Are you sure you mean 8086, or 8088?

There are multiple projects:
https://www.malinov.com/sergeys-projects/
https://www.homebrew8088.com/
https://monotech.fwscart.com/details/p6083514_19777986.aspx
https://arstechnica.com/gadgets/2023/07/going … -like-its-1981/

Just to name a few

I mean 8086 with 16 bit memory bus.
8088 is just a tad bit too slow for my personal taste.

If it comes to simplicity of the pcb design and the still available parts, it's much easier to make new 8088 systems. 8086 are rarer and more complex. And the actual real world difference between a 8088 and 8086 is quite small, so it's not really a good pick if I had to design and sell a retro pc.

Mungo wrote on 2023-11-17, 21:39:

I mean 8086 with 16 bit memory bus.
8088 is just a tad bit too slow for my personal taste.

I wondered the same for a long time.
A retro system with normal ISA bus would be so nice to get VGA cards, network cards, IDE controllers and 90s era soundcards going without sacrifice.

From what I've found, this one is the closest thing when it comes to 16-Bit retro development :
https://alexandrugroza.ro/microelectronics/sy … lane/index.html

If only the 8086-80486 processor is important, then CPU cards for S-100 bus are an alternative.
It's the bus used in the 1970s, by CP/M machines.

http://www.s100computers.com/My%20System%20Pa … CPU%20Board.htm

Edit: Another idea would be to design a "parasite" card that attachs to an existing XT motherboard (from back+then or retro development).

Such a card might be based on existing XT era CPU accelerator cards and their software/drivers (say, Orchid Tiny Turbo).
They'd contain an 80186 to 80386 CPU and some glue logic.

While that wouldn't solve the bottleneck problem of the original PC bus,
it would at least make Windows 3.1 and 32-Bit DOS extender games available to the retro scene.

Edit: And there's also the possibility of using a clock-doubler board that runs a high-speed model of an 8088/NEC V20
asynchronous at its own, rated speed or simply at twice the speed of the motherboard.

Such a daughter card could also contain the PC BIOS, running at higher performance (if its EPROM's timings allow for it).

The RAM would still be on the motherboard amd still be slow, though.
Without a small cache/buffer, the CPU would still have to wait for the RAM..

That's what separates it from clock-doubled 486DX2 processors.

That being said, it's also possible to operate CPUs without RAM - by using the processor registers directly to store information.
But that's a rare craftsmanship these days.

Diagnostic EPROMs for the IBM PC did use this approach, I believe.
It made sure that the diagnostic program could run even if the DRAM was faulty.

There's plenty of software made especially for 8088 @ 4.77 MHz, and probably not a single program for 8086.

My dream PC/XT remake would include:
- 8088 @ 4.77 MHz, with all the surroundings making it cycle-exact with 5150/5155/5160
- 10 MHz Turbo mode
- graphics adapter compatible with modern monitors (VGA or HDMI), at the software side perfectly compatible with CGA and MDA (and maybe even Hercules?), including CGA Composite artifacts
- certified for compatiblity with 8088MPH and Area5150 🤣

Grzyb wrote on 2023-11-18, 01:03:

There's plenty of software made especially for 8088 @ 4.77 MHz, and probably not a single program for 8086.

I think that one 'problem' is the mindset/opinion here.
The western PC community acts as if the IBM PC was the center of everything, while in reality, "MS-DOS Compatibles" were around early on.

Edit: To be fair, the C64 community is no different here.
The successor, the C128 was almost completely being ignored.
Despite the fact that separate, optimized C128 editions of existing C64 programs could have been shipped on same floppy disks no problem.

I mean, the original idea of DOS and BIOS was exactly "not to" tune programs for IBM PC 5150/60.

Before ~1985 or so (before it got a cult following), the PC was just "another" PC running DOS and DOS programs were CP/M style, without hardware dependency.
Edit: What made it very popular/sought after was IBM's build quality, though.
IBM had an excellent reputation for producing long-lasting office equipment.

Microsoft itself used the Tandy 2000, an 80186 PC, to develop MS Windows - rather than an IBM machine with IBM EGA™.

Anyway, I'm not saying that the IBM PC was irrelevant, but with the introduction IBM XT 286 and IBM AT, programs had to be flexible.

Anything released past 1984 had to be expected to also run on non-IBM PC timings. 🙂

Edit: What I mean to say is, that there used to be notable non-IBM PCs, as well.
Some who were used in practice, in larger quantities.

The US-American Tandy 1000 series was so popular, that many European PC emulators supported its super CGA graphics modes.
Edit: Those for Atari ST and Amiga, I mean.

Then there were European PCs like the Amstrad/Schneider PC1512 and 1640,
which had similar importance over here.
They were the closest we had to DOS compatible family PCs or household PCs.
Edit: In relation to popular home computers.

Then there were affordable (more or less) Eastern PCs like the Poisk, which could be assembled like a kit.
Edit: In relation to popular ZX Spectrum compatible systems.

Last, but not least, there was the Olivetti M24/AT&T 6300 that was popular enough in both Europe and US that other PCs and PC emulators had adopted its proprietary 640x400 CGA mode. 🙂

Edit: If I remember correctly, PC compatibility used to be provided if the MS Flight Simulator can be run properly.
That's what PC magazines back then had used to ultimately determine PC compatibility, if I'm not mistaken.
So if vanilla version of MS Flight Sim does run without glitches, the PC running it can be certified as compatible. 😉

Jo22 wrote on 2023-11-18, 01:39:

Before ~1985 or so (before it got a cult following), the PC was just "another" PC running DOS and DOS programs were CP/M style, without hardware dependency.

You're probably right about applications.
In the first half of the 80s, there was still plenty of 8-bit computers with CP/M being used as business machines - see eg. Kaypro.
So yes, developers aimed at hardware independence, to sell their products to users of various 8-bit hardware, IBM PC, and MS-DOS compatibles.

But there were also games...
See eg. Digger (1983) - it's hard to play on anything faster than 8088 @ 4.77 MHz, and requires register-level CGA compatiblity.
With games, hardware independence was pretty much impossible!

I guess there were also games targetted at certain "MS-DOS compatibles", but not nearly as many as for the IBM PC.

Overall, the only x86 machine from that era that's worth of faithful remake - even to the level of cycle accuracy! - is the original IBM PC series: 5150, 5155, 5160.

Grzyb wrote on 2023-11-18, 02:29:

Overall, the only x86 machine from that era that's worth of faithful remake - even to the level of cycle accuracy! - is the original IBM PC series: 5150, 5155, 5160.

Come to think of it, maybe not...
There was also the IBM PC Junior, together with Tandy 1000 - plenty of good toys for that platform!

And adding "PCjr/Tandy mode" to the remake of 5150/5155/5160 probably wouldn't be very difficult...

There have been some attempts at something faster than 8088 using the Vortex chips.

Really though, XT-class machines being made new is something we should be grateful for. You can do a lot with a 10 MHz 8088 system.

What we really lack, however, is modern day displays that handle ancient signals (e.g. 400 line VGA, 320x200 CGA/EGA TTL) *very well*. It absolutely can be done - projects such as the RGBtoHDMI prove this handily - but nobody's put it into a nice display yet.

The ideal would be a very high resolution LCD (Retina level pixel density) in a 4:3 form factor that scales everything nearest neighbor, or can apply CRT shaders for those who prefer them - and has sub-second mode switching.

Grzyb wrote on 2023-11-18, 02:57:

Come to think of it, maybe not... There was also the IBM PC Junior, together with Tandy 1000 - plenty of good toys for that plat […]
Show full quote

Grzyb wrote on 2023-11-18, 02:29:

Overall, the only x86 machine from that era that's worth of faithful remake - even to the level of cycle accuracy! - is the original IBM PC series: 5150, 5155, 5160.

Come to think of it, maybe not...
There was also the IBM PC Junior, together with Tandy 1000 - plenty of good toys for that platform!

And adding "PCjr/Tandy mode" to the remake of 5150/5155/5160 probably wouldn't be very difficult...

In my opinion, the most important of them all was the IBM PC/AT 5170.

If we're looking for software requirements of soft- and hardware of the late 89s/early 90s, chances are good the AT is required.

And even if minimum specs were mentioning an XT, an AT usually was "highly recommended".

This goes so far that certain electronics books bundled with an DIY PCB for an 8-Bit (!) PC card interface card
had been targeted to run on an AT class system rather than a native PC bus system (XT class PC).

I've also seen monochrome clip arts of that era which resemble the 5170, rather than the IBM PC 5150/5160.
The chassis/front bezel is the wider AT version.

Unfortunately, there was little initiative being shown by the retro community to either create a replica of the 5170 motherboard or make cycle-exact emulators.

- Again, the Commodore community has been similar here.
The Amiga A500 board design had been cloned a couple of times, but not the more fascinating A1000 or A2000 models.

The closest I could find in terms of AT emulation, back in the 2000s,
was a Russian emulator called "Script PC".

Now there are PCem/86Box, of course, but no specialized AT emulator (not that I know of). 😔

There is no need for a cycle-exact emulator of the 5170 because there is no (or very little? Don't know of any) software that relies on its exact speed.

The 5170 came in two speeds anyway, and by that time software developers were either wise to the speed control problem or were writing software that would run at an acceptable speed on a range of systems, even if it ran faster.

keenmaster486 wrote on 2023-11-18, 03:40:

Really though, XT-class machines being made new is something we should be grateful for. You can do a lot with a 10 MHz 8088 system.

I don't mean to be ungrateful, but "back in the day" being forced to use an XT class PC was nothing to be grateful for.

It rather was synonymous to suffering.

Of all platforms, users who had to operate such a PC had been felt sorry for by users of other platforms.

I don't thing I'm overacting here, my country's PC magazines of the late 80s (ca. 1987 omwards) were all agreeing here.

I remember reading a few times that video game magazines talked about better playing this and this game on an AT.
So there must be some truth within, at least. 🤷‍♂️

Edit: Please don't get me wrong, I'm not meaning to devalue the Turbo XTs in any way, whatsoever,
but even back in the late 80s there certainly was some kind of demand or "interest" in ATs and clones.

That's why I think that the Tiny Turbo 286 and other accelerators were sold to XT owners with a small budget in first place.

The same could be offered to the XT system users of today, still.
Modern FPGAs could be used to replace the custom ASICs used on these accelerators.
The old 8088 could even be moved to the accelerator card and still be used, if needed.

Those modern XT repros with a passive backplane can use a PC bus CPU card based around an 8086 or 80286/80386.
That way, the rest of the modern XT ecosystem does not need to change.

The basis for this can be found on the S100 sites.
They have 80386 CPU cards made, with a monitor/debug program that can boot MS-DOS.

Edit:

keenmaster486 wrote on 2023-11-18, 04:08:

There is no need for a cycle-exact emulator of the 5170 because there is no (or very little? Don't know of any) software that relies on its exact speed.

The 5170 came in two speeds anyway, and by that time software developers were either wise to the speed control problem or were writing software that would run at an acceptable speed on a range of systems, even if it ran faster.

But something similar could be said about other platforms, also.

- Why writing an MSX1 emulator, if MSX2 is backwards compatible?
- Or why emulating a specific ZX Spectrum model, if the latest one would do as well?

What I think is important that the AT platform has its roots in the Model 5170.
It has the same value of being preserved or recreated like the other systems.

An 5170 with EGA was a baseline system that most EGA games ran correctly on..

The animation speed (Sierra SCI0 games), the added DMA/IRQ channels needed to operate later expansion cards in 16-Bit mode etc.

Edit: I'm not saying that the performance was optimal here.

But it was a configuration that the original game/application writers had been familiar
with.
So this alone is worth to be able to re-experince.

Because, it may help to understand us why a game/application was coded in a specific way.

The limitations of the baseline setup are good to experience thus.

In the year ~2050, when no originals are more around, this might be important.

Grzyb wrote on 2023-11-18, 02:29:

ducts to users of various 8-bit hardware, IBM PC, and MS-DOS compatibles.
With games, hardware independence was pretty much impossible!

I guess there were also games targetted at certain "MS-DOS compatibles", but not nearly as many as for the IBM PC.

Not really my experience. I had Amstrad PC1640 (8 MHz PC) back in the day. I got all sorts of games which were running well, if they were not explicitly targeted for 286 or better computers. All the Sierra Online games (Larry 1-3, Space Quest 1-3, etc.) worked fine. As did Ivan Ironman's Super Offroad Racing, Xenon 2, Tetris, Maniac Mansion, AD&D Gold Box series of RPGs, Commander Keen, Elite, LHX Attack Chopper, Indianapolis 500, F-19 Stealth Fighter, Arkanoid, Populous, etc.... The list is really really long. Trying to play those games on a 4.77 MHz machine would have been futile. 8 MHz was enough to run them at playable speeds.

4.77 MHz was considered ancient and really slow already at the late 80s.

Sure, 8086 @ 8 MHz is much faster compared to 8088 @ 4.77 MHz.
But it's hardly faster compared to 8088 @ 10 MHz.

And 8088 @ 10 MHz can be switched to 4.77 MHz, making it fully compatible with the earliest PC software.

8088 is therefore more universal than 8086 - and this is my answer to the question why there's no 8086 PC remakes.

Mungo wrote on 2023-11-17, 21:32:

There are all kinds of authentic remakes of retro computers, like C64, ZX Spectrum, MSX, Amiga. Some of them are full remakes, some less so. This made me think: why hasn't anyone made a retro PC hardware remake? It would be really nice to have a new 8086, 8 MHz PC with 640 KB of RAM and CGA or EGA graphics output. Floppies/HDD could be emulated with SD card or such and mouse/keyboard could be handled with USB. Adlib or Sound Blaster emulation would be available on the motherboard.

Has anyone even considered such a project?
I'm pretty sure retro computing scene would love a product like that.

well, there have been a few small projects but think about the PC compared to the 'home computer' c64, msx, amiga and so on. while those computers had versions they generally didnt have as much variation as a PC in terms of components

also early PCs were not gaming machines, not likely to be filled with nostalgia for childhood and so on, maybe in the mid 90's on when games on the PC were more widespread but those games can be played on dosbox or bought from gog

in short there isn't a big market for it like c64 and so forth, so what happens is a few small projects here and there

An 8086 Remake would be an 8mhz IBM PS/2 or executive remake with MCGA graphics.

Nothing really targeted the Original model 25/30 so I don’t see it as a large audience

Mungo wrote on 2023-11-18, 10:45:

Grzyb wrote on 2023-11-18, 02:29:

ducts to users of various 8-bit hardware, IBM PC, and MS-DOS compatibles.
With games, hardware independence was pretty much impossible!

I guess there were also games targetted at certain "MS-DOS compatibles", but not nearly as many as for the IBM PC.

Not really my experience. I had Amstrad PC1640 (8 MHz PC) back in the day. I got all sorts of games which were running well, if they were not explicitly targeted for 286 or better computers. All the Sierra Online games (Larry 1-3, Space Quest 1-3, etc.) worked fine. As did Ivan Ironman's Super Offroad Racing, Xenon 2, Tetris, Maniac Mansion, AD&D Gold Box series of RPGs, Commander Keen, Elite, LHX Attack Chopper, Indianapolis 500, F-19 Stealth Fighter, Arkanoid, Populous, etc.... The list is really really long. Trying to play those games on a 4.77 MHz machine would have been futile. 8 MHz was enough to run them at playable speeds.

4.77 MHz was considered ancient and really slow already at the late 80s.

Yah I had one before I got super cheap deals on a 5160 and then later a 5170 6mhz, and The 5160 XT felt about half the speed, and the 5170 which I thought would be an improvement, wasn't much.... there were one or two things that were integer math or FP on integer intensive where it beat the 1640, but for general i/o and screen speed it felt like a slug, noticably quicker than the 5160, but felt slower than the 1640, by about a quarter, so it kinda felt halfway between. I still kinda harbor a dislike for the slower 286es even now they feel kind of pointless next to Turbo XTs

A "remake" and/or of an Amstrad PC1640 would be nice, too, since it was a common system of the day.
The EGA chipset in the recreation can be more generic, maybe.

Thing is, torwards end of the 80s, the clone market in general already was leading.
Developers oriented themselves on what the users had used, rather what was being blessed by IBM.

VGA as a pseudo-standard was the last big success under IBM's leadership.
And to a stretch the 8514/A, maybe, since many companies silently made their graphics cards compatible to it.
What really mattered, though were the VGA cards by third-parties.
My books say the were available since February 1988, at least.

That being said, the 8086 PS/2s were available to game/application developers,
just like the IBM AT Model 5170.
With the difference being that the AT could do EGA since 1984, the graphics standard most games had used.

The AT also supported 80186+ instructions which were being supported in popular compilers via optional switches since 1986 or so (Mix Power C comes to mind).

Of course, the NEC V20/V30 were similar capable here. That's why they were so popular.
The NECs allowed XT users to be a bit closer to the AT world.

The added instructions set was needed for certain performance related products,
like disk caches or drive compression programs.
Later releases refused to run on plain 8088/8086, even.

That's another point that saddens me.
The retro community refuses to acknowledge that the NECs were relevant.

Because, the first thing unhappy XT users did was to replace the 808x by a NEC.
It was a period-correct upgrade, almost as old as the first IBM PC.

But still, actions do show that the NEC chips are being ignored, as if they never had existed.
Emulators are the same. There's a Pentium II emulation available, but barely a NEC V20/V30 core. In the 2020s!

Of course, I do understand that there's a technical side, also.

The "8080 emulation mode" is hard to implement, so there's sort of a hesitation.
But the developers could use an 80186 core and make it more NEC-like.

That would be good enough to make old V20 PC BIOSes run or get Windows 3.0 EGA/VGA drivers going with the need for patching.

But that's another story, maybe.

It's just a bit sad at times that, sas, the PC-98 community has various emulators covering a handful of machine types in great detail,
whereas in our part of the hemisphere the focus is so strong on just a single configuration.

To be fair, though, here's more hardware homebrewing going on.

But maybe you understand what I'm trying to say, nevertheless.

What the OP essentially is asking, I think, is something like:
"Hey, Atari 2600 fans! Anyone interested in tinkering with the ColecoVision? Are there any projects planned ? 🙂"

PS: The AT Model 5170 was slow also because of the many, slow DRAM chips of the day.
They had to be used with waist states, which did slow down things quite a lot.
That's why the XT/286 was quite a bit more snappy, I think.

PS: What's also relevant is the first (?) 386-based PC compatible, the Compaq DeskPro 386.
It even looked like an angry AT Model 5170.

Source: https://en.wikipedia.org/wiki/Compaq_Deskpro_386

Edit: Now that I think of it, maybe it had been better if I hadn't replied to this thread (so much).
Personally, what I missed to take into account is how much the IBM PC means to some people. It's their C64, so to say.
Asking for a successor, like the equivalent to a C128, hurts their feelings and causes their refusal perhaps.
I mean, the Linux community is also a bit, um, religious at times when it comes their stuff/architectural changes.
Boy, there's so much to take care of these days. Things may be deeper than they seem at first glance.

Grzyb wrote on 2023-11-18, 11:32:

Sure, 8086 @ 8 MHz is much faster compared to 8088 @ 4.77 MHz. But it's hardly faster compared to 8088 @ 10 MHz. […]
Show full quote

Sure, 8086 @ 8 MHz is much faster compared to 8088 @ 4.77 MHz.
But it's hardly faster compared to 8088 @ 10 MHz.

And 8088 @ 10 MHz can be switched to 4.77 MHz, making it fully compatible with the earliest PC software.

8088 is therefore more universal than 8086 - and this is my answer to the question why there's no 8086 PC remakes.

Sounds plausible. I haven't benchmarked 8 Mhz 8086 vs. 10 MHz 8088. How close are their speeds in practice?