Is it just me, or does this video feel incomplete? Right when it was getting started, it just suddenly ends! Dude spends all this time talking about the history of putting a Pi into an Amiga, and then just ends the video without doing so himself. I'm disappointed.

Damn straight! Love to you all in blighty from the antipodes!

I must admit I have a couple of TF accelerators which I fujd are great, as I do like running stuff on orgianal Motorola chips, and 030s are still relatively easy to get hold of. Its the cost of 060 chips (and all the fakes) thats making it harder for me to go for an 060 in all the machines I would like to.

@@RetroBytesUK You can blame me for that. After the 1260 release the 060s went from £90 for a rev6 060 to £250 overnight.

@@TerribleFire no good deed and all that!

Please listen to me when I write this. I'm an electrical engineer, and although I don't work in the field, I'm trained as a VLSI designer, I.E. I was trained to make chips. This job sucks, and I'm glad I didn't stick with it, but I know the field. There's no difference between original hardware and a software simulator of the hardware, IF the hardware if fast enough. If I were to plug in a raspberry pi into say, a PS1 or an Amiga or SNES, and throw in a simulator into that, and do some tricks to get the NTSC output work and connect up all the original hardware to connect to the pi, you would NEVER KNOW THE DIFFERENCE between that, and the real thing. You might be able to detect it, if you hooked up a frame grabber, or checked color levels, or did extreme timing tests with external hardware, but you'd never be able to tell the by using the machines. There's no advantage to original hardware.

@@fuzzywzhe The difference is that i know its not original. Retro is all about nostalgia and I liken it to classic car collecting. Its irrelevant that you can or cannot tell the difference. EDIT: i'm also an electrical engineer trained in VLSI and have been designing accelerators for 7+ years.

Thanks that's nice of you to say.

Hell I'd take an A600 if only to just have an amiga again. But other than that happy with my amiga mini I made with a pi 400 + amiga forever.

FWIW I can confirm this was my personal experience too. I do recall playing a boat racing game which may have been called "Rapids". Beyond this I didn't use a BBC Micro until a vintage computer fare where I wrote a simple BASIC program.

Pretty much the same. They were around, but not used much. Not really acceptable, the amount of tax went into buying them, they should have been used constantly untill the letters wore off the keys.

Whereas we had a couple of really good math and science teachers who ran break/lunchtime and out of hours computer sessions for any of the kids who were interested and just encouraged us to get involved … if all you wanted to do was play games then fine but a number of us were doing programming in assembler too.

If a teacher said I could stay after hours to use the computers, I would have been so happy, waking up, excited to go to school. But, nope, they made it boring, and I hated every second of it.@@philharris9631

On the HD front I think it fits the broader point I was making, that someone has to write the code to drive the HW directly to access storage. There is no OS routine, that can just be used that abstracts out the hw, its a much harder road to walk than faking a basic block storage controller and calling fopen, fread, fwrite to do all the work on storing it. I assume there must still be exposed device on the pi side of things that your code is using the access the pis hw. I take it your code provides an alterative to scsi.device thats deals with spi its self then.

The library I mentioned at the beging is in use for pi tube direct, which is what I ment at that point. Re-using the term baremetal pi, to also mean just targeting the hw directly (baremetal) was probably not the best choice on my behalf however. As i did create a situation where I could be easliy miscontruded when it came to emu68. Sorry for the clumsy turn of phrase there. I must admit I also had thought the driver project was part and parcel of the emu68 project. So thanks for pointing that out.

@@RetroBytesUK only one thing is done on pi side - mmu mapping which exposes the entire MMIO space of raspberry (all peripherals) within 32bit address space. The rest is done on m68k side - basically the device driver which follows the same protocol (API) as any other driver. Thanks to that AmigaOS can communicate with it as with any other hard drive, no arm involved. Still, it is not Emu68 code anymore.

Also thanks for the work on the driver much appreciated.

@@RetroBytesUK you are welcome! I must admit the entire Emu68, the way it evolved from my initial idea, is a great journey and I feel absolutely happy hearing that so many people are using it :) when started I never expected it to be where it is today :)

I suspect exposing the raw ethernet device is the easiest to achieve and get working. As they could very thinly wrap the physical device so its addressable by the emulated 68k, then its a case of creating a SANA-II driver, like you would for other ethernet cards. As the reset of the networking stack exists in Amiga OS nobody would need to write that again. Wifi would be a whole different keetle of fish as you would need the wrieless stack, including all the WPA suplicant stuff.

Thanks Scott, thats very kind of you.

A1200's built-in CPU is 68EC020 which has a 24-bit address bus. The trap door's 24-bit or 32-bit address capability is based on the active host 68K CPU.

I did decide to gloss over that detail a little, as I did not want to make the video longer, and I thought it might confuse poeple trying to explain how the slot does allow for 32bit addressing, yet as part of the cost reduction in the ec version of the 68020 there is not the 32 address pins. The main bit I was hoping to get across was the increase in the number of address lines pi storm 32 had to deal with.

Thanks for rasing it in the comments section, as for those interested in that level of detail its usful to know.

The trapdoor exposes 32 bits of address bus, but the topmost 8 bits are free floating on a1200 board

@@MichalSchulz Yes there is nothing inside the A1200 that lives in the part of the address space as the ec version of the 68020 could not address it.

Indeed it would be. The v4 standalone does not need any original amiga bits. There is a v4 based accelerator for the A1200 called ice drake.

Or a MiSTer, if you want to have the CPU reimplemented in the FPGA, or a minimig, if you'd rather have a real 68080 in there. Quite a few options, actually!

There is still a fair bit, the original gfx/sound custom chips are still in use for most games. All the original input devices, keyboard, mouse, and joystick are all still in use. You also still have the abilitiy to use orginal floppy disks and drives.

There already is one for the Atari

That is true for the original PiStorm (although there are plans to make this more universal in general, as it solves Amiga problems too). For the PiStorm32-Lite, the entire bus is implemented, the Verilog was basically redone from scratch.

@@LemaruX The PiStormSTE version is still a work in progress. Atari uses FC lines and depends heavily upon bus arbitration and interrupts. The Atari version doesn't have Amiga's fast Emu68. PiStorm32 Lite is a separate development and uses Efinix FPGA. PiStorm32 is canceled due to PCIe latency.

@@semicuriosity257 I am well aware that the Atari development is still in development. You said a branch is needed for the Atari, and I just pointed out that there already is one. And no, PiStorm32 is not cancelled. The PCIe latency issue is related to the Raspberry Pi 5, NOT the PiStorm32

@@LemaruX From Claude Schwarz @Claude1079 Sep 30 _Did some experiments on CM4. Doing timed mmio reads from a PCIe Gen2 controller memory BAR region._ _Yeah looks like the internet is right, PCIe mmio latency is high. ~1.3us , average access time over 1 million reads. Meh_

The HW requirements for the amiga version are so incredibly high hardly anyone could run it. As you needed a PPC accelerator, and a 3d acceleratored RTG card that was supported by a particular 3d library. So it was the nitch, of the nitch, of the nitch. I have finally got the right 3d card, but the case pyhisically gets in the way of me fitting it, so I prpbably need to make a relocator board, assuming I can find the connector it uses.

I worked on the original PlayStation version of Wipeout (and the subsequent versions)… at the time I was doing a lot of work for Psygnosis and was handling the translation of their (natively English) games into foreign languages including all the audio assets used in the games.

As one or two people have mentioned it, I went crawling through to spot where it was. It looks to be a rendering problem I did not spot. In the editor the media is there, and the frame is correct with the media there, in the rendered output there is a big red media missing error. No idea how davinci managed that one.

The vampire started off with similar goals to make a cheap accelerator, and for a while the Vampire 2 for the A600 was relatively cheap its just the price of FPGAs moved significantly on them. So they where left with a choice of just stop, or target performance, and go with the rising costs. They chose to keep going. It does mean for those that can afford it there is a premium high performance option, and there is a much more affordable option in the form of the pistorm for the reset of us, which while not quiet hitting the vampires performance level does let us do things that where unobtainable to us back in the day. It would be nice if the two projects could co-operate and standardise AMMX so pistorm could implement that extension to the CPU too.

The orginial chip set is in use for none rtg graphics, orginal peripherals works, like joysticks, and on machines that are not the A600 lots of other hw expansions. You can also use your orignal floppy disks.

The only emulated part is the CPU. The rest (RTG and hard disk) is direct drivers to the underlying Pi hardware. The entire Amiga chipset is used as normal.

@@linuxjedivideo This is a 100% accurate statement 🙂

​@@EdgeOfPanic I know, I'm one of the PiStorm community admins and was a developer of the legacy Linux based PiStorm code 😊

For driving the ethernet port, borrowing the code from risc os would work, saddly risc os has no wifi support to borrow from. I think utilising the IP stack from risc os would prove tricky, as while its a good ip stack targetting arm, it would need to run of the Amiga OS side of the stack so amiga apps can make use of it

@@RetroBytesUK From Feb 2022 - "First steps to Wifi released yielding increased SD performance: RISC OS Developments are pleased to announce the release of the first steps towards Wifi on RISC OS - updated SDIO drivers for most platforms making the on-board wifi chips visible/accessible." From Dec 2023: "ROD TCP/IP Stack 7.04 released: We are pleased to announce the release of v7.04 of the ROD TCP/IP stack supporting IPv6 and many other modern network technologies." If you read the announcement, they hint the WiFi drivers, for the new Stack, will follow in the New Year. So there's a bare metal, ARM optimised, Apache licensed, C source solution, that can configure and create an abstraction for the PI's and several other ethernet/wiFi adaptors, there's code that creates a virtual RISC OS network adaptor, with communication likely handled through a couple of Words, in a flat 32-bit, common memory model, on setting / handling a few interrupts. There's a new IPv6 RISC OS network stack, that adds all the layers atop what it picks up / sets in a couple of words, on an interrupt being handled or set, using no more than C and its standard libraries, in a simple, flat, common, 32 bit memory map. I'd at the very least have a trawl of the source, for inspiration, if tasked with doing almost the exact same thing, in a very similar architecture, but the application layer in 680x0, and hooked into Emu68.

I think its great that we have options as you said. I think price point is the key concideration, I would love todo everthing with orginal hardware, nothing would ammuse me more than putting an 060 in an A600. Its just I'm not sure my wife would forgive me for spending the money required todo that. I'll aettle for pistorm for this A600 though, and my kids are loving playing with it now its in the living room.

The point is that the original custom chips are still used, that’s where the magic is 👍🏻

If they has survived long enough the zip or ls120 disks would have been a great choice of removable media for the amiga. Alot of big box amgia onwers (and A1200 tower conversions) went with zip drivers or later on ls120. If commodore had hit the original price point they intended to, and released it as the A300 there would be much warmer memorirs for the A600. I quiet like the A600 now, I like its form factor, and its a nice addition to the Amiga lineup. At the time it felt like a big mistake, as they marketed as an upgrade to the A500, and stopped producing the A500 for what felt like a worse machine. When they finally release the A1200 the range and the A600 made so much more sense.

Its definitely worth getting one, and there is a huge software collection to explore.