The Killer POKE

It's time to demystify the Killer POKE. What does it do? How does it do it? We're about to find out.
You can get your free $20 credit for Linode at: www.linode.com/...
Check out Adrian's Digital Basement here: / @adriansdigitalbasement
André's Killer POKE write up: www.6502.org/us...
Playlists of more stuff like this:
Computers: • Computers
Electrical Engineering: • Electrical Engineering
1980s: • 1980s
Other Links
KZbin: / akbkuku
Github: github.com/AkB...
Thingiverse: www.thingivers...
Patreon: / akbkuku
Discord: / discord

[@8_Bit]

Fun video. As far as I know, the Killer POKE only sped up programs that used the KERNAL to print to the screen, which would be BASIC programs using the PRINT command, and machine language programs using the CHROUT routine. Any program that directly POKEd or stored to video memory would already be at full speed, but would also cause the snow effect.

[@adriansdigitalbasement]

I wanted to test some programs that wrote directly to RAM but since my BASIC 1 PET lacks working IEEE routines, I can't use my PETSD solution to try to load things that actually do that. I guess back then most things used KERNAL routines and BASIC anyway? The KERNAL also blanks the screen during scrolling to eliminate snow which is why we get flashes, even with killer poke turned on.

[@8_Bit]

@@adriansdigitalbasement If you're doing a follow-up video, try this one-line BASIC program: 0 ?CHR$(147):TI$="000000":FOR X=32768 TO 33767:POKE X,77.5+RND(1):NEXT:?TI You can eliminate the white space in the line, it's just there for readability and will run slightly faster without it.

[@adriansdigitalbasement]

@@8_Bit I will be and thank you! Will try this out!

[@JacGoudsmit]

@@8_Bit I'm pretty sure Adrian's PET doesn't have TI$. That's something that was introduced in later versions of BASIC. So Adrian, you'll have to assign TI to another value (say TS) before starting the loop, and at the end of the loop print (TS-TI)/60 you get the number of seconds that the loop takes.

[@8bittimes]

@@adriansdigitalbasement In the VICE emulator, if you look into the source of it, I had included a ROM patch, that you could (optionally) apply to the BASIC 1 ROMs, to actually make the IEEE488 work. Maybe you can try and put that into an actual BASIC 1 ROM and check it out

[@graealex]

I like that you have a sponsor where you're really behind it. You're basically just promoting an actual business partner. Not some sketchy mobile game like many others do.

[@JessicaFEREM]

Yeah I absolutely love it when the KZbinr legitimately uses the program and even pays for it. I definitely trust Shelby a lot more than some other KZbinrs because of that, and the fact that he's a pretty nice dude

[@JamesR624]

Or that would be the case if Linode wasn't this week's "KZbin Machinima". The company that's currently getting any KZbinr that needs money to sponsor. Last week it was Raycon Wireless Earbuds. A few months ago it was the Raid: Shadow Legends game. A few months before that it was NordVPN, and many years ago, the first was Machinima.

[@graealex]

@@JamesR624 If Shelby says he has been using them before, then I unconditionally believe him. I also do use NordVPN myself, and used a coupon, although not from Shelby, as it's actually a useful service, although the pitch they make the KZbinrs do isn't completely honest. And let's not talk about mobile "games".

[@mEnTL32]

One of the few ads I did NOT skip because I was interested in what he had to say about it

[@DOSBrony]

Sponsors have no place on youtube. If a business tries to circumvent an adblocker and interrupt a video/waste video space, then they absolutely do not deserve to stay in business.

[@EagleKenG1]

"I am serious, and don't call me Shirley"....

[@EngineeringVignettes]

I see we have an _airplane_ fan...

[@kylemcisaac]

@@EngineeringVignettes It was at this moment that led to my drinking problem. *pours over own face*

[@EngineeringVignettes]

@@kylemcisaac - Captain Over, the Mayo clinic on the black phone... Captain Over, Mr. Ham on the white phone. _OK operator, gimme the ham on white, hold the mayo_

[@kylemcisaac]

@@EngineeringVignettes Male PA Announcer: The red zone is for immediate loading and unloading of passengers only. There is no stopping in a white zone. // Female PA Announcer: No, the white zone is for loading. Now, there is no stopping in a RED zone. // Male PA Announcer: The red zone has always been for loading. // Female PA Announcer: Don't you tell me which zone is for loading, and which zone is for unloading. // Male PA Announcer: Look Betty, don't start up with your white zone shit again. There's just no stopping in a white zone.

[@EngineeringVignettes]

@@kylemcisaac - "Roger". "huh?" "Request vector, over" "What?" "We have clearance Clarence". "Roger Roger, what's our vector Victor?" "Tower, radio clearance, over." "That's Clarence! over.... over." ... How did these guys not crash? (Karim as the co-pilot... epic)

[@computeraidedworld1148]

That bar at the bottom during the ad is very clever, I don't see many doing that.

[@onometre]

I think this is the first time I've seen it

[@fabiosarts]

the're mostly seen on programming videos

[@computeraidedworld1148]

@@fabiosarts I usually don't watch programming other than, Well I forget his name but he is working on a really cool OS

[@BrunodeSouzaLino]

First guy I've seen that bar was Alex French Guy Cooking.

[@Sharklops]

yeah Corridor Digital does that as well and I really like it

[@8_Bit]

13:31 The reflection! :)

[@cbmeeks]

And don't call me surely!

[@JacGoudsmit]

Thank you, that was very thoroughly researched. I'm glad you and Andre got in touch to set things straight (I like to think I helped a little with that). That being said, pulling a TTL output low with a 6522 is not recommended of course; I expect that the 7408 has strong transistors so the signal keeps being generated; the killer poke tells the 6522 to use its output transistors to fight the output transistors of the 7408. Two things can happen I think: Either the 7408 wins and the 6522 breaks, which is not a big deal but the killer poke may stop working, or the 6522 wins (like in Adrian's case apparently?) and the vertical sync stops going to the monitor. Possibly eventually the 7408 breaks and will need to be replaced but of course when the video goes blank you're probably going to hit that off switch really quick. Either way, regardless of what Killer Poke does, you really shouldn't do it, not on an original PET and not on a CRTC PET. By the way, small detail: As far as I understand, the flickering only happens on the original PET because the video didn't run in sync with the processor; the Kernal routines to print a character on the screen would make sure the display was in a sync period before it would access the video memory so that simultaneous access from the video hardware would not cause a conflict (shown as flicker). So the speed increase only helped with BASIC programs that used PRINT, or with machine language programs that used JSR $D2FF for printing characters. Directly accessing screen memory was equally fast, regardless of whether the killer poke had been done or not. When the CRTC motherboard was designed, Commodore designed it so that the video hardware accesses the video memory during the first half of each clock pulse, whereas the processor only accesses memory during the second half of each clock pulse. That way you wouldn't have to worry about flickering anymore. This does require the CPU frequency to be based on the line frequency of the display but that wasn't a big problem. PAL machines with 64 microseconds per line, would have the CPU running at 1 MHz exactly, and the dot clock at 8 MHz. On NTSC machines with 63.556 microseconds per line, the CPU would run a tiny bit faster so the video line was still divided into 64 clock cycles and 64 * 8 dots. By the way, your 4032 can be converted to 80 columns: in 80 column mode, the video memory is divided into an even bank and an odd bank, and the video hardware retrieves two bytes at a time for each CPU clock cycle. Pretty nifty!

[@tarstarkusz]

Well, Adrian's machine is gonna die soon anyway. His CRT is weak. The dot in the screen that remains after you switch off power is indicative of a weak CRT that could fail at any time. Given the expense of shipping a CRT and the fact that it probably isn't being used all that much, it may very well last decades and is not worth fixing if it doesn't die. But if this machine is going to be used at shows running 10 hours at once, it is only a matter of time.

[@adriansdigitalbasement]

@@tarstarkusz This is how all 2001-8 PETs are and is a design flaw in the analog board. Has nothing to do with the CRT being "Weak" -- IZ8DWF and I are doing a video to correct the issue on the analog board which eliminates the dot on power-off...

[@adriansdigitalbasement]

The 2001N machine used the stock 2001 KERNEL with the slower "wait for refresh" routines but do not have the snow problem because of some changes to the main board. (A latch if I remember correctly.) So the 2001N machines would benefit from the permanent killer poke mod I did -- all the speed without any negatives.

[@tarstarkusz]

@@adriansdigitalbasement Believe what you want to believe. It is a weak CRT. Go ahead, put it on a CRT tester. Weak CRTs do this, a lot. So much so it is in a lot of the service manuals.

[@WesleySmallz80]

tarstarkusz I’m not gonna go to the technical side. But a sign of something does not mean it automatically is the issue. Different causes to something can have the same effect

[@zachbrown7272]

Wayne from Wayne's World has gotten really into computers

[@SuperTechIT]

I was totally thinking that...

[@Hiraghm]

The PETs are _still_ the coolest looking computers.

[@FMHikari]

I've never had or seen on personally, but i really would love a modernized design based on it.

[@Hiraghm]

@@FMHikari I've been thinking of 3d printing such a case, but need the angles and proportions

[@FMHikari]

@@Hiraghm most of the case can be 3d printed in pieces, the keyboard section would have to be separate and screwed in place to keep things simple for the printer.

[@Hiraghm]

@@FMHikari yes, but I would want to keep the proportions (ratio of the keyboard area to the monitor area, for example) and the angles (like of the monitor section, which keep it from being just a box) as much like a real PET as possible.

[@FMHikari]

@@Hiraghm That's not a problem, the internal layout can be adjusted to compensate, so at most you'll get a seam where it was fitted

[@AmstradExin]

"What is getting grounded?????" Me in the 80s using the poke on Dad's computer.

[@Jossandoval]

Wow, what a bad luck, just when you were going to be free to go out again quarantine arrives.

[@johnpassaniti4417]

There was a machine with a "killer poke", but I don't remember it being called that. There was a computer called the Compucolor II that (allegedly) had a video controller where you could change the horizontal (or was it vertical?) refresh frequency. Set it to the right value and the frequency went to something illegal that the hardware couldn't deal with. The result would be damage. I don't remember all the details, but I do remember this: I was 15 when the business department of my high school got a Compucolor II in 1978. I and a couple other prototechnogeeks would play with that machine until they would kick us out because the school was closed for the evening. One of the teachers was a member of the Compucolor II users group and would leave for us their newsletter. I remember reading about how the computer could be "destroyed" with a poke, and this fascinated me. So I did what any moderately destructive teen would have done-- I left a piece of paper with the poke command lying around, waiting for the inevitable student to come along and try it. The next time when my pals and I wanted to use the computer, we were informed that something was wrong with it and it was returned to the computer store for a replacement. Was this the result of the destructive poke? Did I get in trouble? Of course not. Well, not for that incident...

[@nismo2070]

I remember typing in pages and pages of machine code out of magazines to play free games. It was so rewarding to have it run when you finished hours of typing!

[@THRASHMETALFUNRIFFS]

Then get "syntax error" trying to beat the high score on Snake, right? Or can't find a cassette to save over...

[@MrJonline]

@@THRASHMETALFUNRIFFS This! And to add: As a very young kid i did some of those type ins. On my few first attempt i got some with faulty codes. On a certain line i would get an error upon hitting return. My young brain tried to avert this by typing in whole programs as a SINGLE line. Still those hours lost where fun and educational. I quickly learned how to debug a line of code and just fix the error when i encountered one.

[@0xbenedikt]

What a well researched video. The technical aspects check out fine. Good job! On a side note, shorting out push-pull IO pins is something many repair technicians do. It just feels so wrong.

[@mickeythompson9537]

I can remember this from school, where the were approximately 10 4016 PETs, and there was an range of effects from slight to strong distortion. (There was a game going round with a sort of 'Star Wars' scrolling text effect that used this poke effectively.) However, one or two of the PETs would go completely blank, and start to get very hot at the top, and the 'hot dust' smell came out - we reckoned that the poke redirected the electron beam in those PETs right up into top of the CRT, perhaps overloading the direction coils. I don't recall anyone leaving a machine on, but I do recall writing password protected programs that invoked the poke as a penalty.

[@sdrc92126]

I have been waiting forty years for this!

[@tombarber8929]

You and Adrian are the only 2 youtubers I have the notification bell set for, so to see both of you in 1 video is awesome!

[@adriansdigitalbasement]

Thank you!

[@hi-friaudioman]

So what you're saying is that your a sucker for commodore and you don't like variety. :-P

[@hi-friaudioman]

I just came across a new channel called retrospectre78 but he does mostly dos based 386'ish machines.

[@tombarber8929]

@@hi-friaudioman Exactly! haha. (Adrian in one of his videos said he has some Kaypros he wants to do a video on, and I seem to be one of like the 12 kaypro fans in the world, so looking forward to that video too.) Yeah, retrospectre does really good content too. Few other small retro channels that I'll shout out, Branchus Creations(vintage Mac repairs mostly), 65scribe(Documentaries about different apple products, informative and humorous), Mac84(vintage Mac repair, does live streams), tx dj(hasn't done videos in a while, but lots of modem and CPM related stuff), Macintosh Librian(new channel, only a couple of videos but good production, she did a good overview of Scsi2SD and floppy emulator for old Macs), Joes Computer Museum(various 8 bit computers, most of his recent videos are about his own homebrew computer he's making) then there's obviously the big channels, LGR, 8 bit guy, this does not compute, etc

[@supra107]

I have a bit of critique, the way you're using the teleprompter makes it clear that you are looking under the camera. Technology Connections came up with a simple contraption which shows the text exactly where the camera lens is, so maybe you could watch his video about it to get some inspiration on improving your setup.

[@TechTangents]

This is something I am working on fixing. I've got parts on order for it already.

[@evensgrey]

@@TechTangents i hope the USPS is running better than Canada Post. The sorting plants are all at least a week behind, and that's not likely to improve any time soon. Since there's no other way to get a LOT of things except online orders, there's a lot more ecommerce than normal these days.

[@ichitensho7075]

@@TechTangents I used to be the "text" guy and they used monitor facing up with a glass 45 degree angle and it will reflect the monitor screen on the glass with the lens looking thru the glass.

[@evensgrey]

@Lassi Kinnunen I immediately suspect they've just hired a bunch of people to allow them to do more sorting and moving. Post offices are often horribly inefficient in their use of their physical plant. When they get snowed under like Canada Post is, the obvious thing to do would be to lay on another shift in the sorting plants, but the union would be trouble if they did that. (This same union raised a horrible stink about the loss of jobs caused by ending home delivery some years ago, despite the fact that Canada Post was projecting they would need to add thousands of new workers even without mail carriers.)

[@absalomdraconis]

@@evensgrey : The top-level reps would obviously be able to skim off bigger paychecks with more people to represent, and a lot of unions have this deranged idea that the true purpose of business is to pay workers instead of to trade money for goods or services.

[@Derpy1969]

Discount Michael Meyers sure knows his 70s computers.

[@freedustin]

Wayne's World! Party time! Excellent! Weeeeoooooweeeeoooooweeeeooooo!

[@Wolfwood428]

I audibly snorted

[@FirestormDDash]

OH THATS WHO HE REMINDED ME OF.

[@Christopher-N]

Discount Dana Carvey wears _Retroware_ emblazoned clothing.

[@rdubb77]

Mike Myers. Michael Myers is the guy from Halloween ;)

[@AiOinc1]

You know, we used to call this the fastprint... YES! Office software most definitely DID need this speed boost! What does a Word processor do? It moves large amounts of text around on the screen! Of course they're going to take advantage of the speed boost so that they can update the screen faster and have a smoother scrolling document displayed! Talking about CRTs: The PET's CRT is analog as well. It doesn't have fixed timings like some other computers and all televisions typically have. It's fully analog and that means whatever signal is being sent to the display is what will be shown on screen. Also, sorry to say Adrian, but the fact your PET's monitor goes to a long lasting white dot in the center when you cut power... Not a good sign for it's lifetime. Monochrome CRTs will do this when they're worn out.

[@LetsPlayKeldeo]

Ahhh without your logo in the corner I could have sworn it would have been a LGR video haha !

[@TheClockUpOnTheWall]

You have the best quality closed captions. It is obvious you are manually doing them and not letting KZbin auto generate them.

[@AlainHubert]

I wasn't aware of this "killer poke" thing with the PET. Now I finally know why in my Sinclair ZX81 Basic Language tutorial book (back in 1983), they specifically mentioned that no physical (electrical) damage to the computer will happen by poking to memory, but the computer could simply lock-up in certain cases and need to be rebooted (especially when sensitive system memory areas are addressed). Thanks for sharing.

[@8bittimes]

Thanks so much for that great research effort you (and Adrian) put into it! I'm really glad this has been finally clarified. Now I have to update my killer poke article :-)

[@TechTangents]

Thank you for the original article and the help in doing more research! Feel free to use any screenshots from the video or the pictures I put up when updating the article!

[@absalomdraconis]

@wargent99 : I'd bet there were one or two combinations of tape drive and computer that at the very least would burn out the tape control transistors, due to lack of isolating relay.

[@FyberOptic]

I think a lot of the legitimately scary things from back in the day were related to damaging a monitor, and this is no exception. That being said, it was easy to pass along unprovable myths back then, especially once BBSes connected to FidoNet became more commonplace. I knew a guy in school who tried to convince me that he wrote an AI for his computer that wouldn't let him delete it, which he told with a campfire horror story level of conviction. From my experience, you'd often get these guys who were into distributing warez who didn't hesitate to spin tales. I think the more gullible in our group may have believed him, and those are always the ones who would spread those kinds of stories.

[@radio655]

Excellent explanation. I recall using that poke on my good old original Pet 2001 with chicket keyboard in 1979. That went well for some time but eventually the 6522 ceased to work and had to be replaced. Might have been a coincidence but I obviously blamed it on the poke. Still miss that machine. It was a dearly loved creativity workhorse and design masterpiece. Lots of stories to tell from back then. I could not afford a printer so I had a huge old telex machine hooked up to it, driven by a tiny assembler program in the second cassette buffer (through the user port). Printing 8k of basic took half a day and shook the entire building. Those were the days.

[@LesNewell]

This reminds me of POKE&HFFD7,0 in the Dragon32/64 and Tandy CoCo. This over clocked the CPU so everything ran fast. The only major down side was that programs saved with the poke active could not be loaded back.

[@TzOk]

If this POKE changes the GPIO from input to output, and there is another output connected to it, then you connect two outputs together. As they are "strong drive" outputs (i.e. not an open collector) you potentially could damage the output driver of one of the chips, by exceeding maximum output/sink current when shorted together outputs are in opposite states.

[@steveg5122]

This sounds like something that got amplified like a game of telephone. "did you hear that someone in the computer lab broke the computer with a keyboard command" gets turned into "hey someone burned down a computer with a keyboard command" over time.

[@benco1509]

I love these crossover style videos, featuring content from two of my favourite youtubers. Thanks so much, keep up the good work.

[@supra107]

Aaand the monetization is gone, instantly.

[@SlavTiger]

I miss the old youtube

[@buddyclem7328]

It was probably the oscilloscope video.

[@stefarossi]

Why, though?

[@libertine5606]

My dad, a refrigerator repairman, had a Simpson multimeter just like the one over your shoulder. Good times!

[@rfvtgbzhn]

I think there are scenarios where the computer might run long enough that the CRT could be damaged, e.g. if someone starts a game and goes away while it is loading, then is distract by something important (e.g. a call by his boss) and then forgets to turn the computer off. It is not very likely that something like this happens to you, but considering the sales numbers of the PET and because it seems noone reported damage, so maybe it really can't cause any damage even if you let it run for a few hours.

[@svenpetersen1965]

Usually, it is not a good idea to connect two outputs together. But the original 6522 VIA has current limiting resistors in the output. So there is no real short circuit. The 7408 can drive this load without problems. The „Retrace“ bit at PB5 will be read as zero. So, it is disabling the „wait for retrace“ function in the PRINT. With the modern WDC 65C22, it would be a problem, since this does not have internal current limiting resistors in the port outputs.

[@monsalai]

Audio appears to not be sync'd after the linode ad :(

[@mibnsharpals]

i have had an old comodore 80xx in 1989 and there was an poke to speed up older PETs the display and there was an warning, this can damage the CRT(-Board), if you run it.

[@j0eCommodore]

I remember reading about the screen blanking somewhere, most likely either in The PET Revealed or in Programming the PET/ (I think PET Revealed, though I'm not finding it now...). I had tried it on a 2nd gen PET (not the original MB not the Fat forty) and that didn't work. So I figured the monitor off was only on the original circuit (which had diagnostic routines included, which is why it worked on that machine?). Ahh - PET revealed page 89, "PB5 - video on control" Programming the PET/CBM also has interesting bits and covers the 40/80 column CTRC chip. That's starting around page 265. You might find it interesting.

[@kevy1967]

Oh wow! You have a Merlin as well as a PET 4032! Two of my favourite things from my childhood, thanks for the dose of nostalgia

[@TheVintageApplianceEmporium]

Made my day seeing both 2000 and 4000 series PETs in the same video

[@roboverholt9959]

I can't remember what poke I used, so many years ago, but I made a double Dragon clone where the screen looped around and I could draw over it, so I used it to make the background.. I probably only used text base graphics for the game .. I don't have one anymore to try and show you guys what I mean, but it was cool!

[@anotheruser9876]

Your prime number finder program has an error, 1 is not prime but 2 is.

[@sprybug]

The Tandy Color Computer 3 had a similar one. It was the hi speed poke (poke 65497,0). If you tried to write to the disk drive while on high speed poke, you could risk damaging the data on your disk as the timing was now thrown off for that operation.

[@anumeon]

Ah yes, prime numbers. Arthur C Clarke made a nice nod to those in the book RAMA II. Did you know that if you take the number 41 and add first 2, then 4, then 6 and so on. To get the sequence 41, 43, 47, 53, 61, 71, 83, 97 etc. That the first 40 numbers are all primes. And that no other similar numerical sequence of that length exists. :)

[@bradyelich2745]

I just finished a 32 hour straight stint at 7DTD, and am trying to wind down. Even this techno vid could not put me to sleep.

[@LightTheUnicorn]

Was looking forward to seeing this, super nice to get some detail on what was quite a curiosity for the PET!

[@mspeter97]

The funny thing is that I've just finished watching Adrian's PET restoration series. As always, very informative and interesting

[@thedungeondelver]

Adrian and Shelby - in the same video? YES PLEASE!

[@redstickham6394]

Interesting story. I heard a similar story about Atari. The story was a student typed some stuff on the keyboard of an Atari 400 and smoke started coming out of where the cartridge was and ended up burning it out. I found that story hard to believe of if it was true, it was probably just a coincidence, i.e. the smoke had nothing to do with what was entered on the keyboard. I had the Commodore VIC20, then the 64, then 128. All great machines that will be missed.

[@cbmeeks]

Loved the "don't call me surely" reference. :-D

[@offperception]

Great video and collaboration. Very informative and live that Lesley Nielsen joke there.

[@jensmaa]

Uhm, you have to poke alternating values very fast into the register :)

[@k_kon131]

Very well made video, I read André's post a while ago, but now, it is much easier to understand what the Killer Poke does. I still would not want to try it on my 2001 PET or 8032 :D

[@TaswcmT]

You should try to patch that hole in the monitor bezel - it should probably be possible to make it look like new. The 4032 is a beautiful machine - I remember it from the Canadian movie "Hide and Seek" about a rogue AI named P-1. My old school (Norwegian equivalent of high school) had a classroom with at least one (think it had a printer hooked up to it). I only found out visiting 5 years after I graduated, as my class used Apricot MS-DOS machines and some Sanyo IBM PC XT clones. When I visited in 1994, I also observed IBM PC XTs in a basement, where they seemed to still be used to run some "typing tutor" software.

[@PaoloMarcelli]

Very nice video :) and good for the collaboration with Adrian, I've been following both of you for quite some time now. Do you think you'll ever make other videos on typewriters? Those are really fascinating to me and I'd love to see more. Same story for mechanical calculators and other stuff.

[@jerryvr]

Love the Serious Sam shirt!

[@alierengam1749]

Such quality video. It's hard to find such videos these days

[@ikonix360]

Another option would be to cut the trace to that pin and insert the proper value of resistor that will permit proper operation without the poke, but allow the poke to work without the chip sinking too much current and without reducing the signal.

[@jack8407]

the research is top notch level!

[@tedvanmatje]

Although it doesn't kill the PET, Adrian was seemingly hesitant when he gingerly typed in the command and hit enter, The killer poke does sound like something Bruce Lee would've used, lol Thanks for posting brw! First video of the day - the best way to kickstart the brain in the morning :)

[@TaswcmT]

The poke that worked at 09:40 didn't work at the second attempt at 15:28? I first thought that was a different PET, as nothing was said about the fact that it had already worked on the same computer. Just proves that this poke isn't entirely harmless, even on the oldest revision.

[@Adam_Lyskawa]

You could make 8-bit Atari computers a little faster or slower with pokes messing with the display chip. Shame I don't know why it worked that way. But it definitely was the reason the screen showed those horizontal lines during loading programs in turbo. It was made to show anything on the screen without actually using normal display mode. So it was just the graphical chip drawing raster and data read was used to set the color of pixels of whatever area was actually drawn. There was no synchronization between setting of pixel colors and the raster drawing, so the pattern on the screen was constantly moving.

[@magreger]

Thank you for this in depth look at the death poke. I've been intrigued ever since I heard The 8-Bit Guy mention it in an episode.

[@TheMovieCreator]

Some of the IBM monochrome monitors are quite prone to incorrect sync signals. The IBM 5151 does, for example, not have a phase-locked loop on the horizontal drive. Jim Leonard has an account of a friend of him frying something in the CRT-driver circuits, possibly the horizontal drive, of up to several IBM 5155 portable PCs.

[@8bittimes]

In fact, with the CRTC in the PET you could very well set the timing of the HSYNC to strange values. The game in the last sequence of the video uses part of the old BASIC's interrupt routine in ROM, but when run on the shown 4032 with CRTC, jumping to the same address sets the horizontal total characters to zero - which means switching off the HSYNC as it seems. Other values may drive the HSYNC with, say, higher frequencies that may in fact damage the CRT.

[@kingneutron1]

Excellent technical explanation. / never had a Pet // still interesting

[@monkeyman767]

13:33 I had the subtitles on, I see what you did there with the Shirley joke, and I very much appreciate it

[@iCatchLupin]

I have a PET 4032 just like this one! It's not good for much, but it's such a wonderful piece of art.

[@TestTest-zt1lx]

5:41 Nice Jerobeam Fenderson reference

[@survivor303]

Compatibility isnt a issue, if you run exactly same system as i have, says devs back in the day . And read the manual, there was documented all the things what you need modify before running the applications :D

[@ejonesss]

one way i can think of that the killer poke could cause damage is if the poke happened while writing to disk or tape and the screen went goofy people would power off the machine and corrupt the files on tape or disk.

[@apl175]

It reminds me of an old SVGA tube monitor I used to have that stated that in XGA resolution (1024x768) interlaced 50hz the monitor did not meet FCC Class B Certification and as such that mode was only intended for use in professional or commercial environment where FCC Class A was acceptable.

[@marcocet]

Great video! Any updates on the Data General?

[@pawanyr360]

So quick question - if the poke doesn't work on Adrian's PET, how did he run your test program faster and with all the static the second time?

[@GatlingNG]

At 19:09, he grounded the pin in hardware.

[@TechTangents]

I addressed this towards the end of the video, if you disconnect the I/O chip's pin from the circuit and directly connect it to ground it is effectively the same as running the poke but done in hardware instead. There is no functional difference, I felt it was better to demonstrate the poke as if it were working and then go over the issue later.

[@FlamingPhoenix40]

My initial comment was wrong, I hadn't finished watching yet. At the end Shelby explains Adrian had to simulate the poke command by manually pulling a specific pin to ground. Why it wasn't working with the actual poke command, they both don't know, but they suspect something might be wrong with Adrian's machine causing the poke to do something different.

[@joshm264]

I've been a subscriber to Adrian and thought it was cool to see him get some publicity

[@SikSlayer]

I'm LOVING this channel. Another great video, and didn't expect a cameo from Adrian, nice surprise.

[@arado240dd]

if the 7408 AND gate, sources little current when a "!", but can sink lots of current if a "0".

[@Jenny_Digital]

Just a thought, but whilst the killer poke isn’t really a killer on a stock PET, what I wonder is what would happen if the 6522’s pin was configured to logic 1. The old NMOS 6522’s output was significantly weaker as a source than a sink.

[@Ekasmer549]

Omfg the green advertisement status bar is brilliant man. Subscribed just for that. Brilliant mate!

[@AmyraCarter]

10:16 Looks like white noise... Like messing about with the vertical size potentiometer on the rear of older monitors/TVs, in a way.

[@electronash]

Watched the second half of the vid a bit later on, clicked on it, and Adrian appeared. I was very confused for a few seconds. lol

[@brianclimbs1509]

Of course, the oscilloscope deflects the electron beam with an electric field rather than a magnetic field so it can respond to faster signals.

[@Sparkette]

Is the "persistent phosphor coating" you mentioned the same as the stuff that makes things glow in the dark, just with some filter to prevent regular light from making it glow?

[@immortalsofar5314]

I remember reading an article in a short-lived C64 magazine which claimed that the 6502 could be switched between two frequencies (ie sped up) with a single program, which I dutifully typed in, disassembled and realised it broke down to a single command - poke 205, 10. All it did was double the rate at which the cursor flashed!

[@tetraquark2402]

Peek and Poke felt like magic at the time

[@drdysl3xia795]

14:15 it would be interesting to use a thermal camera to see the differences in temps on the motherboard.... before and after. On the C64 the graphics chip already had heat problems. Could that Poke causes weak chips to fail on the Pet from a light increase in heat? Great video.. thanks. I learned lots.

[@drdysl3xia795]

OK if I would've waited my question was answered.. kind of. I still would like to see a thermal test done. Again great video.. you both rock. 8 bit forever.

[@jasonjason5325]

@@drdysl3xia795 interesting but they are not responding

[@truedeth]

Great video. It's endlessly fascinating to see all this info on old PC's in a non dry and boring way.

[@proxy1035]

i wonder what would happen if you were to add a resistor from pin 15 of the 6522 and the rest of the line so that it wouldn't "drain" the signal directly to ground... i assume it would squish the lines not as bad

[@JuxZeil]

The PET 2001 the chap had also had more memory. Probably the memory address allocation table on the ROM is slightly different if it's an upgrade option that the machine had. 👍

[@HelloKittyFanMan]

So what's the original value in that location, and what happens if you poke that back into it? And what about all the other values from 0-255?

[@lauratiso]

I'm here since the beginning and I need to say how I like how this channel has grown. Best kudos, Akbkuku!

[@hene193]

I wonder if the sink current is just the maximum the chip can handle by the spec which is 1.6mA and the actual sink current is higher? Could that have lead to broken chips and dead systems? Afaik how the sinking works on modern chips is the current it passed through a resistor. Which can be a common one for the package or the pin. Now if the signal is constant 5V and the chip is designed for it then it shouldn't be a problem because of the internal resistor. But now the CRT signal is analog which could mean the voltage is higher and leads to higher sink current and which leads to heat. But it could be that the voltage is low and it's fine.

[@Brpwndood]

My second computer in high school. The first was the smaller pet with the chicklet keyboard,

[@chupathingy5862]

4:00 almost felt like a Technology Connections bit.

[@flatfingertuning727]

As a different example of a POKE-related myth, it's well known that on the Apple II, evaluating PEEK(-16336) will cause the speaker to click, and I've read a gazillion books or magazine articles stating that POKE -16336,0 will cause two clicks. Although "sta (ind),y" instruction used to implement POKE -16336,0 will access the speaker's I/O port twice, the two accesses are less than a microsecond apart. Consequently, the effect isn't two clicks, but no clicks. Somehow, though, the myth got printed in a gazillion books.

[@saganandroid4175]

You're very vague: "earlier" and soforth. 11:26 can you update this to tell us which PET models predated this 1980 fix? Or where 1980-onwards editions of the same earlier-model-number PETs undated with this new design?

[@polybius223]

Something about this video is really confusing me… At 9:35, Adrian says this, and then there’s a barely noticeable, but still visible ‘cut’. Here it shows the POKE making the computer faster. Yet later in the video, this “clip” is reused 15:23 , but there’s no cut, and it shows the POKE not working the same way as it did moments before, therefore contradicting the earlier clip. Can anyone explain this?

[@abrahamcorrales3214]

Wow! It was good to see Adrian here. Thumbs up and cheers from México.

[@martindejong3974]

My recollection of why the "killer poke" was dangerous was that it was valid only for later PET's that had video logic that was based on an 6845 CRT controller, and the story was that you could manipulate a register of the 6845 so that it increased the frequency of the horizontal sync, this would increase the frequency of the horizontal sweep gnerator which would increase the voltage of the high voltage generator to a point that it dramatically started to spark inside and that COULD damage your PET. Maybe it isn't the same poke shown here, just that such a poke could exist on a PET. But then the 6845 was used in many other video cards especially those used in PC's, like the CGA, EGA and many other video cards, and I have never heard of a killer poke (or OUT command) for these so perhaps it was an urban legend after all. by the way, the poke that drove an I/O pin low against an 7408 could be simply solved by adding a small resistor (470 Ohm or so) so that the PIO's output or the 7408 would not be damaged by this I/O conflict, the 470 ohm would do nothing when the I/O was used as input, but when reprogrammed as an output the 470 Ohm would significantly lower the output current outputted by the I/O pin.

[@guilherm502]

6:19 Ahhh, THAT'S how Sync signals work! I never thought much about it, but I always wondered... how does the tube controller "knows" how fast it should draw the line? Is the Vertical Sync signal analog? The reason was: I thought they were issued at the end of the line in Horizontal Sync, and analog, as, directly controlling the raster line. I never realized: When It's pulled high, it makes the "dot" go... It's pretty dumb, I know, but again, I never put much thought into it... Thanks for that and the awesome videos!

[@jenselstner5527]

Adrian with a RPN calculator on his smartphone, fantastic!

[@strangeluck]

So the killer poke on some machines causes a short to ground and so the source chip is driven to it's current limits trying to keep up. Not designed to run this way, some chips off the assembly line tolerate it better than others. (Omg, it's not because one machine had the killer poke run often.) Just because nether computer smoked immediately doesn't mean this isn't awful to do your computer. Myth not busted at all, imo.

[@adriansdigitalbasement]

The term was coined because of possible monitor damage on later PETs, it was not called the killer poke on the original 2001 machines. Even if it was bad to short the chips like that, countless people used this all the time to take advantage of 2x graphics speed-up in BASIC.

[@supercompooper]

TRY poke 59587, 42 hehe. Characters will wrap around backwards. Higher numbers will start making flyback squeal ! A higher number will make it pop! I blew up like 50 PETs :)

[@paragonpartners2058]

I remember hearing about this as an urban legend when I was in computer class as a junior in high school, in 1982. The way my teacher explained it, this command had the power to melt down your computer like something in a Star Trek episode.