A New A4000 Accelerator-Wishful Thinking

grelbfarlk · 31 August 2018, 01:28

Before I ruin another thread, I'll just start a new one.

So it goes like this, we don't have any A4000/A3000 accelerators as good as the classic ones like:
-Warpengine
-Cyberstorm
-Trex-II
-Apollo A4060
-Quikpak A4060 XP

Since these Kickstarters are all the rage I'm wondering what a reasonable goal or cost for hiring someone to build one would be. Or even buy the design to one of the above, but that doesn't seem possible.

Yes Chucky's A3660 is a fine achievement and fills the gap of building a new A4000 out of newly designed motherboards, the lack of a fast HDD controller and onboard memory is something that should be corrected.

So in a perfect world base features that would be required:
-040/060 socket
-SDRAM or better, can be soldered on or SODIMM socket.
-DMA HDD Controller: IDE, SCSI, or SATA (or something else?)

Stretch goals which might greatly increase the complexity or cost:
-Some kind of 040/060 local bus connector that could eventually be used for a fast PCI Interface.
-Built in PCI bus controller
-SPI bus
-USB
-Something else?

And before someone says it, the Vampire team doesn't provide board designs, so nobody other than them can reproduce them if they decide to stop assembling them. Which is fine, that's their choice.

I wouldn't expect the board to cost less than $500, which is fine actually would be amazingly cheap compared to what an A4000 accelerator costs nowadays, A3660 excluded.

So what's a reasonable cost to hire out something like this for the base features, $50,000? $100,000?

Or would it make more sense to take the A3660 design, reintegrate all the GALs into a larger CPLD and add the rest of the stuff.

Or take the GBA-060 board design and adapt that, adding an HDD controller.

The goal is not to make money, just to provide an advanced A4000 accelerator for the community and people can do the whole SAMMELBESTELLUNG thing if they want to.

alexh · 31 August 2018, 09:02

The problem is that the vampire V4 is just around the corner and they are touting A4000/A3000 compatibility. MikeJ of FPGArcade is touting hybrid FPGA CPU (running a JIT emulator on an embedded multi GHz ARM inside next gen FPGA's) and finally the scarcity of Rev6 68060 CPUs (lots of Chinese fakes) I don't think there's much enthusiasm to spend the NRE on remaking a classic A4000/A3000 accelerator

Mr.Flibble · 31 August 2018, 09:18

SakuraIT made a start with their 030 CPU card - https://github.com/Sakura-IT/decelerator4030

Everything is open source so it might make a good starting point, I believe they were mulling a more advanced option at one point, but they’ve been quiet for a little while now.

AJCopland · 31 August 2018, 11:45

If Chucky can get the additional memory adapter working and then integrated into a newer revision of the A3660 then you're already most of the way too a new accelerator.

The DMA HDD connector though... are there any open examples that could be integrated?

Mr.Flibble · 31 August 2018, 18:35

That's a good point, I missed that Chucky had started to go down that route.

Getting rid of all those impossible to find legacy components is a damned good step.

NorthWay · 31 August 2018, 19:29

SATA. I don't think there is any native SATA controller so far (no, bridge solutions don't count in my eyes) and I'm tired of spinning rust so that would be utterly important IMO.

Someone do a Zorro 2/3 (autosensing) one if not a full cpu card...

patrik · 31 August 2018, 20:30

Quote:

Originally Posted by NorthWay

SATA. I don't think there is any native SATA controller so far (no, bridge solutions don't count in my eyes) and I'm tired of spinning rust so that would be utterly important IMO.

A good DMA IDE-controller designed to only be used with with an onboard IDE->SATA-bridge should be fine. Using a bridge would most likely make it much cheaper and less complicated to build as it is a bit tricky to interface directly to sata.

You would not notice that it was using a bridge and the speed you experience will only be a factor of the quality of the controller implementation.

tnt23 · 31 August 2018, 21:40

As a proud owner of A3000 with 060 and thus non-working SCSI, I'd appreciate SD or even plain IDE

Hedeon · 31 August 2018, 22:27

Quote:

Originally Posted by tnt23

As a proud owner of A3000 with 060 and thus non-working SCSI, I'd appreciate SD or even plain IDE

Thus?....How so?

trixster · 31 August 2018, 22:31

Yeah, I have an 060 in my A3000 and scsi works fine

zipper · 31 August 2018, 22:41

Quote:

Originally Posted by Hedeon

Thus?....How so?

Probably missing Int2 mod?

plasmab · 31 August 2018, 22:55

Quote:

Originally Posted by Mr.Flibble

SakuraIT made a start with their 030 CPU card - https://github.com/Sakura-IT/decelerator4030

Everything is open source so it might make a good starting point, I believe they were mulling a more advanced option at one point, but they’ve been quiet for a little while now.

That project is years old.

amiman99 · 31 August 2018, 22:58

Quote:

Originally Posted by alexh

(lots of Chinese fakes)

What is a "Chinese fake CPU", can you explain?
Thanks

idrougge · 31 August 2018, 23:02

http://www.amigawiki.de/doku.php?id=...k#faelschungen

plasmab · 31 August 2018, 23:05

Quote:

Originally Posted by idrougge

http://www.amigawiki.de/doku.php?id=...k#faelschungen

Summary version is that people take low end chips, scrape the markings off them and reprint the markings for the best known version on them.

When they do this they get a ridiculous markup.

Given that a high end 060 is 120 quid buying a batch of them is a high risk game.

plasmab · 31 August 2018, 23:07

Ultimately the reason we dont have these accelerators is because the risk is too high for such a small market. Although the demand for these accelerators is high and the supply is low the risk to create new ones based on the old 060 accelerators is too high.

Its why I plan to do an 040 accelerator not an 060 one.

plasmab · 31 August 2018, 23:21

Let me put another way. The board you want to make is about 10-50 times more complex than a raspberry pi (because the pi is mostly a SOC). How much did the pi cost to design? I know Eben did some initial prototypes but he eventually gave it to a board designer to do.

People keep wishing for this but the design is actually the easy bit. Its the years of testing @ around 500 euros a day for a contract electronics engineer (because you arent going to get someone to leave their job and do this for normal wages.. they'll want contract rates).

nexus · 31 August 2018, 23:29

maybe that will change soon since there is a company making 68k again
https://www.embedded.com/electronics...s-back-to-life

plasmab · 31 August 2018, 23:30

Quote:

Originally Posted by nexus

maybe that will change soon since there is a company making 68k again
https://www.embedded.com/electronics...s-back-to-life

Dated "NOVEMBER 11, 2014".. it hasnt happened.

nexus · 31 August 2018, 23:32

yes it has just not 68060's etc yet

31 August 2018, 01:28	#1
grelbfarlk Registered User Join Date: Dec 2015 Location: USA Posts: 2,902	A New A4000 Accelerator-Wishful Thinking Before I ruin another thread, I'll just start a new one. So it goes like this, we don't have any A4000/A3000 accelerators as good as the classic ones like: -Warpengine -Cyberstorm -Trex-II -Apollo A4060 -Quikpak A4060 XP Since these Kickstarters are all the rage I'm wondering what a reasonable goal or cost for hiring someone to build one would be. Or even buy the design to one of the above, but that doesn't seem possible. Yes Chucky's A3660 is a fine achievement and fills the gap of building a new A4000 out of newly designed motherboards, the lack of a fast HDD controller and onboard memory is something that should be corrected. So in a perfect world base features that would be required: -040/060 socket -SDRAM or better, can be soldered on or SODIMM socket. -DMA HDD Controller: IDE, SCSI, or SATA (or something else?) Stretch goals which might greatly increase the complexity or cost: -Some kind of 040/060 local bus connector that could eventually be used for a fast PCI Interface. -Built in PCI bus controller -SPI bus -USB -Something else? And before someone says it, the Vampire team doesn't provide board designs, so nobody other than them can reproduce them if they decide to stop assembling them. Which is fine, that's their choice. I wouldn't expect the board to cost less than $500, which is fine actually would be amazingly cheap compared to what an A4000 accelerator costs nowadays, A3660 excluded. So what's a reasonable cost to hire out something like this for the base features, $50,000? $100,000? Or would it make more sense to take the A3660 design, reintegrate all the GALs into a larger CPLD and add the rest of the stuff. Or take the GBA-060 board design and adapt that, adding an HDD controller. The goal is not to make money, just to provide an advanced A4000 accelerator for the community and people can do the whole SAMMELBESTELLUNG thing if they want to. Last edited by grelbfarlk; 31 August 2018 at 01:37. Reason: random spaces added to post

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Why anyone is thinking to develop an accelerator for CDTV?	Omolungo	Amiga scene	13	25 March 2017 20:51
gog.com for retro games (wishful thinking)	jman	Retrogaming General Discussion	17	01 December 2011 20:51
Thinking of selling my A4000's ...	Retro-mania	Amiga scene	10	01 February 2007 08:51
WANTED: A4000 accelerator and daughterboard	T_hairy_bootson	MarketPlace	16	24 October 2006 13:56
Ebay: Cheap accelerator for A3000 and A4000	Smiley	MarketPlace	10	26 March 2005 12:07

31 August 2018, 09:02	#2
alexh Thalion Webshrine Join Date: Jan 2004 Location: Oxford Posts: 14,337	The problem is that the vampire V4 is just around the corner and they are touting A4000/A3000 compatibility. MikeJ of FPGArcade is touting hybrid FPGA CPU (running a JIT emulator on an embedded multi GHz ARM inside next gen FPGA's) and finally the scarcity of Rev6 68060 CPUs (lots of Chinese fakes) I don't think there's much enthusiasm to spend the NRE on remaking a classic A4000/A3000 accelerator

31 August 2018, 09:18	#3
Mr.Flibble Registered User Join Date: Jun 2015 Location: UK Posts: 472	SakuraIT made a start with their 030 CPU card - https://github.com/Sakura-IT/decelerator4030 Everything is open source so it might make a good starting point, I believe they were mulling a more advanced option at one point, but they’ve been quiet for a little while now.

31 August 2018, 11:45	#4
AJCopland Registered User Join Date: Sep 2013 Location: Beeston, Nottinghamshire, UK Posts: 238	If Chucky can get the additional memory adapter working and then integrated into a newer revision of the A3660 then you're already most of the way too a new accelerator. The DMA HDD connector though... are there any open examples that could be integrated?

31 August 2018, 18:35	#5
Mr.Flibble Registered User Join Date: Jun 2015 Location: UK Posts: 472	That's a good point, I missed that Chucky had started to go down that route. Getting rid of all those impossible to find legacy components is a damned good step.

31 August 2018, 19:29	#6
NorthWay Registered User Join Date: May 2013 Location: Grimstad / Norway Posts: 839	SATA. I don't think there is any native SATA controller so far (no, bridge solutions don't count in my eyes) and I'm tired of spinning rust so that would be utterly important IMO. Someone do a Zorro 2/3 (autosensing) one if not a full cpu card...

31 August 2018, 21:40	#8
tnt23 Registered User Join Date: Feb 2008 Location: Saint-Petersburg / Russia Posts: 324	As a proud owner of A3000 with 060 and thus non-working SCSI, I'd appreciate SD or even plain IDE

31 August 2018, 22:31	#10
trixster Guru Meditating Join Date: Jun 2014 Location: England Posts: 2,337	Yeah, I have an 060 in my A3000 and scsi works fine

31 August 2018, 23:02	#14
idrougge Registered User Join Date: Sep 2007 Location: Stockholm Posts: 4,332	http://www.amigawiki.de/doku.php?id=...k#faelschungen

31 August 2018, 23:07	#16
plasmab Banned Join Date: Sep 2016 Location: UK Posts: 2,917	Ultimately the reason we dont have these accelerators is because the risk is too high for such a small market. Although the demand for these accelerators is high and the supply is low the risk to create new ones based on the old 060 accelerators is too high. Its why I plan to do an 040 accelerator not an 060 one.

31 August 2018, 23:21	#17
plasmab Banned Join Date: Sep 2016 Location: UK Posts: 2,917	Let me put another way. The board you want to make is about 10-50 times more complex than a raspberry pi (because the pi is mostly a SOC). How much did the pi cost to design? I know Eben did some initial prototypes but he eventually gave it to a board designer to do. People keep wishing for this but the design is actually the easy bit. Its the years of testing @ around 500 euros a day for a contract electronics engineer (because you arent going to get someone to leave their job and do this for normal wages.. they'll want contract rates).

31 August 2018, 23:29	#18
nexus Registered User Join Date: Aug 2017 Location: USA Posts: 728	maybe that will change soon since there is a company making 68k again https://www.embedded.com/electronics...s-back-to-life

31 August 2018, 23:32	#20
nexus Registered User Join Date: Aug 2017 Location: USA Posts: 728	yes it has just not 68060's etc yet

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)