Flea86 Retro Gaming System Project

35 replies [Last post]
Valentin Angelovski
Valentin Angelovski's picture
Offline
Joined: 12/24/2010

Flea86Flea86Dear All,

My name is Valentin and this is my first post to Armchair Arcade, although I've been lurking for a little while now :-)

I thought I'd start off with a little background, followed by some details on the biggest retro gaming project I have ever undertaken...

Well, I've been into electronic games since the early TV 'pong/tennis' clones hit the streets back in the late 70's/early 80's. My microcomputer gaming experience began with a Dick Smith VZ-200 (rebadged VTech laser-200) with star blaster bundled in (which was the bomb! lol). After finding this system was found wanting to do homework on, finally ended up with a XT-turbo clone with EGA in late 1987. At that time, there weren't terribly many gaming titles available (in Australia) for the IBM-PC (though that changed fairly quickly later on..).

Now, as uninspiring as 'old beige' might have seemed to some (especially to those who owned an Amiga or Atari ST), the machine made a huge impression on me... (please read on)

Fast forward to around six years ago, when I first got the idea to build my own video games console, initially based on one my best arcade favorites - Space invaders. Armed with a background in electronics along with around five weeks of Christmas holidays to kill, I set about building a tiny atari-2600(ish) clone of space-invaders called 'ALIENZ-2051'... http://www.oocities.com/vz300/vgc.htm

While the end-result worked well, I wasn't quite happy with the fact the ALIENZ-2051 game unit could only play one (admittedly awesome) game, so I decided to try again. I didn't get another chance however until around two years ago, when I decided that I would build own tiny retro-PC system (complete with custom mainboard, so no off-the-shelf mini/nano-ITX mainboard etc), capable of playing many classic PC (DOS) games...

To make it really small, I decided to use a tiny 8-bit processor and shoehorn-in an equally tiny custom PC emulator (entirely of my own design, since there was a 30KByte program code limit inside the processor!). This project evolved over a period of 18-months from IBM-PC (CGA games), extending to Tandy 1000 games (TGA) and finally through to PS/2 (MCGA/VGA) games. The following is the result of 18-months (part-time) worth of effort:

Introducing the tiny 'Flea86' PC/XT gaming system:

Early Flea86 system case (click on image to enlarge):

Latest Flea86 mainboard (click on image to enlarge):

Latest Flea86 running 'Dangerous Dave' (click on image to enlarge):

Youtube videos based on some early test hardware:
http://www.youtube.com/user/dosgamer74#p/u

Technical details about the project hardware can be found at my 'Tandy 1000 replica' web-corner:
http://members.optuszoo.com.au/pioneer10/index.xhtml

I welcome any comments, flames or distinct feelings of nostalgia this project might bring up. Thanks for reading! :-)

Sincerely,
Valentin Angelovski

Ryan (not verified)
Super Cool

Super Cool. Arcade games on PC i miss that

Valentin Angelovski wrote:

Flea86Flea86Dear All,

My name is Valentin and this is my first post to Armchair Arcade, although I've been lurking for a little while now :-)

I thought I'd start off with a little background, followed by some details on the biggest retro gaming project I have ever undertaken...

Well, I've been into electronic games since the early TV 'pong/tennis' clones hit the streets back in the late 70's/early 80's. My microcomputer gaming experience began with a Dick Smith VZ-200 (rebadged VTech laser-200) with star blaster bundled in (which was the bomb! lol). After finding this system was found wanting to do homework on, finally ended up with a XT-turbo clone with EGA in late 1987. At that time, there weren't terribly many gaming titles available (in Australia) for the IBM-PC (though that changed fairly quickly later on..).

Now, as uninspiring as 'old beige' might have seemed to some (especially to those who owned an Amiga or Atari ST), the machine made a huge impression on me... (please read on)

Fast forward to around six years ago, when I first got the idea to build my own video games console, initially based on one my best arcade favorites - Space invaders. Armed with a background in electronics along with around five weeks of Christmas holidays to kill, I set about building a tiny atari-2600(ish) clone of space-invaders called 'ALIENZ-2051'... http://www.oocities.com/vz300/vgc.htm

While the end-result worked well, I wasn't quite happy with the fact the ALIENZ-2051 game unit could only play one (admittedly awesome) game, so I decided to try again. I didn't get another chance however until around two years ago, when I decided that I would build own tiny retro-PC system (complete with custom mainboard, so no off-the-shelf mini/nano-ITX mainboard etc), capable of playing many classic PC (DOS) games...

To make it really small, I decided to use a tiny 8-bit processor and shoehorn-in an equally tiny custom PC emulator (entirely of my own design, since there was a 30KByte program code limit inside the processor!). This project evolved over a period of 18-months from IBM-PC (CGA games), extending to Tandy 1000 games (TGA) and finally through to PS/2 (MCGA/VGA) games. The following is the result of 18-months (part-time) worth of effort:

Introducing the tiny 'Flea86' PC/XT gaming system:

Early Flea86 system case (click on image to enlarge):

Latest Flea86 mainboard (click on image to enlarge):

Latest Flea86 running 'Dangerous Dave' (click on image to enlarge):

Youtube videos based on some early test hardware:
http://www.youtube.com/user/dosgamer74#p/u

Technical details about the project hardware can be found at my 'Tandy 1000 replica' web-corner:
http://members.optuszoo.com.au/pioneer10/index.xhtml

I welcome any comments, flames or distinct feelings of nostalgia this project might bring up. Thanks for reading! :-)

Sincerely,
Valentin Angelovski

Valentin Angelovski
Valentin Angelovski's picture
Offline
Joined: 12/24/2010
Re: What is the status of this..

Hi Omar,

Thanks for your comments. You are quite right - it has been a while since my last update! Flea86 is still an active project and things have progressed since then (though not as quickly as I would have liked due to other commitments). Will try to post a new blog update in the next few days.

Regards,
Valentin Angelovski

Omar Bomrad (not verified)
What is the status of this

What is the status of this proyect? Any updates? We have no news for almost a year. Is it for sale? I'd like to try some games with this!

Valentin Angelovski
Valentin Angelovski's picture
Offline
Joined: 12/24/2010
Flea86 development blogs

Dear all,

Just a brief note to those who are following this project with interest - I will be posting regular development blog updates (like this one). These updates will chart the overall progress of my Flea86 project, so please look out for my update blogs and enjoy! :-)

Thanks and regards,
Valentin Angelovski

Chip Hageman
Chip Hageman's picture
Offline
Joined: 10/06/2010
.-·^·-.
Valentin Angelovski wrote:
Nathaniel Tolbert wrote:

The PPC is the equivalent of the Pentium chip from intel in terms of performance and speed.

True, though most definitely not in terms of complexity - RISC architectures like the ARM and PPC presumably do not impose overly taxing demands on the overall resources within an FPGA, probably much less so than anything x86-related..

While the gate count ultimately speaks to the potential design complexity, you don't need to reproduce the hardware at a gate level.. you just need to be able to reproduce the correct results. I'm not sure how Hardware Description Languages (HDL's) and FPGA's hold up against Moore's law- but it seems to me that FPGA cores are really just hardware emulation layers.. so transistor count / performance may be way different.

Also, later Intel chips are not exclusively CISC.. there is a lot of RISC concepts swirled into the design as well. Like I said, if you can emulate it well enough.. why worry about the "gate level" details?

n/a
Chip Hageman
Chip Hageman's picture
Offline
Joined: 10/06/2010
.-·^·-.
Nathaniel Tolbert wrote:

No, that is not true. While looking through designs, they have in certain FPGA's a pre-programmed CPU that you can replace. The average speed of these is 300MHz. Not bad when you consider that you wouldn't need anywhere near that speed. Judging from the design information that I can find, the only reason that the Zet Project is currently sitting at 12.5MHz is down to just needing to fine tune the speed. Every indicator I see points to an 100MHz FPGA 386 being very much possible. It wouldn't be practical at this point because it's cheaper to just purchase an old 386, but it is doable.

I was talking about the impressiveness of being able to get away with using a microcontroller as a dedicated CPU. I have no doubt you could reproduce the functionality of a 386 processor with an FPGA, although I haven't heard of any projects doing this yet.

n/a
Nathaniel Tolbert
Nathaniel Tolbert's picture
Offline
Joined: 11/06/2010
I don't think so.
Chip Hageman wrote:

I think it's clear that any real horsepower is going to require an actual CPU with an emulation layer or an FPGA with a completed core design for the target system.

That said, what Valentin has pulled off with a microcontroller (which are really only for simple dedicated tasks; say, controlling the cooking time in a microwave) is beyond mind-blowing. I'm not sure which MC is being used in the design, but these things are really only meant to replace a handful of logic chips.

No, that is not true. While looking through designs, they have in certain FPGA's a pre-programmed CPU that you can replace. The average speed of these is 300MHz. Not bad when you consider that you wouldn't need anywhere near that speed. Judging from the design information that I can find, the only reason that the Zet Project is currently sitting at 12.5MHz is down to just needing to fine tune the speed. Every indicator I see points to an 100MHz FPGA 386 being very much possible. It wouldn't be practical at this point because it's cheaper to just purchase an old 386, but it is doable.

n/a
Valentin Angelovski
Valentin Angelovski's picture
Offline
Joined: 12/24/2010
CISC vs RISC
Nathaniel Tolbert wrote:

Remember, transistor count doesn't equate to logic gate use.

Correct, but what is even more important (particularly for FPGA's from what I understand) is what routing scheme is used. This ultimately determines what kind of 'tax' that will represent on the (somewhat limited) interconnect resources of the FPGA. It's the limited routing resources (and not necessarily the gate count) available on the FPGA that present the biggest obstacle (my opinion) to the implementation of high-end CISC architectures like the 386/486 and up

The table I posted, was meant to show that roughly up to an order-of-magnitude in total FPGA resources may be required, in order to jump from 8086 to 80386.

Nathaniel Tolbert wrote:

The PPC is the equivalent of the Pentium chip from intel in terms of performance and speed.

True, though most definitely not in terms of complexity - RISC architectures like the ARM and PPC presumably do not impose overly taxing demands on the overall resources within an FPGA, probably much less so than anything x86-related..

Nathaniel Tolbert wrote:

If that were true then that PPC FPGA design would be massive and it only takes up 135000 logic gates, and it's rated at a speed of 150MHz.

It may also help that IBM probably spent a shipload of R&D dollars optimizing/tweaking their lowest-end PPC405 softcore for use inside a commodity FPGA?

Once again, these are just my own opinions.. Perhaps someone more qualified in FPGA-systems design than myself could provide a more accurate insight i.e. perhaps a member the zet processor design team?

Cheers Valentin

Chip Hageman
Chip Hageman's picture
Offline
Joined: 10/06/2010
.-·^·-.-·^·-.-·^·-.-·^·-.-·^·-.-·^·-.-·^·-.-·^·-.-·^·-.-·^·-.-·^
Nathaniel Tolbert wrote:

Remember, transistor count doesn't equate to logic gate use. If that were true then that PPC FPGA design would be massive and it only takes up 135000 logic gates, and it's rated at a speed of 150MHz. The PPC is the equivalent of the Pentium chip from intel in terms of performance and speed. That being said, without knowing the exact set of instruction capabilities on the PPC chip, I couldn't tell you if they removed redundant features from the chip, or if everything is cycle exact. I want to find out more information about this sort of idea, but unfortunately between work, school, and my regular podcast (which is currently readying everything for this years KantCon, 3rd year.) I don't have the time to research anything too carefully.

I think it's clear that any real horsepower is going to require an actual CPU with an emulation layer or an FPGA with a completed core design for the target system.

That said, what Valentin has pulled off with a microcontroller (which are really only for simple dedicated tasks; say, controlling the cooking time in a microwave) is beyond mind-blowing. I'm not sure which MC is being used in the design, but these things are really only meant to replace a handful of logic chips.

n/a
Nathaniel Tolbert
Nathaniel Tolbert's picture
Offline
Joined: 11/06/2010
Transistor count
Valentin Angelovski wrote:

CPU Transistor count
--------------------------------------
8086 ~29,000
68000 ~68,000
80386 ~275,000
Cheers Valentin

Remember, transistor count doesn't equate to logic gate use. If that were true then that PPC FPGA design would be massive and it only takes up 135000 logic gates, and it's rated at a speed of 150MHz. The PPC is the equivalent of the Pentium chip from intel in terms of performance and speed. That being said, without knowing the exact set of instruction capabilities on the PPC chip, I couldn't tell you if they removed redundant features from the chip, or if everything is cycle exact. I want to find out more information about this sort of idea, but unfortunately between work, school, and my regular podcast (which is currently readying everything for this years KantCon, 3rd year.) I don't have the time to research anything too carefully.

-edit- while researching I found something interesting, by rough count your transistor count to logic gate count is as follows logic gate=4*transistor count. So if I understand this correctly you make use of one logic gate and that equates to 1/4 of a transistor? Or do I have it backwards?

-edit2- while researching further I found a nice power point presentation that shows something interesting. An FPGA with a gate count of 250K can implement a transistor count of 75 million Not that much if you are implementing full system internally but still quite a large number of transistors. But when I run the numbers, they don't match up.. Look 75,000,00/250,000 /= 4 it equals 300. So which is correct? This power point presentation with information pulled from a Professor's presentation on FPGA's? Or information I found in a forum? I don't know at this point.

n/a

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Images can be added to this post.
  • You may quote other posts using [quote] tags.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.