - 8:00 AM. One month they changed to two shifts, 8:00 AM - 8:00 PM and 8:00 PM - 8:00 AM. The next month they moved everybody to the other shift. I worked each of the 3 normal shifts for several months each. For me the most difficult was midnight - 8:00 AM because of the difficulty sleeping in the daytime in a noisy barracks. I found that for me the best way to cope with this was to work a shift, stay up all day, and work another shift, after which I would sleep for 12 hours.

In civilian life I worked normal 8-5 jobs and it has always been difficult for me to get to sleep and wake up early. One time I worked a normal shift, worked through the night with no interruptions, and worked the next shift. I was able to accomplish more in that time than I normally would have in two weeks.

Now that I am retired I go to sleep when I am tired and get up when I wake. There is no discernible pattern to it.

It is much easier to stay up later than it is to get to sleep earlier. The easiest way to adjust to a new sleep pattern is to stay up later, delaying sleep until it has moved to the desired time. You can move it by two to four hours a night this way. Trying to sleep earlier you can only alter your sleep cycle by about half an hour a night.

My first program compiled and ran in the summer of 1967.

It was not unusual for a computer to have 4K of core memory. We all knew that it was 4096 bytes. The first computer I owned came with 4MB of memory. It was really 4194304 bytes. My current machine has 32GB of memory (34359738368 bytes).

It was the marketing people who wanted bigger numbers. Video games touted ROM sizes in bits to make them seem bigger. When hard drives got into many MB manufacturers switched to calling 1000 MB(1048576) a GB to make them seem bigger.

<opinion>

If you don't know about computers, you're not likely counting bytes. If you know about computers you should not have a problem with 1K = 1024. I don't think they should use 1000 = 1K for computer memory sizes. I'd hate to have to look for 8.589934592 GB Ram modules for replacement if one of mine fails. Microsoft didn't raise computer literacy, they lowered the bar.

In the 1960's I learned the Metric System in school and was told that soon everything would be metric. My 42" 4K UHD monitor displays 3840 x 2160. What's 4K about that?

</opinion>

I skip past articles and their comments if I know that I will not understand them, e.g. details of a language which I do not know.

Sometimes I will look at articles which are about things I have never heard of just to find out what that is. Sometimes, the article does not sufficiently explain what it is, like when they only tell about the improvements over the previous release. In these cases I try to find what it is from comments.

If an article is tl;dr sometimes I look to the comments to see whether I can get the gist in brief.

For articles within my realm of interest I will usually read the entire article, then the comments.

I usually avoid making comments on articles I have not read unless the comments provide sufficient explanation for me to add my input.

I truly appreciate the intelligence and courtesy of the comments on HN and try to bring the same to it.

leereves is correct.

The sequence is 2/3 + 2/9 + 2/27 + 2/81 + 2/243 + ...

and it does approach 1

In your second term you should take 2/3 of the remaining 1/3, not (2/3)^2

I saw some odd concepts in this article.

Robots are stealing jobs? Maybe they are used to replace humans in a job which they can perform better, faster, and more economically than humans, but the only robot I have seen stealing was a vending machine.

Tax the robot? Robots typically have no income. If the robot fails to pay tax would it be locked in prison where it could power down until its sentence was served? Robots typically increase the profit of the company which uses them and that company pays taxes on its profit. With the progressive tax system we have they pay tax at a higher rate. It would seem that there is already a tax on robots.

I worked for a company that made automatic label applicators which would apply an adhesive label to a box coming down a conveyor. I have helped install these in manufacturing plants where the employees were extremely hostile towards us. The person who applied the label in the past was no longer needed in that position. My response to them was that in the majority of businesses where I had installed a machine, they took the people who had applied labels and started another production line, increasing their output. If their company failed to do so, it was an issue that they needed to talk to management about. I got the definite impression that they intended to sabotage the machine.

Many years ago I had the idea that it would be possible to make general purpose robots with most of the capabilities of a human. These robots could be used to build a factory which produces more robots, then a factory which produces e.g. shoes, then pants, then cars, and then mining and agricultural robots. With no labor cost involved it would be difficult or impossible to compete with them. With fewer people employed it would become increasingly difficult to find consumers for the goods produced and the manufacturer would be left no choice except to provide a basic income for all people. Those who want to rise above a basic income would need to engage in design work, developing new styles of shoe, pants, cars, etc. They would be compensated according to the demand for their particular design.

If the robots are capable of repairing themselves or other robots it would still be necessary for a human to monitor the overall operation to preclude the possibility of a malfunctioning robot "repairing" other robots to malfunction.

We are still a long way from having such general purpose robots but the use of robots is already taxed.

I met that guy.

I was working for Tandy, repairing TRS-80 Model I circuit boards when he and his partner, Paul came to Tandy Apparatus with some engineers from Tandy R&D. They had the first masked ROMs for Level II BASIC. They put them on a board and tried it and it didn't work.

I took the board I had just repaired and went to them and told them that they were in a repair area and all of the boards were in need of repair. I offered the board I had repaired and told them that it should be able to run their ROMs.

They moved the ROMs to that board and powered it up. It came up with a MEMORY SIZE? prompt and William said to press enter. I did and it gave a READY prompt. I typed in a one-line program to print numbers and it started scrolling numbers down the screen.

William was so happy that he offered me a job at his company. He said that he had about a dozen employees. For personal reasons I was unwilling to locate outside Texas and turned down his job offer.

I found the method I was taught at https://en.wikipedia.org/wiki/Methods_of_computing_square_ro... under the heading Digit-by-digit calculation.

It does not appear that this matches any of the methods tested. A long time ago I implemented it in assembly for integers. It should be possible to write a floating point version but my programming experience has been in embedded systems and I never needed floating point. I would need to study floating point representation but don't have much interest.

I really wish this article had explained how each method works. It's going to take me some time to analyze each method to see whether any of them are the method I was taught in elementary school.

It extracts the square root to any desired precision one digit at a time and it looks a lot like long division. In binary you get one bit at a time and there is no guesswork so it goes pretty quickly.

I worked for Datapoint from 1975 to 1977 as a technician in board test and repair. I started out working on power supplies, later moved to mod/demod boards (for the tape drives), then to processor boards.

I only repaired a few of the old serial processors, these were boards sent in from existing customers for repair. The production boards were the TTL boards for 2200 and 5500.

The 2200 boards used a pair of 74181 ALU chips and a pair of 7489 RAM chips for the registers. They had 121 total ICs, ran at 8MHz, and instruction decoding was performed with mostly SSI chips.

The 5500 boards had 131 ICs, the ALU and registers were 74S series chips and ran at 20 MHz. Instruction decoding was done by microprogrammed ROMs. The instruction set was very similar to the Z80.

While the 8008 executed the 2200 instruction set, later chips made by Intel and Zilog changed the register numbers at Datapoint's request so as to prevent them from running Datapoint's code which represented a major investment for the company.

On Datapoint processors, a register/register load instruction was (in octal) 3 dst src. The later chips changed this to 1 dst src, with the instruction type (high 2 bits) having the high bit inverted from Datapoint.

Dst and src were similarly changed, Datapoint having A=0, B=1, ...,(HL)=7 and the later chips having A=7, B=0, ..., (HL)=6.