Amiga Thread.


I have an Apollo Firebird on the way, it's Essentially a Vampire V4 in an Accelerator configuration and it has a bit more features and memory than the V2 Vampire accelerators. It seems that they have dropped the Vampire name from their latest line of product, Firebird for (A1000,A500,A2000) and IceDrake for A1200. I think this is the case as these are being made by the Apollo team directly that creates the FPGA cores and not through a third party (Majsta or Kipper) like they have been in the past.


Only issue I have the computer I intended to put this in I haven't even started to assemble yet the new Replica A2000 motherboard... Prior post about that: https://pyra-handheld.com/boards/threads/amiga-thread.79281/post-1675268
 
Hmm I'd expect it to be faster than the Vampire, but still likely more obtainable and cheaper and it's good enough for most things Amiga.
When running Dr. Michal Schulz' JIT on PiStorm it actually is many times faster in the benchmarks, than Apollo Team's FPGA. I think the only way to make the Apollo Core faster is to prepare it for production as an ASIC using an equivalent die size as a RasPi 3.

Gunnar von Boehm, the head of the Apollo team was unprepared to do that though. The up-front costs of an ASIC are very high and if the production run of the Apollo SoC (System on a Chip) is insufficiently numerous, the costs of production will not be defrayed. That could drive the Apollo Team to bankruptcy. The stakes are too high.

As for the RasPi 400, it loses speed on the chipset emulation even if it runs on a separate core as it should.

The best of both worlds, in my opinion, would be a RasPi 4 compute module plugged into a board with an FPGA to run the chipset core only. This would make AMMX emulation pointless because the second thread of the Apollo 68080 core is used for blitter emulation. That would be unnecessary if the FPGA was dedicated to chipset implementation.
 
When running Dr. Michal Schulz' JIT on PiStorm it actually is many times faster in the benchmarks, than Apollo Team's FPGA. I think the only way to make the Apollo Core faster is to prepare it for production as an ASIC using an equivalent die size as a RasPi 3.

Gunnar von Boehm, the head of the Apollo team was unprepared to do that though. The up-front costs of an ASIC are very high and if the production run of the Apollo SoC (System on a Chip) is insufficiently numerous, the costs of production will not be defrayed. That could drive the Apollo Team to bankruptcy. The stakes are too high.

As for the RasPi 400, it loses speed on the chipset emulation even if it runs on a separate core as it should.

The best of both worlds, in my opinion, would be a RasPi 4 compute module plugged into a board with an FPGA to run the chipset core only. This would make AMMX emulation pointless because the second thread of the Apollo 68080 core is used for blitter emulation. That would be unnecessary if the FPGA was dedicated to chipset implementation.
Interesting, yeah I would suspect that as being faster, was just surprised that how the Pistorm interfaced with the A500 didn't make it faster.

I've talked to Gunnar quite a bit on IRC long ago, we both work in the semiconductor industry and at IBM Microelectronics at points in our careers, so yeah I know a bit about how much would be at stake with getting an ASIC to market, it would need a bit of sourcing of money and I doubt they would even attempt to do it unless they felt their core was at completely finished state, still to this date they're tweaking things.

Honestly I'm a fan of the original Amiga hardware, I like the Vampire approach of acceleration and how it works in tandem with the original chipsets while giving modern features to older hardware. While I respect the work done with things like the V4 Standalone, Mister and such they just doesn't interest me as much without the actual original physical hardware present. But my whole adventure with Amiga has been restoring the original hardware.
 
Started populating the A2000 PCB, so far it feels like I've done a ton work with just the regular resistors, diodes, ferrate beads and fuses, but it looks like I barely got anything done. This is already about 4-5 hours worth of work. I'm about to tackle around nearly 200 Ceramic capacitors soon...


IMG_20211031_111109480.jpeg
 
Worked a bit on the project today, more components in, looks like I forgot to order some electrolytic caps, those coming in Wednesday.

Potentially get IC sockets in tomorrow.

20211225_183132.jpg
 
Last edited:
Made considerable progress as it was a Holiday weekend mixed with a winter storm that allowed me to have some time to work on the project. I have two IC sockets I need to come in the mail and the IC sockets are done. Also I have a slight alignment issue on one of the Zorro II bus slots, it bugs me, and I will address it later, but want to make progress on getting this working and it's likely a slot I won't use anyway. After test fitting the Firebird Accelerator It looks like I will need to move two of the electrolytic capacitors around the 68000 slot and the RTC battery location is in the way too, luckily I have a coin style battery socket with wires.

20220102_214855.jpg
 
Last edited:
It lives!, still not 100% tested and still some quirkyness with either the OS or hardware... I need to run a thorough hardware test. But I have signs of life for sure. More to come.
 

Attachments

  • 20220113_180253.jpg
    20220113_180253.jpg
    1.1 MB · Views: 189
  • 20220112_213549.jpg
    20220112_213549.jpg
    1.3 MB · Views: 194
I see the Vampire board on there. That's a big boost to any real Amiga!
Yeah that's their new Accelerator card, The Firebird. Not sure why they dropped the Vampire from their naming scheme...

 
  • Like
Reactions: rSl
I think I discovered a bit of the quirkyness, it would run for a bit and then just freeze up on me, with a regular Motorola 68000 I could only get it to boot every once an a while. I think the accelerator is bypassing some of the functions of the Fat Angus chip and allowing it to work in some capacity, just not properly.

After experimenting I could get better results by swapping out the Fat Angus chip with another one I had on hand and I could get to the kickstart ROM more consistently... However in the process of swapping out these chips I damaged the PLCC socket and cracked it... so I desoldered it and waiting on a new one to come in, Along with a proper PLCC chip extractor...

Few steps forward and some steps back.
 
Last edited:
When Igor was kicked from the team, he took the Vampire v2 with him. The v4 boards have mostly new names now with Cyclone5 chips.
I know that now, but back in January when I commented, the information about Igor being kicked was still Nebulous...
 
Last edited:
Back
Top