8088 & 8086 CPUs... Why 16 bit Came Before 8 bit [Byte Size] | Nostalgia Nerd

[@Das1x]

Said eight 97 times in 3:34minutes. That's almost 0.5 eights per second and an almost flawless 30 eights per minute. Not too bad for a short talk about the 8088. But the true 8088 experts reach the 60epm with no sweat; with some even breaking the 120epm for short durations of talking time when challenged by others.

[@user-tm3fz7qx3s]

How much free time did you have to make that comment? XD

[@Flofutz]

Now we need eight more likes to be perfect

[@backwardsface3046]

never like this again

[@snooks5607]

1. click three dots menu (above subscribe button), 2. open transcript, 3. disable timestamps, 4. copy-paste into vscode, 5. ctrl+f "8" -> "1 of 89" + "eight" x3 = 92 times, by my calculations

[@ShainAndrews]

@@snooks5607 A useful comment. Seems to be rare anymore, especially when somebody shows their work.

[@LGR]

Really fascinating subject!

[@DyoKasparov]

My man! ;d

[@idhamsyakir6355]

found you here :)

[@BertGrink]

I feel as if all the retro guys are in reality one big, secret brotherhood haha

[@pferreira1983]

Oh you're just happy you don't have to cover it. Ha, ha!

[@enschedegabber87]

after almost 40 years the 8086 is still in the legacy code of the x86-64 arcitecture ;)

[@mspenrice]

Though thanks to the various protected addressing modes that a typical BIOS and OS will almost immediately kick a modern chip into (never mind the different motherboard peripheral chipsets, video card support, etc), you'll have difficulty running a lot of 8088/8086 native code on a modern PC, unless your chipset and card have good built-in emulation for their own older standards, and you find a way to force it to start execution of code stored on a floppy or other such device without first booting the main OS or a new-style UEFI BIOS (though that's really the wrong term, as it sidesteps the normal "BIOS" altogether... I'm just not sure what the true term is, unless it's _just_ "UEFI"). In fact a lot of the opcodes and maybe even microcodes can themselves be traced back even further, to the 8085 or 8080 (which is the root upon which both the x86 and Z80x architectures are built, being a significant advance over and also a significant codebase break from the original 8008). Because when it comes down to the really rudimentary things like ADD, or LDA, MOV, etc, why would you bother changing them from one generation to the next? And once you get down to the level where each of those only takes a single cycle to complete, you can't even really optimise them any further, other than maybe automatically recognising where you have a string of them in a row, particularly when each operates on narrow-bitsize data, and can parallelise them using SIMD and/or multiple cores. But you still have to load them in sequentially and output the data in a similar way unless they're all coming from AND going to adjacent memory locations. There's only so far you can pare down the fundamental, atomic operations, and there enough examples of code floating around in the wild for 20 years or more (and it's usually OS incompatibility that kills the true long-runners rather than CPU issues), so there's not really any reason to junk them or make huge changes to the way they operate and the opcodes they use.

[@gordonlawrence4749]

Some maniac got windows 3.11 running on a 2.4GHz quad core i5 a couple of years back. You would not believe how fast it was. There is a hell of a lot of bloatware in modern operating systems that slows things down. That said the spoof version of DooM that was in one version of Excell was hilarious as the end boss was Bill gates in his underpants.

[@statinskill]

Relax. S/360 24 bit instructions endured until the early 2000's. S/370 and S/390 ran these instructions ran these instructions with early /370 really only adding a TOD clock and a MMU (DAT). I think the early zSeries even still implemented 24 bit addressing. That's only 16M. About the size of a microcontroller by today's standards. There nothing wrong with maintaining compatibility. Nothing at all.

[@mspenrice]

+Gordon Lawrence ... well, you don't have to be that much of a maniac. I've seen a co-worker install it onto another's (Core 2 Duo) workstation whilst he was away on holiday for a joke before. It's a fairly forgiving OS, so long as it can read and write some kind of disk, as well as human I/O (getting the mouse to work is a bit of a game if you don't have a PS2 or serial one around, but USB keyboards work fine) and you have a compatible video driver installed (VGA does just fine), it'll fire up without any issues. As for speed, Win16 was never particularly bothered about how powerful your CPU was (and, really, most modern versions still aren't, so long as you have a decent amount of cache to speed up task-switching). It's all about the memory, to cut down on how much it has to hit the disk, and to some extent how fast your video card runs and/or how much it has in the way of acceleration features. Requirements which any modern computer will fulfill in spades, even though it challenged a lot of contemporary machines when it came out. You're right about how bloated some modern OSes have got, but Moore's Law seems to have outstripped them somewhat with Windows 7 thru 10. I've yet to use a computer running any of those which didn't feel just fine, at least so long as you had an SSD installed (it's still somewhat bound by kernel, service and program code loading times, and also swapping if you're low on RAM; hard drive throughput has massively lagged almost everything else in the computer speed sphere, and random-access times for spinning-disk drives have barely changed in thirty years), but I ran across quite a lot of dogs trying to run XP and ending up limping badly when it was contemporary. MS seem to have found a nice plateau in terms of how much chip power, memory and disk space they need in order to offer a well-oiled, fully featured modern operating system (quite simply, they've run out of things to add, I think, and are now focussing on polishing what's already there), and affordable hardware has had a good bit of time to stretch out and accelerate beyond the bare minimum needed for the salesmen in PC World to be able to keep a machine running at a good enough pace to make a sale. Ten to fifteen years ago, the cheapest machine you might be able to get brand-new would be based on a hopelessly outclassed, P4-derived Celeron, and even with 2GB onboard would choke on XP Pro; with less memory, you may as well just forget getting anything productive done if e.g. it was time for a virus scan, or Windows update detected something new to download, even though the stated minimum was 256mb (and in some cases 128). Nowadays, I have a £40, Atom-based tablet that runs Windows 10 more smoothly than that, and the lowest end processor that's worth looking at is actually somewhere in the Pentium class (Celerons still exist, but they're looking a bit like a solution in search of a problem these days because they don't offer enough of a cost saving to justify the performance hit - even so, they are themselves streets ahead of the old P4s and probably on a par with the early Cores, maybe even Core 2s), which will punt along perfectly nicely under most usage regimes. You only really need an i3 or higher if you want to do more intensive stuff, like creating your own videos, gaming etc... It's probably not even particularly valid to buy something like an i5 on the basis that it'll "last longer" in terms of being able to keep up with the loads placed on it than the lower grade chips. The CPU in the machine I'm writing this on is a second gen mobile i5, way behind the curve and not even really up to date when I bought the laptop five years ago... but the only time it ever shows any sign of stress is if I bulk compress a load of stuff with WinRAR, convert a long video, play a game that's maybe a little too new, or a browser process runs rogue, and even then the main problem is noise and heat, the system as a whole keeps running fairly smoothly, showing that most things on a modern OS need fractional percent of full performance to work A-OK, and efficient multitasking/task switching is no longer anything to worry about. Plus the rest of the time, including right now, it's just idling... At the same age, the predecessor Pentium-M (which was a pretty high performance chip in its day) was already starting to struggle a bit with some things like overly complex webpages, even under XP. By the time I replaced it at nearly 7 years old, it was borderline unusable for daily tasks. I don't see that happening with this one, even though it's got basically a direct equivalent processor, maybe even slightly lower grade (possibly a higher rated i5 or one of the lower grades of i7 would be a better comparison)... Probably not even if I finally bite the bullet and upgrade to Win10, though there's little real reason to do so.

[@mspenrice]

+statinskill ... I'm not entirely sure what you're trying to say here. Who needs to relax, and what's the relevance of the IBM mainframe architecture (something that traditionally stays in service much longer after introduction than any given desktop equivalent - in fact, it was adapting to that faster pace which gave IBM so much of a headache breaking into the market, to the point where they basically had to spin the new desktop computer division off as its own autonomous entity so as not to be tied down by their wholly irrelevant big-iron-mainframe and five-ton-tabulating machine business philosophies) to the discussion? NB, I actually ended up skimming past some pages on the System/370 line (which was only actively sold from the 70s to the 90s, though of course support lasted a LOT longer) as a sideline from browsing stuff to do with the 5150 PC's development (as initially there was even a thought that they might try to miniaturise the entire S370 architecture into a desktop box, and do the processor in-house, like with the earlier 51xx's and the SCAMP), and they introduced extended addressing fairly early on. Maybe you could configure a replacement machine to run in the same modes as an older, broken one for compatibility's sake, and wouldn't have to both updating your code which might quietly make use of the "extra" 8 bits in a 32 bit address word for other reasons, but you could get IBMs with 32 bit addressing (well, 31 - they themselves still reserved the final bit for signalling purposes, but that still added up to 4GB total if you used 16-bit words exclusively, same as 23-bit could make 16MB) from somewhere a couple years either side of 1980. I mean, what's the point of investing in a huge rack-based computer system if it's only going to be able to use as much memory as a desktop machine? They had to stay ahead of the curve, and the support for older addressing modes was likely just the same as Intel retaining the original modes in their later processors - for legacy code compatibility (and direct machine replacement ability) only, and because *someone* would yell blue murder if they didn't, despite 99% of users having no practical application for it any more.

[@strife711]

man i am trying so hard to watch this but its hard when you say eight oh eight .. when i am so used to calling them eighty eighty eight or eighty egighty six.. i know im being lame but it bugs lol

[@seanld444]

No I totally agree with you, man.

[@galier2]

Imagine for me, it always was "quatre-vingt quatre-vingt-huit" or "achtzig achtundachtzig" (yeah, sorry I'm bilingual and living across the French and German border)...

[@anidnmeno]

> "achtzig achtundachtzig" Rolls right off the tongue.

[@weldmaster80]

strife same.

[@allesklarklaus147]

galier2 More like Achttausendachtundachtzig... But nobody uses that I hope

[@annasmith9398]

If you check your history, the first two 8 bit processors were the Intel 8008 & 8080. The 8086 was the first 16 bit, while the 8088 was internally 16 bit, with an external 8 bit bus. The 8088 was essentially the 8086 on the inside. (BTW, Intel created the 4004 before all that and it was a 4 bit processor, I believe user for a calculator at the time.)

[@ArchaeanDragon]

Things are a bit more complicated. While the 8086 did indeed have a 16-bit data path, there were times where it still took two fetches to retrieve a full 16-bit word from memory, which was when the word was aligned on an odd address boundary, rather than an even one. There were also some delay artifacts related to the 8088's shorter 4-byte prefetch queue (vs the 8086's 6-byte), so "twice as fast" is largely theoretical.

[@gabrielschilive7675]

According to the manual, the exception is if it is fetching instructions. It says it should only apply to data.

[@RobertDeloyd]

I was always confused about this back in the day... Thanks for clearing this up for me 40 years later :)

[@ashleighcartwright5186]

A man after my own heart. Absolutely love your series and look forward to your videos.

[@Folopolis]

TL;DR - Intel realized after they released the 8086 that 16-bit motherboards were too expensive.

[@gorillaau]

Folopolis The 8086 was probably requested for military applications first off.

[@epiendless1128]

They were mainly trying to avoid being pushed out of the market by 16 bit CPUs from Motorola, Zilog and Nat Semi. Once they had their 16 bit CPU released, then they could relax a little and try to mop up the more cost-sensitive end of the market.

[@gwenynorisu6883]

The 8088 was essentially meant as an easy upgrade path for OEMs wanting to transition to a 16-bit architecture, but without the ability to immediately start building 16-bit motherboards with all the necessary 16-bit (or doubled-up 8-bit, as some 68000 computers went with) peripheral chips in one fell swoop. Get the customers running on 8086 code, then when the constrained performance is no longer enough but they're well invested in that codebase, shift over to the full 16-bit setup. Thing is, they they sort of outpaced themselves and ended up getting the 80186 then 80286 out the door within about 6 short years, before the early adopters of the 8088 had time to exceed its limits and start thinking about upgrading to the 8086. Weirdly, it never really seemed to be a problem for Motorola... takeup of the 68000 vs the 680x (and indeed the MOS 650x that cannibalised its sales) was a little slow, but once it reached a certain price point sales exploded, and the 68008 was never really that popular. Unlike the Intel ecosystem where the 8080 and even 8085 were a touch disappointing and low performance, the 8-bit flavours of the Motorola architecture were pretty hot and there wasn't as much drive to upgrade from them straight away, and the idea of spending on an expensive 68k that was throttled closer to 680x/650x performance, having to change your codebase over to suit it with only a fairly mild bump in speed vs the much cheaper 8-bit solutions just didn't appeal. There were a good few computers that actually used the 8086, though... they just all got kerbstomped by the cheaper and often not that much slower IBM PC, so you don't hear much about them any more. I think the main examples were all mutually competing Japanese models that only had mild success outside of the domestic market, in fact. And possibly an Amstrad or two?

[@herrfriberger5]

The reason behind the 8088 version of the 8086 had very little to do with circuit boards ("motherboard" was an uncommon term at the time). The thing was that the 8088 could easily use mature and cheap 8-bit based peripheral and support ICs. Such as those intended for the 8080 and 8085, but also from other companies. An 8-bit bus (8088) also often demanded fewer EPROM ICs and (1-bit wide) DRAM ICs in order to build a functioning computer. This could be important for "low end" embedded systems. (Personal computers was not the main focus.) The 8088 modification of 8086 also worked as a quick learning experience for Intel's then (1978) new semiconductor laboratory in Haifa, Israel.

[@VintageTechFan]

@@epiendless1128 Motorola did the same, see the 68008.

[@jecelassumpcaojr890]

Intel actually started working on a 32 bit processor first: the iAPX432. When it became obvious that it would be very late (started in 1976 it was only released in 1981) they did the 8086 as a quick and dirty project so clients could have something to use while they waited. Despite being 32 bits internally, the iAPX432 had a 16 bit external bus with multiplexed address and data which is one of the reasons it was so slow that Intel likes to pretend it never existed. The success of the 8088 with the PC (I don't remember seeing any product with it before that) inspired Motorola to do the 68008 used in the Sinclair QL machine (and nowhere else that I remember). The problem with all these narrow bus processors is that they didn't have a block memory mode like the ARM did to take advantage of the "page mode" in DRAM chips. If they had, they would not be roughly twice as slow as the versions with full buses.

[@mspenrice]

Funnily enough, that's something Atari DID take advantage of in the Falcon, along with giving the video chip 32-bit (and burst mode) access through a backend trapdoor into memory, which is why it doesn't suffer anywhere near as much as would otherwise be expected from having a 68030 stuck on a 16-bit bus.

[@bobblum5973]

I remember the iAPX432, the architecture was totally different from previous Intel CPUs, or at least enough that writing code for it was a steeper curve than going from the 8080/8085 to the 8086/8088. That's one think I liked about the Motorola 68xx series of CPUs, it was a gradual progression up the product line until you hit the 68000, which was more of a minicomputer on a chip.

[@atarixle]

Thanks! I always wanted to know what's the difference between both chips, but I was too lazy to google it. The background, and to know that the 88 came after the 86 is a bonus.

[@DANGUmustDIE]

Use start/stop bits when transmitting so much eight characters :D

[@mr.nobody6829]

Eighty eighty-six and eighty eighty-eight are already mouthful enough, Eight O eighty-six, eight O eighty-eight...I'm starting to feel dizzy now.

[@wisteela]

Excellent vid. Looking forward to more of these.

[@Yngvarfo]

One other thing. The speed advantage of addressing 16 bit words at a time with the 8086 could only be realised if the words were located on an even numbered address. Since this made no difference on the 8088, there was no incentive for programmers to stick to that. Several of the internal data structures in DOS had odd numbered sizes, or had 16 bit words at odd addresses inside them. The result was that even when 16 bit memory became more common, such as when the IBM AT came along with the 80286, a lot of software had data on odd numbered addresses which forced the processor to pass it in 8 bit chunks anyway.

[@miroslavmilan]

I read the thumbnail as: "why the BOB6 came before BOB8" and thought: Well, why wouldn't it?

[@raven4k998]

why would they go to 8 bit after 16 bit because it makes no sense to go backwards like that🤔

[@cheater00]

Love the shorter videos - nice and to the point!

[@Vargas3499]

I like how that guy in the beginning explains clock speed and front side bus width

[@anurag_verma_youtube]

I want to know what is the original source of that video, I may watch that show

[@looneyburgmusic]

As much as I love watching your videos, (really brings me back to the late 70's into the 80's, when I used a lot of the computer gear you've talked about), hearing you say "Eight-Eight-xx-xx" over and over again almost made me punch my $1100 38inch monitor. Please, going forward, don't do that again, "Eighty-Eight-xx-xx" works so much better, and doesn't make the watcher want to cut their own wrists after the first 30-seconds ;-) Cheers!

[@iamsemjaza]

In Michigan, we always pronounced it eighty-eighty-eight and eighty-eighty-six.

[@BertGrink]

As did the guy in the vintage video clip

[@gordonlawrence4749]

We always did here too in every place I worked at in the UK.

[@seanld444]

Same here.

[@userPrehistoricman]

everyone except this dude

[@laurentd9980]

Well, in France we always pronounce "four-twenty four-twenty-eight" and "four-twenty four-twenty-six" :-) (We also say "Windows four-twenty- fifteen" for Windows 95 :-o

[@Ali107]

The master of 8s, 6s and 0s!

[@kjamison5951]

I worked in a school and the PCs had 80186 CPUs. The server was a 80286. The next advanced computer was the DOS card in a Macintosh and that was a 80486. When I wrote the spec for 5 Pentiums, everyone wanted one... They also started with Motorola 68000 and worked up to 68040 before buying PowerPC’s. These days, there’s not a lot of change so much as multiplying cores and running RISC CPUs.

[@Dr_Mario2007]

8008, 8080 and 8085 were all 8-bit processors from Intel, with the 8086 being Intel's first 16-bit processor (and therefore the beginning of the popular x86 microarchitecture that we still use today), and also the 8088 is actually a 16-bit CPU, but gelded in order to make it cheaper to build the computer (16-bit circuits required extra DIP chips to form the Northbridge and Southbridge circuitry on the motherboard - long before both Northbridge and Southbridge become integrated, on a single die for each components - nowadays, the Northbridge circuitry is integrated on the CPU die or implemented as a separate Northbridge die in recent Ryzen CPUs), so the cost was also a factor for IBM as they were trying to enter the home PC business back in the day.

[@multimood]

the 8088 was used in the IBM PC and XT, the 20286 was used in the AT, I read that the 8086 didn't come out until the lower PS\2 models, these coming after the AT

[@ci3219]

Anyone else hates it when they call it "the eight oh eight six" - when those authentically from the 80's knew it as the "eighty eighty-six"?

[@TheJamieRamone]

The IBM PC may have been (arguably) amazing, the 8088 was certainly not! Specially among game programmers as it notoriously inefficient thanks to it's neutered data path, causing the instruction pipeline to be empty most of the time as the ALU processed instructions faster than they came in wasting cycles.

[@mystsnake]

This could of been summarized as the 8088 was a cost cutting version 8086. But the problem is BOTH are 16bit chips, plus the timings are a lot more complicated then merely that moves 16bits so its twice as fast as 8bit... For quite a few instructions, they were so long the extra cycles added made small difference. Plus the instruction set is obscure, a instruction is between 8-120bit in length, which is why timing is complicated, memory access aside instructions could vary between 2-180 cycles roughly for both cpus

[@dijoxx]

*could have

[@Rouxenator]

Excellent video 6 years later.

[@Spudcore]

Is that a 303 I hear accompanying the 8088? How very apt!

[@TimHoppen]

My first computer at home was an IBM PC with the Intel eighty eighty eight. We got it in the mid 90's after it was retired from use at the local university that my mom works at.

[@hashirscam]

Finally, found the answer! Thankyou!

[@freddyzdead1]

The big mistake Intel made was to keep squeezing their processors into a 40-pin package. There was actually very little speed difference between the '86 and '88 because the 8086 had to multiplex the data and address lines due to not enough pins. Motorola's 68000 was in a 64-pin package and had none of these drawbacks. For the same reason, Intel adopted the ludicrous segment/offset addressing scheme, which was cumbersome and confusing, and limited the immediately addressable space to 64 kilobytes, just like an 8-bit processor. With that sort of thinking, it's hard to see why Intel became the most powerful chipmaker on the planet.

[@bobblum5973]

The package size/pin count issue was going on even back during the 8008 days: www.intel4004.com/qa8008_8080.htm

[@galier2]

TI-99/4 was 16 bits before that (but it had only 256 bytes of 16 bit memory, all the rest was on an 8 bit bus).

[@Alianger]

So which IBM PCs used the 8086 one? Edit: Seems the most popular PC to use it was the Compaq Deskpro

[@poitsplace]

Actually if you look up the instructions per clock, the 8088 was slightly faster. The problem is that the earliest processors were extremely simple. They more or less had a list of steps to complete for each instruction but made sure each step was complete before doing anything else. The 8088 most likely took advantage of something called "pipelining". In the early forms of pipelining, fetching the next instruction and/or writing completed data were made independent. That way while the instruction was being processed internally, another part of the CPU could deal with the data read/write concurrently.

[@michalzustak8846]

... And this is how IBM delivered the original PC with a CPU slower than most real 8-bit CPUs despite being technically 16-bit. This type of technique is the plague, they claim it offers the power at a low cost, but such bottlenecked systems tend to be worse than the "real" system of a category below. The 16 Mhz 286 pure 16-bit also outpowered the 16 Mhz 386SX crippled 32-bit CPU.

[@HrHaakon]

386 had memory protection and other basic security features though.

[@mspenrice]

I'm going to need some kind of source on that "16mhz 286 is faster than a 16mhz 386SX" claim. The SX still had a more efficient core clock-for-clock even when operating on pure 16-bit code, could run 32-bit software (which, yes, meant a performance hit... _versus the 386DX_ - if the 32-bit instructions were used intelligently, in a mix with 16-bit where appropriate, it should still have meant a boost vs the 286), could use protected mode much more easily than its predecessor (meaning it was much more likely to be found _actually using that mode,_ which conferred advantages like not having to fart about calculating segment offsets for each memory address outside of the first 64k, and being able to make much more efficient and speedy use of memory over 1mb, particularly when it was connected directly to the motherboard rather than sitting on an ISA card), was inherently better at making use of external caches despite not having an internal one (at least, not until the laptop-focussed SLC came along), and even turned up in systems with Local Bus and/or EISA expansion slots meaning it could communicate with those peripherals a lot faster than over regular ISA (which was stuck at 7.16mhz) despite still only having a 16-bit data bus. Possibly there are well made, and moreover very well-tuned 286 systems (especially those which have an x87 coprocessor, or maybe actually use an NEC V40 chip...) that are able to outperform generic, poorly set up, default-setting-using 386SXs on a clock for clock basis; but on a level playing field, the SX should be able to outrun the 286 _even where it's been dropped in to the exact same motherboard as a direct replacement_ (which, after all, was sort of the point of making a 16-bit bus variant in the first place, the same as making both the 8088 - which could use 8-bit boards originally intended for the 8085 or 8080, and the Pentium Overdrives - which adapted the 64-bit Pentium to work in 32-bit 486 and even some 386DX boards). I mean, the difference might not be that much, but it'll still be measurable. And if you were to run code tuned for the 386 itself and thus then 286-incompatible (as it implemented several new instructions even if running in 16-bit mode, and featured wider main registers that could be broken up to provide twice as many 16-bit pieces of internal storage as well as several wholly additional registers vs the 286, ultimately meaning more work could be done wholly internal to the CPU without having to endure the performance penalty of hitting RAM for more data) the improvement should be even more noticeable. And that's not even considering the usability benefits produced by Virtual 8086 mode and all that... or that it had its own 16-bit version of the 80387 (the 387SX) which was faster than an equivalent 287... or that the 16mhz parts were only ever meant as the cheapest possible point of entry to the 32-bit code environment (with the SX itself being capable of higher clocks than any 286, and the DX being faster still) so it didn't even matter if they really were slightly slower than the older chip on the exact same code.

[@RonJohn63]

Very true. Text applications like Word Star were much faster on my 4MHz CP/M KayPro machine than on a 4.77MHz IBM PC.

[@jlinkels]

That was very common in those days. First develop some device with full power, then derive a less powerful crippled version. I assume it was commercially attractive do to so as something could be advertised as further development of the real version and then sold as the crippled version. Wasn't there the 386SX where they did the same trick, 32-bit internal and 16 bit external. And the 486SX which had an integrated numeric copro -- oh no, wait that was the 486DX and they disabled it on the 486SX.

[@swinde]

The "SX" suffix was born when a large number of processors coming off the line did not meet factory specifications. These were offered with the suffix. Later they just disabled to numeric processor to maintain the SX market, which seems kind of a way of cheating.

[@justsomeperson5110]

Oh ye gods! The success of the 8088 for Intel was inversely proportional to the experience of the consumers. This step backward for pricing purposes was such a painful performance loss. My first PC, a Tandy no less, with its amazing Tandy color and sound, should have been stunning for the time ... if not for this noticeable performance loss from the 8088. It was not a smart pairing. So most of my gaming continued to be on the C=64. LOL

[@yrgdrasil6403]

0:20 where can i get original video?

[@laurentd9980]

Outstanding video as usual: clear, concise, and accurate information, delivered through an excellent commentary, and perfectly illustrated and edited. Thanks for all these professional-quality videos on the history of computing, they are real gems to me!!

[@mspenrice]

I'm going to take a stab (before the video kicks off) that it's because 8-bit and indeed 4-bit (and lower) processors already existed before both of those chips; the 8086 was a 16 bit (hence the "6") successor to the 8085, itself an upgrade to the 8080 (and 8008, and 4040/4004). When there was a clamour for an internally 16-bit machine code compatible with the 8086 but able to be retrofit to cheaper and/or existing 8-bit bus motherboards (as its improved efficiency over the 8085, and particularly increased address space, and how the narrower bus wouldn't be too much of a bottleneck for byte-aligned data even though 16-bit opcodes would have to be cut up, meant people weren't happy just sticking with that), they cut the external bus down to 8 pins and took the simple route of designating it 8088. Not sure if the 88 is still that much faster than the 85 overall, given how it's generally starved for bandwidth (the input pipeline, which is kept reasonably full on both the 85 and 86, is near-empty much of the time on the 88, and processes end up being held up by having to output results in a laborious half-speed fashion)... but, eh, it's what the people wanted, and it was at least code-compatible with the big brother chip meaning you could write programs for either and they would run on the other, providing an easy upgrade path once it became more affordable to build 16-bit PCBs, as well as a way of having stratified cost and performance (running the same bytecode) with multiple machines in a manufacturer's range, the faster ones not even having to run at a higher clock, merely have a wider bus and the '86 instead of '88.

[@PearComputingDevices]

They used this same trick with the 386sx to save money and reuse 286 motherboards. Basically put a 32-bit cpu on a 16-bit bus. OEMs loved it because they could advertise a system as being a 386, yet re-use the stock from the 286 machines and thus saving money.

[@MatthewHendrickson]

Are there any chips that still have 8088 compatibility like the x86 line has?

[@ThomasFarquhar2]

Why would you need it to though?

[@dovic2293]

what is the difference between 8088 and a 8088-2 ?

[@scottsullivan234]

Yes, same number of transistors between 8088 and 8086, but the 8088 can address only 64KB of RAM. The 8086 up to 1MB.

[@Decco6306]

can i have a link to that video that you posted at the tintro of the vid? where he talkes about the datapaths and clocks

[@Corsa15DT]

you should do benchmark on both cpus

[@trulsosmundsen515]

Really good video! But.. WAY too many 8's in the first 8 sentences... heheh :)

[@jaybrown6350]

The reason stated in the video for the 8088 is incorrect. The 8088 was a custom concession to IBM. The support chips were not ready for the 8086 at the time that IBM was designing the IBM PC so Intel modified the 8086 to be the chip vendor for the IBM PC.

[@daishi5571]

While it may take 2 operations to push/pull the data through the bus (only 16 bit information) internally it is the same speed (which vary in range of clock cycles for an operation) and due to the fact that not every command has an output (sometimes it's just a setup for the next command) the actual speed difference between the 2 processors was a lot closer than it sounds.

[@mspenrice]

Only if the code wasn't well optimised for the 8086 and happened to be accidentally well optimised for the 8088. The core design has a 4-stage pipeline (ie, 4x 16-bit word read-ahead input buffer), with the 8086 actually being well balanced in terms of memory read rate vs average execution rate (and need for writing output data) that keeps it reasonably full - and thus both the memory bus and execution engine close to full saturation, without either of them ending up idle waiting for the other. Which is basically the optimal way to set up a single tasking, single core CPU and its support circuitry, as you can't really make the system any faster by individually speeding up one side or the other; you either have to increase the clock speed (accelerating both chip and motherboard, and using faster memory to cope), or increase the efficiency of both parts (making the core do more work per-clock, and then having to lengthen the pipeline for better buffering, reducing waitstates and implementing burst mode transfers on the motherboard/memory side...). It only ends up suffering data starvation or leaving the bus idle with fairly pathological instruction mixes that have long strings of ops that each take much less or much more time to complete than maybe three or four memory cycles (one to read the instruction, one or two to read data, one to write the result), when most code would naturally end up with a mix of instructions with different cycle counts, and a dedicated programmer would cycle-count and rearrange code where the exact sequencing wasn't critical in order to maintain as optimal an average as possible. With the 8088, that job is a lot harder, because it takes twice as long to read or write anything to/from memory other than strictly byte-(or nibble, bit-)wide data, and as early-generation as the 8086 was, it just doesn't have that many commonly used instructions that take at least as long to process as 6 or 8 memory cycles (when working on 16-bit words - when working on 8-bit it's 4 at a _minimum_ - 2 for the instruction, 1 to read a data byte, 1 to write a result byte; if working entirely on registers, the instructions still take twice as long to read). So unless you have an extremely smart programmer writing your code, able to sprinkle a good number of genuinely useful long running operations inamongst all the other shorter and more common one, the core will be continually running ahead of what the data bus can provide, leaving the pipeline essentially empty because as soon as something's read into it, the data-starved core is consuming it byte by sparing byte, rather than it waiting in the buffer to be used all in one go as and when needed. So it's almost always going to be labouring under a half-speed cap when working on data-less pure opcodes (all 16-bit) or anything using 16-bit data, and still have trouble managing anything better than maybe 75-80% (3 vs 4, or 4 vs 5 memory read ops) when working on pure byte data unless the instruction mix involves a lot of mults or divs... Weirdly, about the only time an 8088 will end up matching 8086 performance is when processing code that might traditionally be thought of as rather challenging for the 16-bit chip, as the core will be tied up chewing over complex arithmetic work contained wholly within the internal registers and seemingly running very "slowly" as it's not running through very many operations per second, with the pipeline sitting full of successive instructions awaiting processing and the memory bus being largely idle (like, doing nothing at least 50% of the time). In which case, the 8088 is just as able to fill up its pipeline (or at least keep it, say, half full) as the 8086 and so can keep its ALU and other core components continually saturated with new instructions and data. And similarly, any frequently branching code that ends up regularly taking the less-likely branch (so not prefetched by the branch predictor), which is a bit of a problem for an optimally matched processor like the 8086 because you then have to flush the entire buffer and start reading the other branch from scratch and suffering a resulting slowdown, will run almost as fast as pure sequential or endless-looping code on the 8088. On the other hand, just as unintuitively, "light" code that takes hardly any time to process (and so seems "fast"), especially if working on 16-bit data, could be slowed to 50% the speed of the 8086, maybe even slightly less because of the extra delay involved in directly reading the data from memory in response to the core's requests rather than being able to take it immediately from the pipeline. So on the whole it's still somewhat slow, but just how much of a speed hit you take in exchange for a cheaper overall architecture (which gives you spare money to improve performance outside of the CPU-Memory pair, including possibly turning up their operating frequency - or, as seen in the IBM PC, implementing an entirely separate video system with its own VRAM that doesn't load the main bus unless you make changes, on a wholly separate expansion backplane which lets add-on cards talk to each other without touching the main bus, as well as various other little copros like the keyboard microcontroller that further reduce load on the CPU/bus) is dependent on what code you're running... with heavy maths and frequent unlikely branches unexpectedly showing much less slowdown than simplistic toy code that in most cases isn't really hugely dependent on raw performance (especially if it's been upgraded from BASIC to assembler/machine code, and ultimately hailed from an 8080, 8085 or Z80 based 8-bit machine...). Which is probably why the initially unlikely looking architecture of the 5150 PC ended up being so successful - the raw processing side of things was pretty marginal in terms of speeding up simpler old Z80 code, but would have done quite well on heavier number crunching, especially if paired with an 8087 for the floating point side, and all the extra little details IBM were able to build in with the extra budget headroom versus an 8086 (or 68000, 32016, TI9900 etc) based system both helped compensate for that by relieving the processor and memory of a lot of the load they were expected to carry in a lot of other contemporary systems (so they could concentrate more completely on actual data processing instead of endless admin, or redrawing the display from a shared buffer 50-60 times a second) as well as making it remarkably flexible and expandable for a huge range of different applications on top of what the base model could handle.

[@okaro6595]

Remember that you need to fetch the instructions. Many instructions were two bytes that meant 8 cycles to fetch. It does not help it it takes just 2-3 to execute. If the instructions accessed memory then competed with fetching for the bus. Also 8088 had just 4 byte queue whereas 8086 had a three word one.

[@stephenjacks8196]

Umm ... the 8088 was an an upgrade path offered to 8085 system users. The IBM PC precursor used the 8085 so the first IBM PC was an upgraded existing design. Note the PC 8-bit card slots are same as for IBM earlier 8085 offering.

[@seprishere]

Reminds me of the 68008, which the Sinclair QL used.

[@JustWasted3HoursHere]

This reminds me of the fact that the lighter was invented before the match, but only in the sense that the seemingly more advanced thing came before the less advanced thing. JW3HH

[@bobblum5973]

We take so much technology for granted, now that it's cheap to mass produce. Simple things like the ballpoint pen, the Number 2 wooden pencil with graphite lead, paper clips, wooden and paper matches (as you mentioned). And let's not forget the ubiquitous batteries in AA, AAA, C, D and 9-volt sizes, along with many more. Think of all the gadgets we have that would fail if they were no longer available.

[@pete3897]

Whaat? Seriously? (I'd be interested to read more! :)

[@JustWasted3HoursHere]

@@pete3897 If I searched long enough I could probably find it, but that was YEARS and years ago.

[@timelapsedprojects4892]

What music is playing at the end? The one with TB-303

[@nicolasjonasson4820]

Isn't the "visualisation" of 16 bits vs 8 bits missleading? Its shown as twice the size, but it is infact 256 times the size.

[@StrangeDuck]

we had an 8088xt turbo when i was a kid. Not until a few years later when i understood pc's better i found out it had a turbo card. that was pretty weird.

[@freeculture]

Unlikely. Most 8088 were "overclocked" parts, they simply worked fine with higher than standard 4.77Mhz crystals. But some software (ie. old games) didn't account for more than 4.77mhz and would become unplayable, so the pc clone makers introduced the "turbo" button, a switch to go from overclock mode (turbo) into normal mode, now you could play Digger properly again with that 12mhz 8088 cpu put to work back at 4mhz by pressing that button. TL:DR was not a "turbo card", but an overclock switch they called "turbo".

[@iblesbosuok]

I respect to NEC who developed V20, direct replacement of 8088 with quiet low working temperature.

[@okaro6595]

They also made a much faster multiplication in it. Norton sysinfo used multiplication as a benchmark so people thought the V20 was 70% faster than 8088.

[@aidanstenson7063]

where did the 86 name come from because 8086 came before x86

[@slap_my_hand]

aidan Stenson The 8086 and 8088 were the first x86 CPUs. The 80286 had the same basic architecture, but it added a protected mode and a few features that made it easier to develop software for it. The 80386 (i386) expanded the architecture to 32 bits, but it was still compatible with the 8086 and 80286.

[@aidanstenson7063]

StarTrek123456 thank you

[@gordonlawrence4749]

You missed out the 80186, which was released and used by - - - - - -nobody. It had a half arsed DMA controller (nothing like as good as an 8237) and an interrupt controller which was nothing like as good as the 8259 and not cascadeable either.

[@gwenynorisu6883]

The name came from it being the latest incarnation in the 80xx line. Intel started with the 8-bit 8008 (upgrading from the 4004 and 4040), then followed it with the 8080 and 8085. The 8086 was a source (though not entirely code) compatible 16-bit extension of those, and the 8088 its 8-bit bus counterpart. Everything after that was built directly on the 8086 as a base, hence getting a 1, 2, 3 etc shoved in between the "80" and "86". Quite why they didn't instead make them the 8186, 8286 etc, I dunno... it probably scanned better when said out loud, or there may have been other existing chips with those numbers. Or maybe by then they'd just got attached to the "80" prefix and decided to hang onto it to imply a straight progression from the 8080, even though it wasn't quite that simple... and it also risked confusion with the similarly 8080 derived Z80... The "x86" is simply from "x" commonly being used in numbered series such as those used for microchip families to indicate "any number could go here" (and, in the case of the 8086 at least, there might not even _be_ a number in that position). It's also less commonly written as IA86 (Intel Architecture-86) or 80x86, and similarly, the Motorola equivalents are sometimes described as the "680x0" series (and, for the first two incarnations, possibly even 6800x and 6801x or just 680xx, as they both had 8-bit-bus flavours available); we also have the 80x87 and 6888x coprocessors, 5600x DSPs, etc, as well as 680x and 650x 8-bit CPUs... etc etc.

[@kjamison5951]

In the same way that certain logic ICs were often referred to as 74xx series, the ‘x’ simply represents ‘any number. When you see x86, the ‘x’ is a placeholder for ‘80’ or ‘801’, ‘802’ etc, to give the 86 family of CPUs their full names. Eventually, the CPUs became known by the last three digits, 286, 386, 486 and so on. To break with this traditional naming convention, Intel used ‘Pentium’ to separate their 586 from the other manufacturers making 80586 CPU’s. At one point, there was a 686 CPU which intel named the Pentium Pro.

[@weldmaster80]

My first IBM was the 8088 XT, it even had a HUGE 20 mb hard drive, ahhhhhh good times

[@okaro6595]

I had a 30 MB ST-238R.

[@tr4pfloor]

OMG!!!! Just say eighty eight or eighty six. LOL :P Just messing with you. The good old days of 8086, 286, 386, 486(MX & MX Turbo).... Nostalgia for sure. Subbed! :D

[@chanhguo1734]

My first computer IBM PC XT. Intel 8088, DOS 1.1

[@TheMovieCreator]

Now, what if IBM had gone for the 68k for the 5150 (which they almost did)... Things would problably have been very different today to say the least.

[@robertkopp873]

The Pilose One Likewise, what if IBM had selected CP/M as the operating system, instead of MS/DOS from Micro$oft? Much, much different...

[@gordonlawrence4749]

The 68 series unfortunately memory maps everything, whereas the 80 series from pretty early on has an I/O map as well as memory map. It's still not Harvard architecture but that would have required another bus which is too expensive for a lot of applications. Even low end DSP's are now going single bus to cut cost.

[@gwenynorisu6883]

Better sacrificing a relatively narrow range of memory addresses for IO than putting up with that crawling horror of a segmented memory architecture in the x86s, though, as well as not being able to simply treat an IO target as a memory address and throw a value at it and let the support hardware deal with the rest (vs having all the IN and OUT, NEAR and FAR statements etc), with a nice simple one-bus design. Even in something relatively convoluted and not particularly forward-thinking like the Atari ST architecture, there's less than a single meg of address space given over to all the system devices _and_ the multiple 64k ROM areas (the latter being something that the IBM PC loses more than a third of the CPU's total address space to, and ends up having to address memory upgrade cards _as peripheral devices_ - yuck! - and swap blocks in and out of a small part of main memory in order to work around), leaving more than 15mb usable. In a mid 1980s system design, which officially never had more than 4mb RAM installed. Safe to say it wasn't really that much of a problem. Even in the Falcon, which only fit a relatively small number of main RAM blocks in-between the mapped ROMs and IO (in a similar pattern to various ST max-memory cards, presumably for compatibility with software written for them), provided up to 14mb of working memory, which at the time of its launch still seemed quite generous, and it was only limited to that because the 030 was bodged into what was essentially an old 68000 based motherboard design; if they'd managed to make a proper 32-bit one, then the reserved areas would have disappeared into the wider 4Gb address space and could happily have been masked off as a single multi-megabyte block without there being any trouble at all. And, well, don't most Intel-architecture based 32-bit PCs of the modern age end up not actually exposing their full 4Gb range, and being stuck with somewhere between 3.0 and 3.5gb only, because of various reserved address ranges clashing with the RAM anyway? I think that's probably at least as much of an issue if not far worse... losing 512 to 1024 _megabytes,_ a good 1/8th to 1/4 of your entire memory space (or 3/8ths, 384kb in the 5150...), vs less than 1/16th (and less than 1mb).

[@xnetpc]

It seems like they could have made a single CPU, that depending on the availability of data paths, could process either 8 bits or 16 bits data.

[@rooneye]

The video at the beginning, what is that from? Is there a longer version of it?

[@herauthon]

+rooneye - isn't that PACMAN - the game?

[@TheOriginalSycHolic]

8088 is made from the 8086 its the first example of selling a cheaper chip with a few less features... ya should talk about the 80186 and its insane short lifespan (most people never even heard of it)

[@mattl_]

Not to forget the 4004 and 8008, and the 8080., but ultimately the MOS 6502 won out for many due to the price. I believe the 8088 was a reaction to the price of the MOS 6502?

[@GeoNeilUK]

+Matt Lee The 6502, which became a Commodore part through Jack Tramiel buying up MOS after bankrupting them through not paying Commodore's bills (Business is War... and that would make a good title for a biopic or mini-series about him)

[@mattl_]

+GeoNeilUK Sorry, I think I'm missing your point?

[@GeoNeilUK]

Matt Lee Just sharing a bt of history. The 6502 was meant to be a cheap clone of the 6800 and was desgined by people who had left Motorola (the makers of the 6800) It was actually lawyers that made the 6502 what is was! What's interesting is Atari and Acorn were using this Commodore part in their computers which were directly competing which the VIC-20 and then C64 (which is probably one of the reasons why Acorn just said screw it and designed the ARM CPU for the Archimedes, they didn't want Tramiel doing to them that TI did to Commodore's calculator business)

[@Nostalgianerd]

+Matt Lee It was certainly a reaction to the industry not wanting to expand their boards to the 16 bit architecture, so any competition which circumvented were probably winners, until IBM came along and snapped it up.

[@3gdosrsfs]

+Matt Lee the 6502 was a superior processor in many respects even if it was not a 16 bit processor. The instruction set was extremely simple. Most instructions were 1 or 2 cycles max as where the 80XX were horribly inefficient in cycle usage and even though the 80xx had more registers the instruction set was rather limited. Any subsequent 80x86 processor following was even worse as it had to maintain compatibility with it's 16 bit counterpart and the only way it could ever use more memory is through memory segmentation. The truth of the matter is that the 6502 in the end did not win out as you'd probably be using a 64 bit version of an Atari or Commodore computer today. IBM's name was so huge that they could have used dog doo for a processor and it would have won the market share. Bill gates undermined IBM who has long ago left the PC based market they created. Let's put it this way. An Atari 8 bit or a Commodore 64 was much more feature rich and cost WAY less. It was a few years before you could even get any reasonably priced color cards. Even the expensive ones were so slow that the other 8 biters at the time were far more capable. IBM originally toyed with using the 68000 as it's processor, but unfortunately for us, Bill Gates essentially stole CP/M and Unix, put them together and created MS-DOS. Computers like the Amiga or Atari ST were far more capable and less expensive and it took DOS based machines a bunch of years before they could compete technically. In fact the Intel line stepped in shit when IBM went with them. That IBM name held such high esteem that technology had nothing to do with why we use DOS/Windows based machines to this day. These OS's are garbage and continue to be garbage. They are inferior in every way as are the processors. It was not until the Pentium lines came out that Intel started to show some real technical prowess.The only real competition to an Intel processor today is the Arm processor. If this were not true, Apple would still be using the PPC, but even they saw the writing on the wall. Remember...the 8088 was designed to do motor control processing, not home computing. Bill Gates was not a marketing genius. He simply went with a name no one else could compete with. Once he had a stronghold, his marketing was just a pile of BS and most people bought into it. So in that respect, if BS'ing your potential market is genius, then yes Bill Gates was the master. Technologically, he has us all by the short and curlies. I use my PC because it is the mostly written software platform. I'd prefer something much more than what the DOS/Windows based PC have to offer but there is not much out there. Android or iOS as well as OS.x are the only competition and they are all front ends to a Linux Kernel anyway. Linux should be the Os of choice but too many people are afraid to switch because the support is just not there. One day this may change and I always though it would have by now, but you can't walk into a store and buy a fully loaded Linux system, at least no where near as easily and as cheaply as it is to by a Win based machine. I have a Mac, my PC dual boots to win and linux....I always use the win based Os...why? because unfortunately, that is where all the good software is.

[@anchietacruz]

Please do one about de Motorola 68000 and its variants

[@televiciousgoober]

Never knew the 8086 got higher than 4.77mhz, interesting. we had the 8088 at 8mhz because it was cheaper than getting a 4.77 mhz 8086 and said to be about the same speed. It's like the original celeron.

[@thetaleteller4692]

Actually 16-Bit did not come before 8-Bit. Intels first integrated CPU was the 4004, a 4-Bit CPU. It evolved into the 8008, Intels first 8-Bit. The 4004 got a later improved brother the 4040 as the 8008 got the 8080. The latter one released in 1974 got a mostly unkown successor the 8085 in the year 1976. The 8086 followed two years later. The 8088 was mostly a discount offer as the 8-Bit era as we know was starting to take off during these years. It also got a successor in 1982 called the 80188 as cheap brother of the 80186. Both getting not much of attention since the 80286 was released in 1982 as well.

[@PauloConstantino167]

Exactly. This guy is just an idiot

[@srlagarto1]

He never claimed it did. If you watch the video and read the title, he just said that the 16 bit 8086 came before the 8 bit 8088.

[@another3997]

Paulo Constantino I think you were looking in the mirror when you said that. Try watching the video properly next time you muppet.

[@gordonlawrence4749]

@TaleTeller and @Engineer of Wonders. Try actually listening and comprehending, muppets. He said the 8086 was produced before the 8088 only and then said the 8086 was not used because existing motherboard designs were 8 bit. Ever thought about actually learning English Comprehension? Nope didn't think so.

[@HrHaakon]

Well, I mean, yeah, but you had 36bit-word computers from IBM before then. CPU history is fascinating. :)

[@JimLeonard]

I'd love to watch that entire Mark Minasi video; what is it called?

[@Nostalgianerd]

+Jim Leonard It's just a short clip, about a minute long in total. It's from a PC upgrade book with CD-ROM I have

[@Nostalgianerd]

+Jim Leonard I'll put them up on my Extra channel if I get a chance.

[@JimLeonard]

+Nostalgia Nerd No worries; I've tracked down a copy. Thanks for the info.

[@milou-psychicmedium]

How about Motorola 8008 the one bit processor 😂😂, I had to wire-wrap ten of those CPU cards to be used as controllers for a wafer processing machine, it was fun but I didn't enjoy wire-wrapping. We have it easy today.

[@bobblum5973]

Wait, the 8008 was still an 8-bit cpu, the 4004 was 4-bit. The MC14500B Industrial Control Unit (ICU) was a CMOS one-bit microprocessor designed by Motorola in 1977. Was that manual hand wire wrapping you did, or with a powered wire wrap gun? I came across my hand wire wrap tool the other day, had to explain to people what it was and how it worked. 😉

[@wind5250]

Daaaaaamn i havn't seen an isa port in so many years .

[@RockyBergen]

Very well explained!

[@bobcobb2429]

0:22 Nostalgia Nerd's dad!

[@PUBGDRGN]

Intel released i7-8086k for Commemorate 40thof 8086

[@johneygd]

Another interesting subject will be, what's better, a 1mhz 16 bit cpu or a 8mhz 8bit cpu, would be really interesting.

[@38911bytefree]

Hard to tell since ALU and register width on 8086 and 8088 (as I understood) are 16bit. I dont think there is an 8 factor. Speed on the internal registers is tied to Mhz and width. Aritmethics, integer, wont penalize by an 8 factor. None of them can do 32 bits resutls or math or floting point. Both emulate them. It could be noticeable on memory access for program and data, but is factor could be not bigger than 3 .... worst case.

[@youreperfectstudio7931]

Closed Captions said you are looking at Bear bust sizes haha

[@usquanigo]

Good lord. It is exhausting listening to you name the chips. They are the eighty eighty six, and the eighty eighty eight.

[@mspenrice]

As well as the eighty eighty five, eighty eighty seven, eighty oh eight, eighty eighty, and the eighty eighty eighty oh eight nine eight sixty nine sixty nine sixty nine nine nine oh one one three two... four.

[@mspenrice]

Fact: we didn't need to know that.

[@thelazywanderer_jt]

I like how you pronounce "Data Path". It's orgasmic to me for some reason!

[@AiOinc1]

I want to mention that you can simply truncate the physical data path and use the lower half of the bus. Why wouldn't they have done this?

[@epiendless1128]

That only works is the CPU supports dynamic bus sizing. I don't know is the 8086 supports that, but I'm guessing 'no'.

[@JJ-Streaming]

Shame the 8088 didn’t get released with 8 KHz, 8 bit bus, 8 bytes ram with 8 byte flash with 8 byte eeprom with 8 instructions in 1988

[@briantw]

You don't absolutely have to have a 16-bit bus for a 16-bit CPU. This was demonstrated on the TI-99/4 as far back as 1979, which has a 16-bit data path on the CPU, but the rest of the motherboard apart from ROM and RAM was 8-bit. No biggie and easy to do.

[@deannascott3475]

The 8086 was not introduced in 1977, it came out on June 8, 1978...

[@Calm_Energy]

Maybe that's the year it was announced? I mean I don't think chip announcements were as big back then as they are now, for example lots of people looking forward to AMD's 50th anniversary announcement. But I noticed these differences in dates in the Wikipedia pages of 8086 and 8088 and others.

Hey cool. Your pre-logo-video thing has pacman game. And like 10 minutes ago, I was scratching a PAC-man lottery scratch ticket. I only won $5, same cost as the ticket. But it is a fun scratch ticket.

[@lucysluckyday]

The 8088 seems similar to the 6809, though it was the 6502 that diverted me from the 4004 and 8008, although I couldn't avoid the 68000.

[@toby9999]

The 6502 and 68000 were the one's I favoured due to the C64 and Amiga computers. The 68000 instruction set was especially nice in my opinion.

[@citizensteve6713]

bendix k5 ?

[@iggytse]

That explains the difference between an IBM compatible XT and a 286 AT.

[@lawrencedoliveiro9104]

Did “8-bit” versus “16-bit” versus “32-bit” really mean anything? Most of the “8-bit” micros had full 16-bit addressing. The Motorola 680x0 family had 32-bit addressing and full 32-bit registers. But the early members of that family were considered “16-bit”, not “32-bit” machines.

[@whomigazone]

Or even 24 bit machines, on some revisions

[@Carewolf]

The 8086 might have come before the 8088, but the 8008 and 4004 both preceeded 8086 and is it direct 8bit and 4bit ancestors.

[@teravolt1195]

@thumbnail "why the BOBb came before the BOBB"??

[@jmalmsten]

Oh one-one-eight nine-nine-nine eight eight one-nine-nine nine-one-one-nine seven-two-five... ... ... three.

[@confusedwolf7157]

yup that one

[@itchyisvegeta]

"The world will never need more than 640K of memory"

[@bobblum5973]

I still have an OS/2 book with a forward written by Bill Gates, saying how that OS was the future... Yeah, right.

[@ShainAndrews]

Well Chrome didn't exist back in the olden days. Once you explain to them what a browser was... well after you explain the internet, then the browser... drop the bomb that Chrome will gladly tie up more than 4G just watching that first porno.

[@ShamrockParticle]

@@vlc-cosplayer apart from speed, lol

[@ShamrockParticle]

@@bobblum5973 that was before the Microsoft/IBM divorce

[@MAGAMAN]

Metronomes are not just for beginning musicians.

[@dejanpopovic7903]

So what they are saying is that the 8086 was way ahead of its time, by two years infact.

[@AmaroqStarwind]

Why is your playlist arranged backwards? The older episodes should play first, not last. More _directly_ related to the video, when did CPUs come around that would operate in a lower mode for less-capable motherboards?

[@danthompsett2894]

how come they never made an anniversary 8088 800Mhz 8bit like the i7 8086K cpu lol

[@confusedwolf7157]

ok... now do the 80188 vs 80186 video (go for max 8ishness)

[@leberkassemmel]

Let's say, it really took off with the 80386.

[@ianedmonds9191]

303 makes this video an instant win. :-) Luv and Peace

[@JouvaMoufette]

Take a drink every time the word "eight" is said

[@dfbess]

My first time on a pc.. was an 8088