17: Does my key fob have more computing power than the Lunar lander?

When people talk about the power of their computers, we've heard all kinds of claims:

• I have more power on my desktop computer than the Apollo had to put a man on the moon
• The phone in my pocket has more power than the Apollo space program had
• I have more power on my wrist watch than the lunar lander
• And the latest one: I have more power in my key fob than the Apollo capsule had

Are any of these true?  In this episode, we break it down and enjoy a bit of computing history along the way.

Links:

https://en.wikipedia.org/wiki/ENIAC

https://en.wikipedia.org/wiki/Apollo_Guidance_Computer

https://www.ibm.com/history/space-shuttle

https://www.nordicsemi.com/Products/nRF52840

Hosts:
Jim McQuillan can be reached at jam@RuntimeArguments.fm
Wolf can be reached at wolf@RuntimeArguments.fm

Follow us on Mastodon: @RuntimeArguments@hachyderm.io

If you have feedback for us, please send it to feedback@RuntimeArguments.fm

Checkout our webpage at http://RuntimeArguments.fm

Theme music:
Dawn by nuer self, from the album Digital Sky

Chapters

• 0:00 Intro
• 4:08 Feedback
• 10:35 Main topic
• 21:48 Mainframes - ENIAC, IBM 360
• 27:55 Mini computers - DEC PDP-7
• 28:42 US Space Program
• 36:47 Microprocessors
• 52:49 PDA's and Smart Phones
• 57:22 Current generation of CPUs
• 1:03:45 Key fobs
• 1:04:50 Conclusion

Transcript

Wolf: 0:05

Welcome to another exciting ep well, for me anyway, exciting episode of Runtime Arguments. I'm Wolf, and as always, I am joined by my very good friend and co-host Jim. Say hi, Jim. Hey Wolf, how you doing? I'm doing good. Um it's been uh na this, you know, sometimes it feels like uh we record and then boom we record again. This time it felt long. Uh yeah, it kind of is considered two weeks. Yeah. Um and speaking of two weeks, how how was your two weeks?

Jim: 0:50

Oh, busy, you know. Um I I say that every time, right? But uh do you ever have a day where you are so productive? Uh I I had one of those uh yesterday, and it kind of stretched into today. But I think I got more done yesterday by noon than I did the entire week before. Uh some days just drag on, but yesterday, man, I was firing on all cylinders, and I just I I I you know, it in the zone is where I was, and it was great. I gotta figure out how to do that more often.

Wolf: 1:22

Yeah, I love it when that happens.

Jim: 1:24

How about you?

Wolf: 1:25

You been busy? Uh I have, and um I guess two interesting things this week. One, I I have been working on Advent of Code. Um I've been working on day one uh because I'm writing a pair of Python I want to do it in Rust too, but I've been writing a pair of Python implementations. One is here's what you'd really do, and it took me, I don't know, 10, 15 minutes to write. Um, and one is this is an example of how you would do a big project. It has pre-commit, it has tests, it has a type checker in it, it's arranged like a Python package. It's it is a teaching project with comments and asserts and everything that you would want in a big project. Uh what it builds is uh useful if you were planning to solve a whole bunch of projects that all worked on the same kind of things, dials and rotations and stuff like that. Really, you'd use the first thing, the second thing is to teach you about Python. So that's one thing.

Jim: 2:39

And by the way, you're taking that seriously.

Wolf: 2:42

I really am. And and by the way, um all the code in there I wrote by hand, it was all me. But the second thing from the pair of weeks we just went over is uh I have been using Claude. I use Claude for lots of things, helping me with tests and variable names and comments and whatnot. I gave Claude Code a try, and Claude Code is really nice. It's quite a step up from regular Claude. The thing that um AI normally does when it helps you solve problems is it gives you a solution and you don't like it, and you ask to change something, and then it gives you a whole brand new piece of code that's everything all over again but different. But it's the same thing. Claude code um makes the changes you want and then shows you a diff of what it gave you last time and what it wants to give you this time, and the diffs are small, it doesn't rewrite everything. So Claude Code, uh, I'm super excited by. And uh you can install it with homebrew. I mean, there's lots of ways to install it, but yeah uh home homebrew on Mac andor on Linux, pretty pretty awesome.

Jim: 4:02

That sounds so that was my two weeks. That sounds like a game changer. Sounds interesting.

Wolf: 4:07

It really is. Um so we talked about our week. We did get some feedback slash follow-up that I wanted to address, and that is um someone asked, we had mentioned, I believe, uh Richard Stallman and his feelings about open source. Um and we got some some feedback about that. They got the impression from what I said that if Richard Stallman had his way, there wouldn't be open source. And they thought, well, that's exactly backwards, isn't he kind of the guy who invented it? Which I think is hilarious if you put those words together exactly the way they are. And if you said that to Richard Stallman, he might puke. Um I actually know Richard Stallman, not like a friend.

SPEAKER_01: 5:03

Yeah.

Wolf: 5:04

Not um I'm not not actually sure he has friends, but um Richard Stallman is the founder of this idea of code shouldn't cost money. And his version of the idea is called free software. And it was a huge change when he brought this idea to the world because before then, um, you didn't show other people your source code, you charged them for it, and what you gave them was a working program and they never saw your code. And then he had this idea that if your printer does things, you ought to be able to control it. Um and here's the code. Uh, and it's yours for free, but you have to promise that if you make any changes, uh maybe you'll give them back. Maybe that was one of the promises that uh was a possibility. Uh, but certainly you will not re-release the code and then charge for it and not say where it came from. And uh he called this free source. And he had a special license called free software. The he uh had a special license called the GNU um the GPL, the the GNU public license. Um and then when open source was introduced, he went nuts because it was absolutely against everything he made free source for. If you use somebody else's source, open source, you don't have to release it. You don't have to give back changes. Um you can just use it and publish something that doesn't show anybody the source.

Jim: 6:58

Okay, wait a minute. Wait a minute, wait a minute. Open source is a huge term, and there's lots of licenses. Uh the f the GPL is one of them. The MIT license is one of them. The I don't know what else there is. There's a bunch of license, the Creative Commons license is part of it. So uh it's not that you don't have to give back your your code in open source, it depends on what license you pick. And if you pick the GPL, then you're free software compliant.

Wolf: 7:28

You are, and you have to uh give up your source to someone if they ask to see it. And that's why a lot of companies uh describe the impact of the GPL on their c products as poisoning.

Jim: 7:45

Um they use it.

Wolf: 7:49

They use other things. They use the LGPL or they use the MIT license or um, you know, whatever. And that's for instance why uh cute, which some people mispronounce as QT, you can get it under two licenses. One is you can get it, I think it's the GPL or the LGPL, uh, and you don't pay anything. And if you do that, it's because your product is going to be free and you're going to show your source to anybody who asks, or you can buy a commercial license and it costs money, and you don't have to show anybody your source code. Um, and that's kind of an interesting duality that's super useful. Uh, and there's a lot of people who really hate that idea, and Richard Stallman is one of them. So the answer to the follow-up question do you think that if Richard Stallman had his way, there would be no open source? Well, of course, we can't know what that alternate reality would be like, but yeah, if Richard Stallman had his way, my opinion is that would mean there is no open source. He thinks you ought to be able to fix your printer.

Jim: 9:06

Yeah. Okay. Depending on what license you use again, that that's entirely possible. Uh I think a lot of people.

Wolf: 9:12

But he doesn't like the other licenses.

Jim: 9:14

Fine. I agree. But a lot of people just don't pay any attention to the difference between open source and the smaller uh uh uh idea of free software. Uh they they just lump it all together. And you know, to hear that Richard Stallman hates open source, well, they think, well, he hates free software then. That's not true. Uh your your description of it, you're being much more strict in the difference between free software and the much more liberal open source. And that's fine. And I think that's what Ron was uh getting at was uh he doesn't hate open source. He he kind of does, but uh it it's a i it's a matter of semantics, really.

Wolf: 9:58

Um to Richard Stallman. Oh, sure. He's all about semantics and names.

Jim: 10:05

And apparently it's important to you too.

Wolf: 10:08

Uh it kind of is. Yeah, yeah. All right. So that was the feedback that I wanted to talk about.

Jim: 10:15

Um, that was feedback to the previous episode on uh uh uh essential tools and and knowledge and stuff. So yeah.

Wolf: 10:24

All right. Um so this is it. This is the time for the meat of the episode. Uh, and it's you, Jim. You're gonna talk about stuff today. Uh, what are we talking about?

Jim: 10:35

Well, uh a couple of months ago, uh I got a new iPhone, the iPhone 17, because every couple of years I uh I I'm due for one. So I got it and I made a comment to Wolf at lunch one day, something about uh something to the effect of I've got more power in the palm of my hand than the space uh uh the NASA uh Apollo lunar lander had when it put a couple of men on the moon. And and I've used that analogy uh for a lot of years for various pieces of uh computing equipment. And Wolf sort of pointed out that uh, you know, your key fob has more power than the than the Lunar Lander had. And uh and I thought about that and I thought, well, is any of that true? You know, do I really have more power than the than the Apollo uh uh Lunar Lander? Um I I've heard variations on that as well. Like um I've got more power uh on my desktop uh than it took to put a man on the moon. Um uh I've got more power on my wrist uh than the lunar lander had, and um and of course the one that Wolf uh uh pointed out uh about the key fob. Uh so let's talk about it. Uh what do we mean by more powerful? Uh what does it mean to talk about the power of a computer? There's a lot of things to discuss here, uh a lot of terms. Uh the uh you know, one of the things is clock speed. How fast is the clock ticking uh in inside that computer? Because every click of the top uh, every tick of the clock is uh usually something's gonna happen. Uh and then you talk about word size. Well, what is the word size of the CPU? You know, 8-bit, 16-bit, 32-bit, whatever, even very large word instructions could be 128-bit or more. Uh currently we've got a lot of 64-bit computers. Uh, what is that? That's the kind of the size of the instructions uh and the data that can be operated on uh within the CPU. Uh obviously, an 8-bit uh CPU is not as powerful as a 64-bit. Uh and you talk about bus width, uh, that's how fast can it get the data from memory uh across the bus. Uh it could be an 8-bit uh uh path, uh 16, 32, 64. Uh doesn't always match the CPU. Uh that's kind of interesting. Um how much memory can be addressed? That's a measure of the power of the computer. Uh some computers can only measure, can only access like 4K. Uh other computers can access terabytes of RAM. Um modern computers, really, for quite a while now, they have caches on the chip. Uh, you got your L1, L2, and L3. Um, I I did a little bit of digging and I'm going to describe exactly what those are. Uh L1 cache is the smallest, fastest memory built into the CPU core. Uh it's for data needed right now. That's the data that the CPU is processing right this second, uh, or nanosecond. Um, it's typically 32 to 128k per core. Um, so it's not huge. You know, you're not going to operate on your entire problem in that kind of space unless you have a really tiny problem. Uh, and the access time on L1 cache is usually one clock tick. Uh L2, level two, uh, it's larger than L1. Um uh not as fast as L1, but faster than RAM or L3. Uh it's usually 250 256K to two megabytes. Um, it's usually on the same die as the CPU. Sometimes it's per core, sometimes it's shared between cores. It's for data needed soon. Um, not as soon as the L1 stuff. And then the L3 is the the last cat. In fact, some computers have L4, but L3 is usually shared across uh cores, uh, and it can be a couple of megabytes up to over a hundred megs. Uh in fact, there's an AMD chip, the Epic EPYC. It's got a little more than a gigabyte of L3 cache. Uh that's a lot. Uh and measurement of the sp of the processing power of a computer. You got MIPS. I'm sure a lot of the old guys here, uh Wolf and myself included, have heard of MIPS. That's millions of instructions per second. Uh, it's really kind of an outdated uh performance uh measure. Um because you know what's going on in that uh instruction? You know, some not all instructions are created equal. Uh uh an add instruction is really simple, a divide instruction is crazy complex uh and takes a lot more to process. Uh a newer measure, or it's not even that new, flops, floating point instructions per second. Um, or floating point operations per second. Um that's a lot more work, you know. Doing multiplies and divides and floating point operations takes a lot more. So uh a lot of computers now will will give you a rating in flops, and you know, it used to be megaflops uh and and then gigaflops for uh billions of floating point operations. Uh now a lot of the computers are measured in teraflops, trillion floating point operations per second. Um and then you know, if you're on Linux, uh and if you've booted up and see the seen the boot screen uh as as all the messages fly by, you don't see that so much anymore with all the artwork and stuff. Um but there's something called BOGO MIPS. Uh have you heard of that, Wolf?

Wolf: 16:16

I I haven't. And and actually before you dive into that, I have to say, um when I was a young man, uh maybe in my twenties, yeah, uh, and my father uh my father and I actually had a couple conversations about computers, and he was, if I'm remembering the sizes right, um he was super excited about the idea that sometime in his life we were gonna build computers that crossed the megaflop boundary.

Jim: 16:55

The megaflop. Um that's but I have never heard of a bogo. It's funny because I remember that happening. It uh back in the 80s that happened. Um uh anyway, BOGO MIPS is not really a measurement at all. It's kind of a measurement. It's the it's not a good measurement of performance, it's the number of million times per second the CPU can do absolutely nothing. Uh uh, it has to do with clock speed and various other things. And it's calculated by the Linux kernel to calculate an internal busy loop for timing. It's just used for timing. Uh, in the Linux kernel, they use something called jiffies, which is something like 50 milliseconds or something. I I think it's changed recently with the faster processors, but it's not a good measurement of performance. It's just, you know, it's how quickly you can do nothing. Um anyway, that's kind of some of the terms we're gonna be throwing about as I get into some of these other uh uh systems, CPUs. Uh and of course, if we're talking about performance of a computer and and stuff, we got to talk about Moore's Law. Um you've probably heard of it before. It was coined by Gordon Moore, uh, one of the co-founders of Intel.

Wolf: 18:12

Although he didn't use the word law, he made an observation.

Jim: 18:17

Yeah, basically, it's an observation uh that the number of transistors in an integrated circuit doubles about every two years. Uh his initial observation was it doubles about every year, but in 70 1975 uh he updated it to say about every two years the the uh the transistor count doubles in in an integrated circuit. And uh I think if you look, I saw a graph and it tracks pretty well. Uh and a lot of people take that to mean that the power of a computer will double uh every two years. You know, if you double the transistors, is the CPU really twice as powerful? It's hard to say. Uh anyway, that that's Moore's Law. You've probably heard it before. Uh, but let's get, you know, if we're gonna talk about uh computers and performance, let's go way back to the beginning. Um I I've got a couple of notable computers here to talk about. Um and believe me, I really, really enjoyed doing the research on this. Uh I dug into uh the first general purpose uh programmable digital computer, electronic, uh was the ENIAC E-N-I-A-C. Um the electronic numerical integrator and computer. Uh completed in 1945. That's a long time ago. I I don't know what is that, um 80 years ago? Um uh anyway, it was initially created to calculate firing tables for artillery for the U.S. Army. Um I was really surprised to read that, and it kind of reminds me of something. You know, uh my father was in the Korean War. Uh he was in the tail end of it, never actually saw any action or anything. But his job, he was The Marines, his job was to uh calculate uh trajectory for artillery. Um, and that is like what angle should they shoot the uh the mortar at uh in order to make it hit its target at uh you know 1.6 miles or whatever. And to do that, uh he didn't have a computer, he had a slide rule. Can you imagine? I I've I've tried to use a slide rule, and it's just pick up my calculator, it's so much easier. But can you imagine that using a slide rule to figure out how the heaters are?

Wolf: 20:36

I I do want to say one thing, um, and this is not nitpicking. Um, it's actually kind of a statement about the British government. And that is um, as you probably know, they had some super smart people uh solving uh trying to solve Enigma uh over at the British government and and I guess here in the US I can't remember Bletchley Park was a place where um they did this work, and my understanding is uh from reading that book about Bayes Rule, uh the algorithm that uh wouldn't die, is that they had real electronic computers before ENIAC. Really? And they destroyed all evidence of these computers. Right. Which is man, is that sad. That's like not using source code control.

Jim: 21:40

They had to they had to disavow any knowledge of any of that stuff, so they destroyed the computers. That's awful. Anyway, probably said sorry to interrupt.

Wolf: 21:48

I have used the slide roll. That's the answer to that question.

Jim: 21:51

Yeah, yeah. Uh so the ENIAC, um it uh I I'll just give you a rundown of uh uh what was in it. Uh it had 18,000 vacuum tubes, 7,200 diodes, 6,000 relays, these are mechanical relays, 70,000 resistors and 10,000 capacitors, 5 million hand-soldered joints. Can you believe that? Five I struggle with one. Uh uh they didn't use the transistor because it hadn't been invented yet. Uh the in the transistor didn't come along until 1947. Uh that computer weighed 27 tons, was 10 feet tall, three feet deep, and 100 feet long. External storage was all on IBM punch cards. Remember those? I I I I used punch cards when I was in college. That's how we input our computer programs.

Wolf: 22:45

Yeah, me too. I mean, I I wasn't actually in college. I worked at the college. And yeah, exactly like you, punch cards. Yeah.

Jim: 22:54

It had a one kilohertz clock. That that just amazes me. That clock ticked at at a thousand times a second. It's it's insane. It could do 5,000 simple addition or subtraction operations per second. I don't really know how they did that. 5,000 operations with a clock that ticked uh only a thousand times a second. Um, maybe some pipeline, I don't know. 385 multiplication operations, only 40 division operations. Remember, I mentioned earlier division is much more tricky than a simple add or subtract. Uh it could do three square root operations per second. Now, all those operations, you know, if you're trying to calculate uh the trajectory for artillery, uh those are all the kind of things you needed to do. Uh I I'll give the rating of MIPS a lot of these computers that I talk about. This thing had a 0.00289 MIPS. So that's um that's not very fast, but hey, it's faster than uh a human with a calculator or a human with a pad of paper and a pencil, right? Or a slide roll. Yeah, or a slide roll. 28 what I say, um uh 18,000 vacuum tubes, several tubes would burn out almost every day. Uh programming was performed with plug board and switches. There was no there was no terminal or you you used uh Vim to edit your source code. Uh it was it was all done through plugs, boards, and switches. In fact, they would spend weeks, uh, many, many weeks just figuring out what the program was going to be. And then it would take days to actually program the computer with that. And before they would take the time to set the the jumper cables and the and the switches and stuff, they would go over that code over and over and over again manually to make sure it was correct before they took the time to put it into the computer. Because remember, there was only one of these things. So the time uh, you know, they wanted to be really efficient with their time.

Wolf: 24:58

Um and that's why they were generating tables and not individual answers. Yeah, they said, hey, what angle right now?

Jim: 25:07

Yeah. So they'd give you this table and and uh you say, oh, because out in the field, you know, you don't have a uh satellite connection back to the computer. You have this piece of paper in front of you. Uh anyway, uh but this is 1945. That's like the end of World War II. But you know, they were doing it. Um the there are pieces of the ENIAC on display in various places. The Smithsonian uh uh National Museum of American History, University of Pennsylvania, several other places. Uh you can see bits and pieces of the ENIAC. Uh a hundred feet long. That's kind of crazy. So that was plenty of pieces. What's that?

Wolf: 25:46

Plenty of pieces to go around.

Jim: 25:48

Yeah, plenty of pieces to go around. Yeah. Um, so that was like the first real notable computer. Um I I think it's kind of neat. Uh the next one to me anyway is the IBM 360. That was a it was launched in 1964. It was a whole family of computers. Um the the first one was eight, eight-bit, uh well, it used eight-bit byte, but 32-bit words. Uh it was the first to use microcode. Uh, and that is uh, you know, beneath the hood of your computer, uh, you think you're running instructions, instructions that are just built into the silicon, uh, but it's not really the case. You're running microcode, and that microcode is running uh and converting that down into the actual transistor gates and stuff of a of a CPU.

Wolf: 26:36

Is that true today? Yeah, still, very much. I did not realize that. I thought I knew about microcode, but I thought I thought we were done with it. Oh no, everything's using that now.

Jim: 26:49

Uh unless if you get into like an FPGA or something. Uh but if you're talking about a CPU.

Wolf: 26:55

I think I learned about microcode in the book um uh Soul of a New Machine by was it Stacy Kitter? Is that the guy who wrote it?

Jim: 27:07

All about um I I I I don't know what you're talking about, but sure. Okay. Okay. Uh the original Model 20 uh IBM in the 360 family uh had only 4K of RAM uh in it. It operated at 0.0018 MIPS. So that's um uh 1800 uh instructions per second. It's not a lot. Uh the whole thing weighed 1700 pounds. It's quite a bit smaller than the ENIAC that was uh what was that uh uh uh 27 tons? Yeah, so they shrunk it down quite a bit. Only only uh uh point one uh uh not not quite a ton. Um anyway, that was that was the 360. Uh uh certainly notable. Uh the next one to me is uh the deck PDP7, and uh that was introduced in 1964. Um it's notable, at least in my mind, because that's the computer that they created Unix on in 1969. Um wasn't a very powerful computer. Uh it did not use integrated circuits. Uh it was an estimated 10,000 individual transistors. Uh uh oddly, it had uh uh 18-bit words, and uh it could have up to 64k of memory. And that's 64k words, so it's not bytes, but anyway, you get the idea. That was about a half a bip. 500,000 uh instructions per second. Big jump from that IBM. Um okay, finally, we get into the really interesting stuff, and this is where I spent a lot of time doing the research. The United States space program. Um I'm gonna skip over all the early ones, the Geminis and those things, get right into the the the uh the the vehicle that took the astronauts to the moon, because that's really what this conversation started out as, right? Uh so the the the the the rocket that took the uh took the guys to the moon, uh it was uh really had three parts. There was the Saturn V booster, that was the big tall rocket that you saw on the launch pad uh with the huge tanks and the and the giant engines on uh on on the uh on the end of it, um, that had computers. Um it had uh if you look at it, um there's a ring up towards the top just before uh the payload, like at the top of the the stack. Uh there's a ring that contains all the computer stuff. And and I say all the computer, there wasn't that much. Uh the the computer in that thing was uh designed by IBM. Uh it used it was not in the 360 family. Uh it used core memory. Uh you've probably seen pictures of uh those little tiny ferrite coils with uh with with wires going through them, um, and and they could be magnetized uh uh either like a one or a zero. Uh that's what it used. Um it had a two megahertz clock. Um uh the instructions on that thing were 32 bits, and the data was 26-bit words. That's weird. Uh 30 32k of words uh in that magnetic memory. Um a simple instruction like an ad took 168 clock cycle cycles. Oh my god. I think we're doing ads now in less than a cycle. Um uh there were yeah, because we got the pipeline. Yeah, now we have the pipeline, so we can actually in a clock tick, we can get a lot more done. Um uh but the longer uh uh instructions like the multiply and divide took considerably longer than the 168 cycles. Uh it could do about 12,000 instructions per second. Um it was primarily used for guidance, basically controlling the engines, uh the the gimbaling of the engines to uh so that it would it would go the right direction, right? Uh uh there was actually it wasn't just like light the light the candle on fire and watch it go. There were uh there were engines that that that would uh swivel and stuff to to head that thing in the right direction. Um there were three logic modules inside the computer, uh, so they had triple redundancy. Um and it was like a voting system. If uh as long as two of them agreed, it kept on going. If if they didn't agree uh on the answer, uh the emission would be aborted. And that's kind of tough to do when you're a couple of miles downrange and and you know cruising along at uh several thousand miles an hour. Um uh fortunately that never happened. They they they never had to abort because of that. Uh that computer weighed 72 pounds, so we're getting smaller. Uh and it was liquid-cooled, which is which is kind of neat. Uh certainly wasn't water, but um yeah, probably some kind of anti-freeze alcohol or something. Um so the next stage of the rocket, or the next uh the next major piece of it, was the command module. That's the part that the the astronauts rode in to the moon. Uh they they they rode in it the whole time until they got to the moon. Um, but that's like the capsule with this big chunk behind it containing engine and and fuel and oxygen and all kinds of stuff. Um, that used something called the Apollo guidance computer. It was designed by MIT. It was the first computer based on silicon integrated circuits. Remember, this is uh really the mid-60s, they were designing this thing. So that's when they started using uh integrated circuits. It had a 15-bit word length, uh, it ran a two megahertz, uh, only had uh uh 2k words of memory, so that's uh 4k bytes. Um the the um uh uh the core rope memory, uh which was similar to the to the the the the Saturn V booster, um had only 36k words of RAM. Um it was programmed entirely using assembly language. It was about two cubic feet, so it's still pretty big. Uh it only had 12 instructions in the instruction set, and it could do about 43,000 instructions per second that's 0.043. Um you'll see these numbers start climbing drastically. Uh so let's get into the lunar lander. That's the piece uh of the rocket that two of the astronauts got in, and and they basically the the command module and the lunar lander were joined together and they got to the moon and they orbited the moon, and then two of the astronauts climbed into the lunar lander and and they broke away from from the uh command module um and they landed on the moon. Now their computer was the same computer as the command module, the the AGS. Um it it uh I'm sorry, the AGC. Um it's got it had all the same specs, it had a different program in it, um, because it was its purpose was to land on the moon and take off from the moon. Um uh so it's the same computer, not terribly powerful. It they had another computer that was the abort guidance system, the AGS. Uh that was an entirely different thing. That was to use in case they had to abort. Uh it would basically get them off the moon. Uh, or if they were approaching the moon, it would get them out of there and back to the back to the um uh uh uh command module. Um they did a lot of testing with it. They never actually had to use it in an abort situation. Uh so that was that was good. Uh interesting, they did use it. Uh you remember Apollo 13. You've probably all seen the movie, and and you know, some of us were around back then. Uh Apollo 13 uh had the uh explosion on the way to the moon, uh, so they never actually landed. Uh they ended up uh using the uh abort guidance system in the LEM, uh the lunar lander, um, because it it required less uh water for cooling, uh, and they needed to conserve every drop of water. Um uh so they instead of the uh using the uh AGC, they used the AGS. Um less powerful computer, but it worked for them uh because uh they they did use the engines on the uh on the lander uh in order to get around the moon and get back to Earth. Um anyway. Um so that's kind of the the Apollo program. Uh not terribly powerful, and we're gonna compare that with some of the stuff that's out there now. Uh, but one more thing. I'm gonna talk about the space shuttle. Because I've said things like my desktop or my phone have more power than the space shuttle. Um the space shuttle had five IBM AP101s. Uh that was a uh that was a computer that IBM already had. In fact, uh the military was using those computers in the F-15 fighter, the B-52 bomber, and the the B-1B bomber. So that was like a flight-hardened computer. Well, the shuttle had five of those, four of them uh in sync for redundancy, and a fifth as a backup. Um originally it had the magneticore memory, but then they upgraded to solid state. That thing ran at five MIPS or 0.5 MIPS, so still only half a million instructions per second. Um on board the shuttle, they also had laptops for doing science-y things, uh, but those laptops were not controlling the shuttle. Um anyway. This was all in the 70s. Uh in the mid-70s, uh, in fact in the early 70s, in 1971, Intel invented the microprocessor. Finally, we get into the little tiny computers, right? Uh the Intel 4004. Um it was first, it was available in 1971, the first uh single chip CPU, commercially available uh single chip computer. Uh it was only a four-bit size. Uh, it was not processing very much data, ran at 750 uh kilohertz. It only addressed 640 bytes of RAM, 2300 transistors. Now we're getting all into the transistors on these things. Um, it was really designed for electronic calculators. So uh I remember when I was a kid in the early 70s, my dad had a calculator. Um I'm guessing it was powered by a 4004. Uh Intel in 72 created the 8008. Uh that bumped it up to 8 bits, 16k of RAM, 3,500 transistors. And then into the uh the mid and late 70s, a whole slew of microprocessors uh were developed. Uh you got the Intel 8080, the 8086, the 8088, uh, the Z80, the Xilog Z80. Remember that one? Uh the TMS 9900 from Texas Instruments, uh the 6502 and 6510. I'll I'll I'll list a couple of computers that use those. Um probably the kind of an interesting one uh was the Altair 8800. Remember that one, Wolf?

Wolf: 38:15

You didn't have one, did you? I did not. I didn't actually have any computer. My friends did, but I grew up pretty uh pretty poor. Oh I didn't have any computers at all until I got um my first real programming job, and that was right around the time of the introduction of the FatMac, the 512. Oh wow. Um that was my first computer. Now I programmed. My friend had an Apple II. I programmed some big computers using RP uh RPG2, yeah, and uh I did some other programming, but that FatMac was my first my first actual I own it computer.

Jim: 38:58

Wow. Well the first the first microcomputer, um uh the one that really sparked the microcomputer revolution was uh introduced in 1974. That was the Altair 8800. It had an Intel 8080 CPU. That's the computer that Bill Gates wrote uh BASIC for. Uh Bill Gates and Paul Allen. Uh they wrote the basic interpreter for that, flew down to Texas and and uh uh uh gave it to uh uh the Altair guys. I forget the guy's name that that ran Altair or the MIPS company at the time. Um but that's where that's that's really how Microsoft got its start um by writing a basic interpreter for the 8800. Uh another one that uh some of my friends had the the ZX80 from Sinclair that had the Z80 chip. Um we're starting to get a little more powerful here. That ran at 3.25 megahertz, 16 K RM, uh 8,500 transistors, and uh still it only performed about 400,000 instructions per second.4 MIPS. Moving up the Atari 400-800. I remember I knew somebody that had a uh I think it was the 400, and I thought that was pretty cool. That was using the 6502 chip. Uh ran at 1.8 megahertz, 48k RAM, 4,500 uh transistors, another uh uh half a million MIPS. Uh the Commodores. Uh the first computer I ever got to play with, I had a I had a math teacher in high school, Mr. Cox, and he had a Commodore pet. Uh P.

Wolf: 40:31

Man that keyboard.

Jim: 40:33

Yeah, the little little goofy chiclet keyboard. It was really weird. Uh all the symbols. He um uh uh he invited me and a couple other guys over to his house uh uh to to show us the computer and and teach us a little bit about uh computers. And man, I was I was smitten. I was so impressed with that thing. Um that that that's how I got started in computers, that thing. Uh I was really into math at the time. I was gonna go to school as a math major, but that computer really did it for me. Um as limited as it was, uh I thought it was the greatest thing. Uh it had the 6502 processor. The VIC 20 also had the 6502. I my first computer that I bought was the VIC-20. Um, not very powerful. You hook it up to your television. Uh, the keyboard and the the computer were all built together. Uh in fact, I still have that somewhere on the shelf down in the basement. Um, kind of a neat. I I moved pretty quickly up to the Commodore 64. That was a much more powerful computer. It had a 6510 instead of a 6502. Uh ran at the same clock speed, really. It was about the same power, but it had more things. It had better sound, uh, better, better color graphics, I think. Uh, it had something called sprites for doing graphics. So game game developers, they liked that. It was kind of neat. Um then we move up to the Apple II. Again, 6502 processor. See, there's a lot of computers out there that had that 6502. Um that was uh that was kind of a kind of a major thing. I mean, it launched the Apple company, right? Um uh you see all kinds of little videos online and stuff about uh Steve Wozniak and and Steve Jobs and that Apple II and how they did it. Uh I think it's pretty cool. Uh and of course, uh there was a Radio Shack computer, uh I forget what that and a and a Texas instruments computer using their uh TMS uh uh 9900 chip. Um lot of stuff. That was kind of the home computer revolution. Um we get into the personal computer uh IBM, right? And and the compatibles. Uh they started with the 8086, um uh and and some of them had the 8088. I uh I I went for the Commodore 64 to remember the IBM PC Jr. Uh oh my god, yes. I had one of those. Again, talk about crappy keyboards. Uh I had one of those. That had the 8088 chip, and that was kind of funny because um the 8088 was a 16-bit chip, but it only had an 8-bit data bus. So anytime I had to fetch memory, uh data from memory, it had to do it in two operations. You know, grab the the high order uh 16 bits and then the lower order, or the high order 8 bits and then the low order 8 bits. Um limited. Uh the the the the junior also had this weird thing. Um the floppy drive and the keyboard, I think, shared an interrupt. So sometimes it would get confused. It didn't really know what it was doing sometimes. Uh it was weird, but I had it for a little while. Um that thing, uh you know, we're up to 0.66 MIPS now, 20,000 transistors. Um, and then uh I think I don't know, it must have been about 83 or so. IBM came out with the IBM AT 8286 processor. Uh 16-bit, 16 megs of RAM, 134,000 transistors. We're we're getting up there now. Uh it could operate up to usually I think there was six or eight megahertz, but uh the chip itself could go up to 12.5 megahertz uh from 0.9 to 2.66 MIPS. We're starting to get into the MIPS now, not the 100,000 uh K, but the actual millions. Um 85, uh the 8386 came out. 32 bit. Woo-hoo. Uh that was that was some serious computing. Then 32 bits, 33 megahertz, 275,000 transistors, uh, and it also had uh a virtual 8086 mode, so it could run multiple things concurrently, uh uh uh sort of emulating an 8086. Um it I think yeah, it was. Uh, but it's it's awful backward thinking about it now, right? Not backward, but minuscule compared to what we're doing now. Uh but it's a big deal. Uh two MIPS. That thing was cooking along. It's you know, two million instructions per second. That's pretty good. So we're you know, here we are. Much more powerful than the than the early computers, more powerful than the than the ENIAC and the IBM 360 and the space program. Um, that's pretty cool. Uh 486 came in 89, 60 megahertz. Now we're really cooking. 1.2 million transistors. Now we're into the millions of transistors. It's incredible. 15 flops. 15, 15 floating point operations per second. Uh it's not many. Um uh 50 MIPS. 50 million instructions per second. Look at that thing. Uh, it didn't take long, you know. We went uh we're talking about like less than a million instructions per second in in uh like 1981. Uh here we are in 89, we're doing 50 million instructions per second. Still, you kind of laugh at it compared to where we're at now. Uh the Pentium, remember when that was announced? Oh yeah. That thing, I remember getting a Pentium computer at work, and that thing was so cool. Uh that was uh 1993, 32-bit again, 3.3 million transistors. Uh uh at the time, Intel had this marketing campaign, and if uh they had remember magazines? We used to always read magazines. Uh there was an article or a full-page ad in a magazine for Intel Intel. I don't understand. Yeah, you don't uh you use this paper. Oh, it's paper. You're talking about those. Yeah, yeah, you can't do that.

Wolf: 46:43

I thought you were talking about the things that go into a rifle.

Jim: 46:47

No, no, uh uh anyway. They had this marketing campaign and the full page ad, and there was some fine print at the bottom, and it said if if you respond to this by uh we didn't have websites at the time. I don't remember how I did it. If I if I um uh uh maybe I called them up or something, I guess I had to call a number. Uh they would send you a poster of the 8386, uh, and it was like a photograph of the die. And this this poster was huge, you know, it was like a standard poster size, probably 24 by 36 inches. Um and it was a very large print of the CPU. I I got that. I still have that. I framed it and I had it in my office for years. Um when I moved out of my office uh for for my work, I uh I ended up putting the thing down in the basement. I gotta dig that out because that was a really cool poster showing that 3.3 million transistors. Uh you you've probably seen pictures of it. 188 MIPS, 100 megahertz. That thing was flying. It could do all kinds of stuff, right? Um uh and then Intel kind of made a I'm not gonna call it a blunder, but it it it spent a lot of money on a project that really didn't pan out so well. The Itanium. Remember that thing? That was Intel's first 64-bit computer. Uh uh.

Wolf: 48:06

I I remember the word. I didn't remember it being a failure because it didn't intersect my life. It was a failure.

Jim: 48:14

Well, it there they didn't make many computers with it. It just they spent a lot of money on it and it just never really took off. Um, this is at a time when when you know there was x86 compatibility and that didn't have it. It was going to require everybody to rewrite their programs and it just didn't do it. Um 64-bit was released in 2001. You know, at that point the the internet was taking off, everybody had these websites, nobody wanted to spend a lot of time porting their software to another cheap uh CPU. Just it just never really went anywhere. They they they finally discontinued it fairly recently, like in 2001, uh 2021. I was really surprised. Oh my god. Yeah. I I think they had commitments. Um they they had a project uh with Hewlett Packard, so HP was making some computers with that chip. I think they had like contractual commitments, but man. Um something else. That thing ran at 2.66 gigahertz, so we're starting to get up into the speeds that we're used to, right? Um finally we get into the Apple Mac. You mentioned you your first computer was the Fat Mac, the 512.

Wolf: 49:28

Yeah, the second Macintosh they made. And and um uh there was a tiny little company called Levco. It was, I don't know, maybe two guys, maybe five guys. I think they were from Czechoslovakia. The company was named after the guy who founded it, Stash Lavak, and they made a thing that went into your Mac. They called it the Levco Prodigy 4. And what it did was it added the you know, the 512K Mac had 512K. Yeah. And it was a 68,000 or something, whatever it was.

Jim: 50:06

I think it was some variant of the 68,000.

Wolf: 50:09

The Levco Prodigy 4 raised that up to whatever the next thing was. I I'm remembering 68020, but I think that was the numerical coprocessor.

Jim: 50:22

No, 6820 was uh was a real CPU. That was the uh one of the next ones in the in the line of them.

Wolf: 50:29

And it also gave it four megabytes of RAM.

SPEAKER_01: 50:33

Wow.

Wolf: 50:33

So this was a vertical card that squeezed into the Mac, and we bought for our company, um, I don't remember how many programmers we had, I think we had six, something like that. We bought six, and I bought one myself for our own company, and they were ridiculously expensive. Uh we bought so many of these things that the president of the company himself, Stash Lavak, got on the plane, came to our office in Ann Arbor, and installed every one. Um we all worked in one room and we didn't really have working air conditioning. It kept blowing out the circuit. Um, and these things were hot. So once they were all installed, we all had to take the backs off of our Macintoshes. And when you walked into the room, all you saw was like the backs of the CRTs and stuff. And um nobody did.

Jim: 51:34

Oh, because it was all integrated, too. It was like the monitor and everything was built into the same box.

Wolf: 51:39

Right. Nobody died, but there were some electrocutions, uh, some some pretty big shocks, and I'm ashamed to say uh it was me. I got a bad shock. And then I showed somebody the bad shock. I said, Don't touch this. And you touched it. And I touched it. So I got electrocuted twice. Um yes, I was an idiot. Well, you didn't die, so that's good, right?

Jim: 52:05

I feel good about that.

Wolf: 52:06

Every time I don't die, I'm I'm super happy.

Jim: 52:09

Well, yeah. Like me, every time I every day I wake up alive, I'm happy. So uh the the Mac came out in six in uh '84. Uh used the 68,000 chip, as we mentioned. Uh that chip had 68,000 transistors, and I'm kind of wondering if that's a coincidence or if that's how they came up with the name. 68,000 transistors, uh uh 32-bit instructions, 16-bit data bus. So again, for it to uh grab data, it had to do it uh 16 bits at a time. Uh 1.4 MIPS. Um, and then you know, the Mac, it's just been growing and growing and growing since then. Uh so we kind of get into the smaller computers now, the PDAs, right? Remember the Palm Pilot? I had one or two of those. Uh General Magic had the Magic Link. Didn't you work there? Is that the company?

Wolf: 52:58

I worked on two of the things uh that you're talking about now. I worked at General Magic. Um, I wrote their entire text system. If you saw text on the screen or did editing, yeah, that's because I gave you that power. Um and the next thing that you're about to mention, um I worked in the Apple Newton.

Jim: 53:20

Yeah, that was uh uh didn't a bunch of the General Magic guys go to Apple and and work on the couple. Uh including you?

Wolf: 53:29

Uh uh I'm not sure if it was just me or maybe just me and one other guy, maybe. And I didn't really go there because it was like a huge win. Um I went there because uh I was not um satisfactory to the people who were running General Magic, which was essentially um Andy Hertzfeld and um uh Bill Atkinson. Um some things I wasn't doing well enough, and some things uh I did that were uh a little too complicated for them. They didn't like the idea, it was kind of scary to them, and so they decided maybe I would be better someplace that had more money to get it's not so bad. It was kind of a step uh step up for me. It was it was a win. General Magic eventually out in the Bay Area at the time, weren't you?

Jim: 54:29

I I did, I lived right. This is all before I knew you. Yeah. So this was mid-90s, I guess? Something like that, early to mid-90s. Yeah. Yeah. Uh okay, and then another PDA uh from way back in the day, the the compact iPack. Remember those things? Those were weird.

Wolf: 54:47

I don't think I ever touched one.

Jim: 54:49

Oh I might have seen a picture. I I remember at one of the Linux events, uh Jim Geddes and Keith Packard, they were playing around with the with the uh iPack, and they got Linux running on it, and they had X Windows running on it. Um that was kind of neat. And I there was a very high possibility that you were gonna brick it by trying to trying to load your own operating system on it. So I think that happened more than once. Um but it was that was you know, for its day, that was kind of a neat little thing. Um and then, you know, in early 2000s, I don't I don't know the date, but the iPod. Now there is a little computer in your pocket. Uh that thing, it had a 90 megahertz ARM chip. Um uh the the later ones they moved all the way up to the A10 infusion chip. But that device right there probably saved Apple from bankruptcy. You know, there's a few devices along the way. Um uh the the iMac uh was was uh important to Apple. But that that iPod that that um you know that was a pivot for them. And uh I I think that sort of led them on the trajectory that they're that that they eventually ended on.

Wolf: 56:03

Um I gotta agree with that.

Jim: 56:05

Yeah, that's I I had one or two. I uh they were pretty cool little devices, way better than the goofy little MP3 player I had before that. Uh really, really powerful.

Wolf: 56:15

Way better than even stuff that came after, like, and I'm gonna say the word, the zoom.

Jim: 56:22

Yeah, they tried. They they tried. Uh, but the iPod that they had it, you know. The marketing and everything was really good for that thing. Uh so finally we move into the smartphones. Uh what was it, 2007 that Apple uh introduced the the iPhone? Uh and of course, right around then, uh the Android devices, and really before that, the BlackBerries. Uh they all had computers in them. They they were little handheld computers. Um the chips that they put in those iPhones now. Uh, you know, this whole talk started uh uh because uh I had made a comment about my phone uh have having more power than the than the than the uh uh Lunar Lander. Um they got the A19 Pro chip in those things now. That thing is a beast. I'll talk more about that in a minute. Um it's just incredible. So you look at the power, you know, compare that back to the the ENIAC or the or the PDP7. Uh it's amazing. Uh so let's talk just a little bit, uh we're getting near the end here. Let's talk just a little bit about the current generation CPUs um that you might be familiar with. Um it it uh if you're looking for a server CPU and you want a really good fast one in the x86 world, uh the Xeon uh 6980p, that thing is a beast. 128 cores, 256 threads, tens of billions of transistors. Uh Intel didn't it won't say exactly what's in it, but it's estimated. Tens of billions. That's a lot. Uh if you're more likely to see a chip like the uh core i9 1490 KS. Isn't that the chip you said you're trying to get in your laptop?

Wolf: 58:06

Was that the um the one I've specifically requested is the core ultra 9275 XH. Okay. Supposedly that's supposedly that's faster than the i9, but they don't they don't give the exact like it doesn't say 14900 or whatever.

Jim: 58:27

Well that okay, that core i9, um 6.2 gigahertz clock, 24 cores, uh 32 threads, uh 0.8 teraflops. Now we're talking about uh uh 800 gigaflops, 800 billion floating point instructions per second. Um that's incredible. Uh uh uh compare that with the uh uh back in 19 uh 83 there was the CDC Cyber uh 205. That was uh one that uh uh CDC was a computer company way back when. Uh and and they were they were happy that they were getting 800 megaflops. Now here's Intel with a chip doing 800 gigaflops, it's a thousand times faster. It's incredible. Um it's it's nuts. Uh uh of course, then there's the ARM series. There's an awful lot of equipment out there now uh with the ARM chip. The the the the the iPhones are are you know they're the the M chips and the A chips, they're ARM, uh kind of an ARM core, but Apple's kind of you know turned the crank quite a bit on them. Uh ARM was introduced in 1985, originally 32 bit. Uh they upped it to 64 bit in 2011. There's lots and lots of models. It's really hard to get a uh a real definition of how fast they are because there's so many different models. Uh so I I I I I don't really have numbers for you, but they're powerful. Well uh I can tell you some of the numbers of the Apple though. If you get into the Apple silicon, that's all based on ARM, the uh A19 Pro, like in my phone, 19 billion transistors, 4.26 gigahertz clock, um uh even the efficiency core uh runs at a at a modest little 2.6 gigahertz. That thing will do 2.5 teraflops, 2.5 trillion floating point instructions per second in my phone. Um uh Wolf, you got a an Apple MacBook that's got an M4 chip in it, right? I have 28 billion transistors.

Wolf: 1:00:38

The best MacBook Money can buy. I have the best MacBook Money can buy.

Jim: 1:00:42

Yeah, yeah, until the next one, right? Right. 28 billion transistors, 38 teraflops, uh, I'm sorry, 38 tops, those are trillions of operations per second. That's kind of like MIPS, only only better, right? 38 trillion of them, 4.26 teraflops. These things are nuts. And then we get into the wearable computers, right? You got the Apple Watch, the Garmin, the Samsung. Those things all have little CPUs in them. Uh and let's talk just a second about the hobbyist computers. Uh, you got the Beagle Board, uh the Arduino, the Raspberry Pi, uh the Ada. Yeah, let's talk about the Arduino. I know you've got some words about them.

Wolf: 1:01:27

I I I really hate it when people take something that is open source and they buy it, and then they make it be closed source. Now, of course, this the license that they started with doesn't let them uh deny people access to the existing versions that are out there. But their new license says, from now on, um, yeah, it's closed source, and also you are not allowed to reverse engineer it. And that is that's not just against the the old license, it's against the spirit of the whole effort. Um I don't know if Arduino um had to sell or if they didn't research who to whom they were selling in advance. This is just a step. I don't even want to say backwards, they fell off the backwards cliff. This is just bad.

Jim: 1:02:26

There's a term for this. I've been hearing it a lot lately. And shittification. Yep. That's what's happening. And it's it's too bad because uh they're doing it. Arduino is a pretty cool little thing. Um, I I don't have one, uh, but some of my friends do. And and I always thought it was really cool. Um, and the Raspberry Pi, I've got a bunch of those. I I've got those things out in production, uh uh handling uh printing tasks and stuff for for client computers. Uh and the Adafruit, um, that's kind of a neat little one. That's just a little hobbyist thing that you can program with Python and make it do things like turn on lights and listen to uh sensors and stuff. Kind of a neat one.

Wolf: 1:03:07

Yeah, and we even uh both of us, you and me, know uh one of the people who built a lot of the libraries you use for that. Yeah, super talented, super kind uh in fact.

Jim: 1:03:20

She gave me one. I I've got an Adafruit something or other. A really cool little thing.

Wolf: 1:03:25

Um Yeah, she's she's smart, she's done some great stuff, she's empathetic. Uh what a great person to be around. So nice she gave you stuff. Yeah.

Jim: 1:03:34

Um uh anyway, these these little things have super powerful processors on them, especially when you compare them to the you know the computers of the 50s, 40s, 50s, and 60s. So finally, finally, Wolf, we get to the key fobs. This is really what we're here to talk about, right? Uh-huh. It took a took me a long way to get here. Uh, but key fobs, uh, as I dug into it, uh, I I became more and more impressed. I at first I I wasn't so impressed. Uh, I was looking up and they were talking about uh um microcontrollers and stuff, and it didn't really sound that powerful. But it turns out um, you know, to handle the the passive keyless entry and the the uh the PKE systems, and it's got wireless and all kinds of stuff on it. There's a really common microcontroller in those things from a company called Nordic Semiconductor, the NRF 52840. This thing supports Bluetooth, NFC, Zigbee encryption. It's based on a 32-bit ARM Cortex M4, runs at 64 megahertz, like my my key for my uh Chevy Suburban. Uh I I I can't confirm for sure, but it likely has one of these chips in it. Um what does that mean? Well, compare it now to the lunar lander. What does it mean? Let's let's get to the conclusion, right? Uh the the payoff for why we're here. Uh, is is my key fob really more powerful than the Apollo Lunar Lander? Uh yeah. Yeah, by a long way. Uh I think you and I talked like several days ago, and I was not so sure. But now coming across this uh the Nordic semiconductor chip, yeah, my key fob is a lot more powerful than the lunar lander. Um I uh I can't use it to land on the moon, but it's uh uh yeah, there's a lot of power there. Uh and I've seen some key fobs that actually have a little display on it, they give you feedback. You know, you tell it to start your car, and it comes back several seconds later saying car started. Uh those things are powerful. Uh, it's pretty cool. Is my desktop more powerful than the space shuttle? Oh yeah. Yeah. My my desktop is uh uh Apple Studio with an M1. I got one of the one of the early ones. Uh that M1, I don't know how many cores it's got eight cores or something, a whole bunch of RAM. It's much more powerful than the computers that that uh the space shuttle used. I think even if you added up the computers, you know, those five IBM computers that ran the thing, and all the laptops on board that the astronauts used, I think my desktop is still more powerful than all of that. So yeah. My key fob is more powerful than the lunar lander computers, and my desktop is more powerful than the space shuttle. Um I'm what a time to be a cat. Oh man. This is this is a great time, isn't it, to play with all these all these toys and stuff. Um there's a lot of stuff I didn't cover. Uh you know, I talked fast, uh, and we've been talking for uh just over an hour. An hour? Yeah. Of course, uh before we started, I said, Well, this is gonna be a short one. And it's not. There's a lot of things I didn't cover. We didn't cover RISC versus Cisq, we didn't talk about the power PC. There's a chip that's pretty cool. Uh, we didn't really talk about GPUs like the Nvidia's and stuff, and and parallelism and uh optimization techniques like pipelining, uh prefetching, branch predictions. There's a lot of things that go into the microcode of a CPU that make it as powerful as it is. And we didn't talk about supercomputers, we could spend hours talking about all those things. We just didn't get a chance. So uh, but I think I don't know, I think we covered some pretty cool stuff. It was a lot of fun doing the research. I I dove down that rabbit hole many, many times.

Wolf: 1:07:22

Uh, and I managed to pull myself out. And frankly, um, it's the fun we have during the research that is uh one of the really important motivators for us.

Jim: 1:07:34

Oh yeah, it it's a blast. And then to get to uh give you guys the uh as Ron said, the reader's digest version uh of our research. Uh it's a lot of fun. So uh there you have it. Yes, my key fob is more powerful. I I think about it now every time I start my car.

Wolf: 1:07:52

Nice. Um I guess that brings us to the end. Um so the of course the most important thing is you, the audience, thanks so much for listening. Uh, I can't tell you how much it means to us. Um maybe you have thoughts. Please send us your feedback. You know where it goes. Feedback at runtimearguments.fm. Um that address and uh individual contact addresses are available on the website uh where you see this podcast and where you get your show notes, etc. Uh there are links in the show notes, so look there for more information about this episode and uh, of course, how to contact us. Um and uh Jim has included several links having to do directly with the content for today. I think that about closes it up. I do want to say one thing, and that is uh our next episode, we have a special treat for you. I think it's gonna be fun.

Jim: 1:08:59

You got anything to say, Jim? Uh no, just thanks everybody for listening. Uh come back in two weeks for the next episode. Uh we're gonna have a lot of fun with it, and uh uh be safe. All right, see you a little later. Bye bye.