There's at least two album names on this sentence

@@luisclaudiofugolin6250 That's what I thought as well when I read it.

@@luisclaudiofugolin6250 where

_"... obscured by _*_clouds_*_ "_ ☁ 🤓

@@RichardNobel lol

I think the schematic for even a Z58 chip-set with a i7-2700k would be stupendous 🙄🤯🤔

@@MrLunithy You can often find leaked schematics for modern computers. While there's a ton going on it's actually pretty easy to understand if you only look at one piece at a time. When chips are as complicated as they are now they don't need much support circuitry other than power supplies.

@Janek Winnicki that's funny, I started programming at the age of 10 on a acorn electron. I was highly fascinated that I could make a computer do stuff.

Computers still come with manuals... including a rough schematic. You just need to purchase your motherboard separately (and not a fully-built Dell, for example) to get that manual. You can also download them as a PDF from the manufacturer.

@@JonRowlison "rough schematic", it tells you which holes to plug things in, and each type of plug will only fit one type of hole. You've no idea how the LPC bus connects the ROM to the CPU, or if there even IS an LPC bus and what it replaced and why. There's still a basic IBM 5150 in there _somewhere!_ How ever many of it's interrupt controllers and timers are emulated in software at some level by some in-chipset CPU (and there's probably at least a couple of those, and you'll _never_ see the source code!). Just the spec for PCIE is only open to members of it's Special Interest Group, which will cost your company a shit-ton to join, if you Need To Know, and to download it you search for which _pieces_ of it you're interested in. Possibly nobody in the world has read the entire thing. Certainly nobody in the world understands a modern PC from gate level to the OK button on the screen. And there's stuff beneath the gate level. Modern PC "manuals" are nothing _near_ as comprehensive as the ones that came with computers back in the day. They had entire circuit diagrams. A modern motherboard might contain, essentially, 1 big chip, that's more complex than 1,000 chips on the original PC. There are billions of logic gates in there. With a PET or Spectrum or whatever, you could find out what bits to set to get immediate results in the hardware, no OS, no other code interpreting your wishes, you just DID it, right there and then, yourself! You could repair, replace, and upgrade parts that aren't even visible today cos they're all in one gigantic chip, and you'd need something like an atomic force microscope to see the parts if you got the lid off. Really computers today... thousands or millions of times more transistors, no exaggeration. They all do something. Then there's the millions of lines of code, all interdependent, where you used to have perhaps a few thousand that did exactly what they said they did. Now, a program you write might be running on a CPU like you expect. Or it might be emulated. Or hypervised, or trapped and partly-emulated. On a machine with similar or massively different architecture to what you thought you were programming for. It might well be running on some cloud server halfway across the world and only part of it's actually running on the so-called user's PC. Parts of it might be operating on different servers at the same time. You and me might be centuries dead, running as AI's in a virtual world a tiny fraction as interesting and sensual as the real one was. And nobody's told us! We really wouldn't know, and software doesn't, can't know. There is no comparison. The comparison between a Ferrari and a roller skate, really, on that order. Pardon all the hyperbole but I don't feel I've exaggerated at all, computers now really are ludicrously detailed inside. Where the stuff of the '70s and '80s wasn't so far on from the machines they used in WW2. Go look up transistor counts of PC CPUs from the 8086 to the modern Core Whatever. The 8088 CPU of the first PC contained 29,000 transistors. Some recent thing from AMD contains 39 billion transistors! Admittedly a lot of that is cache RAM, silicon monoculture rather than the jungle of other parts, but I think it makes a point!

Thanks for the idea! Not a bad one - could start with turning the machine on, then what happens until it goes input idle, then how it parses and executes the line of code. I'll think about that!

Kinda boring imo, the DOS interrupt vectors are more fun. ;)

@@DavesGarage *Builds TPF file from original sprouting as BMP from default texmod tool* *Upscales multiple ways and purges old .log file with removed associations so its absolutely fresh and associated, while making sure the name matched after removing the non upscaled exacting format* I can't get Jurassic Park back online without Dennis Nedry. -Ray Arnold The brain inflammation stage from staring at a screen for 65 million years is setting in. Ill cycle through the different compression technologies and see if I can definickize this blasted tool which im glad with very little effort I figured at least the above out.

@@DavesGarage My first lead. Gigapixel is a liar. Compression is TIFF despite the setting of keeping it the same.

@@DavesGarage The Eagle has landed. The trick was not using BMP default since Gigapixel cant meet that. JPG in the .log file and outputting in Gigapixel worked. My T Rex looks far less compressed. Ill have to figure out the multiple asset same name, tree of funsies in time since its the only thing left to figure out, then its just pure drudgery.

the IBM PC, while more complicated, was still comprehensible to me on a software level. the hardware design was well past my understanding. DOS 1-3 and the IBM PC BIOS were well documented, and you could lterally debug through most of it and make sense. I learned assembly from the examples in the MS-DOS 3 reference, which also had an appendix with the IBM PC BIOS reference. Amazing that someone (or more likely a team) at Microsoft wrote assembly examples to demonstrate literally every single DOS 1, 2 and 3 function!

Or NOT doing. Me too.

@@linusfu515 The nice thing about having libraries that interact with the OS for you is that they might be OS-independent so it'll work on Windows, MacOS and Linux without you having to write it 3 times.

​@@linusfu515 I have 3 books on my shelf specifically there for computer stuff. O'Reilly "learning Python" , Sobell "A Practical guide to Linux commands, editors, and shell programming." , and Comptia's 7th edition A+ guidebook. Mentioning this, because I personally think you have the right of it mostly on how people kind of gravitate towards Python when first learning, despite having other resources available to them. I don't know if that's a good or bad thing, but I do blame my raspberry pi and its camera for needing to learn Python to use the camera properly (at the time at least) on why it ultimately became one of my first languages being learned; instead of something else like Assembly and C with its variants.

Indeed, but we have so many layers of complexity and abstraction that no single person can fully understand it on anything reasonably "modern" to be useful to an average user today. You can pick and choose which layers are useful/interesting to learn, although some parts have little to no public documentation and are essentially secrets (like firmware and microcode).

The 8-bit Guy did an entire series on Commodore computers.

@@eriksiers true. But Dave has a different perspective that would nicely compliment David's.

Definitely look up the 8 Bit Guy, hes got so much on old computers, Commodores, and the like. He will definitely fill what your looking for.

who invented screen editing? I didn't think it was Commodore

Ah, you should see Vi !

@@greenaum I use vim but it can be annoying that because it's more of a "line editor" you have to put it in edit mode once you get to the line you want and then you have to exit the edit mode before you can move on to a different line

I think my old Apple ][+ did the same thing, including the ReadY thing. (I could be wrong - it's been a REALLY long time!)

It was Microsoft BASIC that did it. The PC's version does the same thing. It also maintains an extra byte in RAM for each line on the screen to indicate whether the text on one line wraps into the next line, so that the screen scraper knows where to stop interpreting.

Perhaps some university programs could have some retro programming classes! Could be fun.

@@wandererstraining My university has some classes that deal with writing assembly and how logic gates and an ALU work. I don't remember if they ever expanded that into the full picture though.

@@cameron7374 OSSU has in its curriculum a course called NAND to Tetris, which is apparently very challenging but fun.

Nothing on planet Earth warrants a visit to Reddit.

I also recommend James Sharman's builds.

My first programming class used a KIM-1 single board computer with a 6502 and a whole 1K of RAM. I got to where I could perform the startup from memory. That class led to a field service career for CNC machine tools that were using punched paper tape, but we had to key in by hand the 30 or 40 step program that made the tape reader usable - no ROM. I really miss front panels and bit manipulations. 🖖

​@@markcoleman9892I'm not saying you LIKE pain, but I'm not saying you don't either. :)

Link to channel please?

​@@LordAssClownclick on his name

@@LordAssClown You can just click his username or profile picture.

thats what im doing right now lmfao. Getting a shit ton of TTL chips and making my own computer, and then moving onto OS.

You still can, although it requires quite a bit of patience, but at least there are tons of easier to reach documentation on how to do it today. Want to build your own TTL computer? You can, there are even entire series on KZfaq on how to design and build your own custom CPU if you want. Want to write your own OS? There are tons of wikis, tutorials and books on the subject, and entire open source projects that you can download for free. Want to build an 8-bit or 16-bit computer? There are kits you can buy and build and tutorials on how to design your own. Microcontrollers (and their development kits) and basic logic analyzers are essentially commodities now. And FPGAs went from a thing you could only dream of, unless you worked for very specific fields, to something you can buy a decently sized one for a reasonable price and build quite complex digital circuits with it. There are even video series on how to develop cores for FPGAs today and tons of open source cores that you can learn from. If you are so inclined, take a look at the MiSTer FPGA project, lots of retro gaming fun and also a very nice learning tool if you want to either dabble or deep dive into digital circuit design. Sure, it's "expensive", but still costs less than a modern video game console or a gaming computer and I promise you, it will give you way more hours of fun than those other options.

@@gcolombelli If you build your own, that is different situation. Also open source and open designs are different. I agree to that. I used to work with closed source systems mainly. But having access to all hardware levels and low level interfaces and functionality is not so accessable for commercial reasons. Nowadays there are exceptions, I agree. In my university years I had access to OS source code from DEC and other companies. But later even working in a computer company, getting access to the source of the software I was selling and supporting was restricted. Open source wasn't really part of the business plans of that company. Inhouse CPU development information was even more separated and information restricted.

Well the newest LTT video had them trying to use a networked swap partition and that definitely starved the CPU

This is also my theory. You'd have to dual port whatever RAM the character map lived in. As it is you had to dual-port the 1k of screen memory. Note the ZX81 didn't only have to deal with dual ported memory it actually generated the video directly, so the CPU was quite busy when video was being displayed. It's possible do the necessary dual-porting but that would have complicated things quite a bit. So they allowed two different character sets and that was that.

RAM and ROM speeds generally kept pace with CPU clock speeds. When specifications needed, you could opt to spend more money and get memory chips that had faster timing. To switch which device was using the address and data buses, you would use tri-state buffers, multiplexers, transceivers using low power Schottky transistor-transistor logic chips (LS TTL). The three states are Hi, Lo, and high impedance (the tri or third state). The tri-state high impedance state would allow some other chip to drive the given bus lines high or low. A bus conflict happens when two devices are driving say the address bus at the same time thereby causing a short circuit state on the lines where one driver was pulling the line to low, while another device was trying to pull the bus signal line high. Chip damage would happen if this situation lasted too long and made the power dissipation per chip exceed design specifications for long enough time to "smoke" the chip drivers on a given pin or maybe the whole IC would fail and need replacement with a working one. Positive holes flow toward negative polarity ground potential causing unnecessarily high current flow or a short circuit. Any downstream device listening to this electrical fighting would likely see the ground or negative driving circuit win these fights because TTL chips had lower impedance to ground polarity. The internal circuitry of the TTL outputs would be specified in the TTL data book that listed the given chips involved. Usually the outputs were using totem pole configuration using complementary polarity transistors--a PNP and and NPN bipolar junction transistor. The upper schematic transistor would be tied to +5 Volts collector circuit voltage or Vcc through about a 260 or so Ohm resistor, whereas the lower schematic transistor would have its emitter tied to ground through a pretty low valued resistance like 10 Ohms or so. The P and N junctions for bipolar transistors are formed by using infusion or doping of the semiconductor material with either a donor or hole material or atomic element whose outer electron shells would either have excess electrons or hole slots in their outer electron orbitals. The crystal lattice would have each atom of silicon have neighbors (top and bottom) as well as forward in and backward out and finally right and left with these excess electrons or excess holes (positive charge carriers). Flow of charges in junctions would only be in favor of one direction. The physics says that if the charge flow opposes and depletes the charges in the native PN junction field, the flow shall continue unabated. However if the charge flow being attempted reinforces the doped field junction, then the electric field magnitude would just grow bigger and bigger until the voltage was so high that breakdown would suddenly allow the current to flow backward through the junction. If the current flow magnitude was too large through the junction, excessive heat shall smoke the device unless the current flow was limited to sane levels within designed specifications allowable by heat dissipation abilities of the device. This would be the situation in a reverse breakdown diode junction known as a Zener diode where the reverse breakdown voltage would be a known calibrated X number of volts. Field effect transistors or FETs use a by design a channel through which flow would be made to happen by enhancing the conductivity of the channel (enhancement mode) or by design a channel through which flow is pinched off or blocked by depleting the field (depletion mode). The channel can be made to carry positive holes as in P-channel or to carry electrons or negative charge carriers as in N-channel FETs. The control lines into these various bus switching governing chips would determine who was allowed to talk or listen to the given addressed memory or I/O component. Way back in the late 1970s the speed of CORE non-volatile RAM magnetic memory was low microseconds. Semiconductor memory was about 1 microsecond down to about 450 nanoseconds. Each successive year the memory speeds got faster for semiconductor RAM. The DRAM chips for the Apple II were around 200 to 300 ns. The first 386 processors at 16 MHz were using 150 ns DRAM chips in sockets on the mainboard and expansion RAM boards. Expansion RAM boards in the early PCs would not necessarily be able to use ultra fast static RAM cache sets, nor could they use faster SCSI bus mastering DMA controller boards unless specifically designed to do so on the newer 486 mainboards with premium level chip sets and bus expansion slot standards that would permit such stuff. CPUs and chipsets started to be fabricated using advanced high speed CMOS--complimentary metal oxide semiconductors. The signaling was using less current flows to swing the voltages involved. MOS transistors used field effect transistors. NMOS and PMOS complimentary transistors to form CMOS. The fields on the signaling gates would either use depletion mode or enhancement mode to turn the transistors on or off. Instead of base input, collector or emitter outputs, the newer transistors would have gate input, drain and source outputs. Gates would require field voltages rather than input currents to do switching. The only current flows required were to charge or discharge the parasitic capacitance involved in the small gate circuitry. Leakage currents were reduced with each newer generation, thereby lowering power consumption. DIMM RAM started to come out for the Pentium series CPUs in the mid to late 1990s. It had RAM speed at or faster than 80 ns and eventually in the early 2000s was running close to 7 ns. Eventually dual data rate (DDR) chips started to come out that would do block transfers to and from the CPU's internal cache a bunch of words at a time. DDR2, DDR3, DDR4, and so on. It turns out that every step forward was done at the scale of the chip die size being shrunken more and more each new generation making the transistors smaller, closer together, have less active reactance charge capacitance or inductance delays. A lot of this was done initially with semiconductor etching masks made of shrinkable fabrics where the pattern was transferred, and then the mask was shrunken and then applied to the semiconductor wafer layers. As these shrinking advances came in, so did power supply voltage drops come into play. Switching times of the complimentary totem pole configuration transistors determined the power used because as the frequency of the signal switching increased toward and beyond the switching speeds of the two transistors, there became more and more a window of time when the bottom and top of the totem pole of transistors could be on at the same time thereby drawing current straight from Vdd to ground from top transistor through bottom transistor causing momentary short circuits causing huge power dissipation. With bipolar junction transistors such heat can cause what is called thermal runaway with excess charge carriers preventing the transistors from shutting off the current flow without a much greater base junction input current to attempt to shut the flow off. From what I can recall, CMOS field effect transistors do not tend to have thermal runaway and in fact the increased temperature would tend to somewhat increase resistance to such flows.

Let me refresh your memory. 150ns was common. Any faster got exponentially more expensive....in the end 286s etc. Ran 60 or 70ns ram, some 10+ years later.

I would solve that differently.. if the character rom was really ram, it could be mapped into a block and CPU access would take priority over the character generator. Given the system would load this ram from rom at boot, the extra blank screen at start would be un-noticed. Given that dynamic access to the character ram would be unusual and minimal, it could be timed during the blanking period without creating visual aberration's.

Shoving the character ROM on it's own bus certainly wouldn't be all that odd a choice- the NES did the same with tiles, and I think the IBM PC monochrome adapter _may_ have done the same with it's character ROM. It would always have been possible to add some circuitry (e.g. a DMA circuit in the video processor) to get characters moved, but that would have (slightly) complicated the video circuitry, and would have required extra chips (and as I recall, ROM was cheaper than RAM).

This is why I migrated to Linux relative early. I even own the T-shirt "Ich bin /root, Ich darf das!" = "I am /root, I am allowed to do that!" I also stopped developing hardware stuff for the PC and migrated to embedded systems only using Ethernet and browser based interfaces. Less pain and no sweat anticipating the next OS upgrade.

@@CC-ke5np thank you for teaching us Germany

Hi ChesslsJustAGame. Did you work on any memory test equipment when you were at TI? I believe the testers I was developing were purchased by TI for those parts. The SRAM business was brutally competitive at that time...

hear hear!

Thanks for the kind words!

I feel your pain. I still kick myself often for not taking the opportunity to snag my family's first computer, an Apple II Plus, when my dad was cleaning out his old house prior to selling it. I also enjoy watching videos of car enthusiasts restoring antique cars. Hearing restorers talk about their finds, I used to find myself shaking my head and wondering why such beautiful machines were just abandoned and left to rot - referring to both the computers and the cars. I have to stop myself and remember that at the time they were replaced, they were just old, obsolete machines, often crippled by problems that weren't worth fixing at the time. They couldn't run modern productivity software, Internet access was beyond them, and there were thousands more just like them here, there and everywhere, so unless except for special editions, they weren't worth anything. And a quick look at eBay suggests that these old gems, still aren't worth much to the market at large, and probably never will be. I currently have a Compaq Deskpro from 1998 or '99 that was retired just a little over a year ago, long after most of its peers had been chucked. I'm currently using it to demonstrate some long-gone storage media to a group of high school students. I'm preservation-minded enough that I'm not going to throw that Deskpro out when I'm done with it, especially since it's in pretty good working order. As far as what I *will* do with it, that's a good question. It's too old to be useful, but too new to be a museum piece.

@@mar4kl Absolutely. Im seeing 5 1/4 floppy drives, old sound and video cards, memory sticks bringing in crazy $$$. Granted these are not making someone an instant millionaire over night but I personally have thrown away many of these items and its painful to pay any $$$ for stuff you once trashed!! What are we throwing away or neglecting today that will have value in the future? Its just difficult to know! On your Apple II Plus over time they along with the first Mac's joined the pallets of unwanted computers.

Glad you enjoyed it!

I found it easier to do calculations in BASIC than using a spreadsheet program. :-)

I did something similar on my ZX80 in around September 1980. I had a little board that I designed that plugged into the ROM socket, but also added some extra logic and a couple of 2114 RAM chips. My ZX80 had the normal 64 characters, plus 63 user defined characters and their inverses as well as the inverse of the 64th user defined character that you could not display normally due to the way the ZX80 display generation worked. Fun days!

Everyone should know how it happens!

I feel you. How much of this video is included in your courses?

@@therealb888 We do learn how computers work at a fairly low level but it's more focused on modern platforms which he mentioned are huge compared to these older machines. This video is like looking at how a go kart works to understand a passenger jet.

@@coolbrotherf127 Good for you man. My retarded college is stuck with turbo c & 8085 (not even 8086). But low level understanding goes a long way. Instead of having a course on modern wireless comm sys like 4G/LTE networks, we were thought the GSM/CDMA standard.

@@therealb888 If you want deep dives, I'm studying assembly x86 as a Comp E student. We didn't go over basic yet, but we either will in future (I hope) or we will go back to C++ I'm typing cuz CS doesn't really go into oldschool, hardware type stuff, that's our territory. CS usually goes into performance engineering and such with more modern stuff. What i mean to say is that they really don't need to focus on controlling hardware as much, CS is more about applications and etc. Maybe @@coolbrotherf127 will go into Rust and the like in the future

I have, he is spectacular. Love his "Hello World" episode on the 6502!

Oh and when you said your left side of the screen, it was your left (stage left), it was our right 🙂 lol

Tell me more! Email me at davepl@davepl.com with the story! I assume you mean the display? I just swapped one and noticed the PCB wasn't a rev, it was COMPLETELY different than the old one!

The character ROM seems to feed data into a 74165, a shift register. So the byte is read from ROM only once per character per scanline, not once per pixel.

Infact ... it could be done, a standard BSD server, users sign in with ssh like expected, then they use minicom or such to a local BBS .... this must be tried!

Even IRC would be a good midpoint, old enough that you can get old hardware to talk to it, but not that old that modern hardware can't also talk to it! You can even get smartphone apps for IRC, just!

I used to help a friend run an Acorn BBS with 3 lines, allowing a whopping 4 people to chat together - well until their parents got their monthly phone bill and banned them 😏

In the *nix world there'd be no emulation required to work over TCP and/or serial line.

I bet he already has 😜

When the CPU was slow (1MHz range), the static RAM once upon a time was more than twice as fast than the CPU, allowing interleaved access by CPU and graphics. So no dual-port RAM needed. Just waiting for the CPU to release the address enable signal and have the graphics card step in and read. I have a home-built Z80 machine using this logic. Nice, clean, and fun to implement.

Easier to debug as separate units and both circuits were already well known. But I was thinking bus contention as well.

I may be wrong, but I think the character ROM on the PET (unlike the C64) was not hooked up to the main CPU--it was only attached to the CRTC. So, the actual hardware made it impossible to relocate the ROM.

@@whollymindless Yes, contention is definitely the issue. But there is a cheat possible in some situations. During CRT horisontal and vertical retrace, the graphis do not need to read out data. So it's possible to switch ownership for accesses. Not a perfect solution for graphics RAM since it limits available draw time per frame. But enough time to redefine sprites in a RAM-mapped character generator. Edit: but it does require some MUX chips to switch if addresses comes from CPU or CRTC or whatever is used to drive the graphics display. And it requires an extra dual-direction buffer chip on the data lines since the character generator data must be muted from the CPU data bus while graphics is being emitted. Easy to build but adds cost and PCB space.

@@d2factotum That is why MUX chips are needed, to switch ownership of data and address lines. And a means ti synchronize the code with the horisontal and vertical retraces

I knew 100% of everything and still found it interesting. 😉

I knew 90% of it and had fun doing it :-)

I knew 0% and enjoyed myself 100% of the time

Sadly no one seems to be watching this one, but they like it when the do, so I'd love to! I have to think of some clickbait name to make people click though :-)

@@DavesGarage I think the title might be putting off people who don't care about the PET. If the title were more generic maybe more people would watch.

Modern video game sizes shouldn't really be compared to those old disk drives, because most of the consumed space is graphics (whether images or 3d models).

I didn't mean (or say!) that. The one little image shows the scanning of each line, which is how it happens. If your screen says COMMODORE first it draws the top of the C, then the top of the O, the the top of the Ms, etc. Then the next scanline of the C, the O, the M, and so on. It's definitely one ROW of pixels, not a character at a time!