Redesign the A3640

Chucky · 13 January 2016, 16:14

I got this crazy idea. why not do the A3640 on a new PCB. add the possability of 060..

then release all data. and let the community do its magic..
maybe remove GALs and change to FPGA. add memory etc..

idrougge · 13 January 2016, 16:31

If you do all that, where's the 3640?

Chucky · 13 January 2016, 16:34

where it is??

idrougge · 13 January 2016, 16:41

I mean, where's the 3640 in that new card, and why even try to clone the worst of all 040 cards?

Chucky · 13 January 2016, 16:44

ALL info of the 3640 is out. GALs etc. that's the reason. by doing that you can produce new cards that does 040/060 (060 on the 3640 is kinda ok:ish) then with that part of work done. maybe someone with knowledge of FPGA can take signals from the GALs. redo them. and then also maybe add memory etc...

do a new solution requires a lot of development. but by doing this.. no real developement is needed. just redo the work and store in a digital way.

hey still better then 030..

idrougge · 13 January 2016, 17:04

Once you add memory, you no longer have anything like the 3640 and the GAL equations can be thrown where they belong.ß

Chucky · 13 January 2016, 17:10

but then you will have a card that can be PRODUCED.

idrougge · 13 January 2016, 17:15

Yes, but the difference between a 3640 and the further development is so big that it will be two different cards.

StingRay · 13 January 2016, 17:45

Quote:

Originally Posted by idrougge

and why even try to clone the worst of all 040 cards?

That was exactly my first thought too!

Chucky · 13 January 2016, 18:09

people redsign the picasso II the absolutly worst gfxcard.. so why not

and as told before. this is the only 040+ accelerator where ALL info is avaible.

Tumo007 · 13 January 2016, 18:52

I guess it all depends on the current price for a fully modded A3640 and the estimated price of the new board.

The new board could include a larger CPLD to integrate all the GALs logic, replace the ugly delay line with a PLL and add support for the '060. It could be called the A3641

How much is the KMEL 200 pins connector anyway?

8bitbubsy · 13 January 2016, 18:59

This sounds like an interesting project.

With SpeedGeek's state machine hack, it's an OK'ish accelerator. Would be awesome to have support for 256MB of onboard fast RAM, onboard 060 socket and what not! It should've been called the A3660.

It probably is more time consuming and frustrating than what it's worth, though.

Chucky · 13 January 2016, 20:30

well.. it is timeconsuming yes.. but well done maybe someone can take that thing and improve stuff etc..

billt · 13 January 2016, 23:15

As hard as it is to get genuine 68060 parts, I'd skip that. Do an fpga only, using tg68, wf68k30, Apollo etc. Maybe in a zync to get memory controller and some soc peripherals from arm, the arm itself for fun, and fpga soft-periphs. The I'd myself make this new accelerator a carrier for comexpress module, which could then be an accelerator on an accelerator for ppc, but that's a but crazy.

Chucky · 13 January 2016, 23:22

well.. with 040/060 socket and jumpers.. so you can choose between 040 and 060. and hey.. rev 1 060s are quite easy to get hold of.. so I will think about this.

strim · 13 January 2016, 23:57

Guys. I don't want to discourage anyone. Actually, the opposite. But it's damn difficult to do something this complex.

@Chucky you might want to take a look at this thread (basically, a clone of A3630 + some stuff added = Decelerator 4030).

TjLaZer · 14 January 2016, 05:09

All these new '030 accelerators being produced now, I just don't get it. Why not 040/060??? While nice, they are just too slow. But I guess perfect for WHDload

Chucky · 14 January 2016, 07:17

Problem with 040/060 is the logic around it that is required.

hooverphonique · 14 January 2016, 13:38

Quote:

Originally Posted by TjLaZer

All these new '030 accelerators being produced now, I just don't get it. Why not 040/060??? While nice, they are just too slow. But I guess perfect for WHDload

The A3000 and A4000 motherboards were designed to fit a 030 directly, so a 030 cpu card is very simple.. 040/060 have a different interface and thus requires logic to translate it to be motherboard-compatible.

tassel · 14 January 2016, 13:46

Quote:

Originally Posted by hooverphonique

The A3000 and A4000 motherboards were designed to fit a 030 directly, so a 030 cpu card is very simple.. 040/060 have a different interface and thus requires logic to translate it to be motherboard-compatible.

Well, then someone should design a new A4000 PCB with Zorro and PCI slots ready for FPGA and 040/060 Cards

Hehe...

13 January 2016, 16:14	#1
Chucky Registered User Join Date: Mar 2015 Location: Karlstad / Sweden Age: 52 Posts: 1,210	Redesign the A3640 I got this crazy idea. why not do the A3640 on a new PCB. add the possability of 060.. then release all data. and let the community do its magic.. maybe remove GALs and change to FPGA. add memory etc..

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
C=a3640	Kin Hell	MarketPlace	19	27 May 2015 22:05
A3640	cosmicfrog	MarketPlace	1	19 February 2010 17:01
DiscreetFX website redesign	Pyromania	Amiga scene	0	21 September 2007 16:45
HOL Front Page Redesign?	Antiriad	HOL suggestions and feedback	3	13 December 2003 19:28
My Site Redesign..	Djay	Amiga scene	32	17 October 2003 23:21

13 January 2016, 16:31	#2
idrougge Registered User Join Date: Sep 2007 Location: Stockholm Posts: 4,338	If you do all that, where's the 3640?

13 January 2016, 16:34	#3
Chucky Registered User Join Date: Mar 2015 Location: Karlstad / Sweden Age: 52 Posts: 1,210	where it is??

13 January 2016, 16:41	#4
idrougge Registered User Join Date: Sep 2007 Location: Stockholm Posts: 4,338	I mean, where's the 3640 in that new card, and why even try to clone the worst of all 040 cards?

13 January 2016, 16:44	#5
Chucky Registered User Join Date: Mar 2015 Location: Karlstad / Sweden Age: 52 Posts: 1,210	ALL info of the 3640 is out. GALs etc. that's the reason. by doing that you can produce new cards that does 040/060 (060 on the 3640 is kinda ok:ish) then with that part of work done. maybe someone with knowledge of FPGA can take signals from the GALs. redo them. and then also maybe add memory etc... do a new solution requires a lot of development. but by doing this.. no real developement is needed. just redo the work and store in a digital way. hey still better then 030..

13 January 2016, 17:04	#6
idrougge Registered User Join Date: Sep 2007 Location: Stockholm Posts: 4,338	Once you add memory, you no longer have anything like the 3640 and the GAL equations can be thrown where they belong.ß

13 January 2016, 17:10	#7
Chucky Registered User Join Date: Mar 2015 Location: Karlstad / Sweden Age: 52 Posts: 1,210	but then you will have a card that can be PRODUCED.

13 January 2016, 17:15	#8
idrougge Registered User Join Date: Sep 2007 Location: Stockholm Posts: 4,338	Yes, but the difference between a 3640 and the further development is so big that it will be two different cards.

13 January 2016, 18:09	#10
Chucky Registered User Join Date: Mar 2015 Location: Karlstad / Sweden Age: 52 Posts: 1,210	people redsign the picasso II the absolutly worst gfxcard.. so why not and as told before. this is the only 040+ accelerator where ALL info is avaible.

13 January 2016, 18:52	#11
Tumo007 Registered User Join Date: May 2007 Location: Montreal, CANADA Posts: 56	I guess it all depends on the current price for a fully modded A3640 and the estimated price of the new board. The new board could include a larger CPLD to integrate all the GALs logic, replace the ugly delay line with a PLL and add support for the '060. It could be called the A3641 How much is the KMEL 200 pins connector anyway?

13 January 2016, 18:59	#12
8bitbubsy Registered User Join Date: Sep 2009 Location: Norway Posts: 1,711	This sounds like an interesting project. With SpeedGeek's state machine hack, it's an OK'ish accelerator. Would be awesome to have support for 256MB of onboard fast RAM, onboard 060 socket and what not! It should've been called the A3660. It probably is more time consuming and frustrating than what it's worth, though.

13 January 2016, 20:30	#13
Chucky Registered User Join Date: Mar 2015 Location: Karlstad / Sweden Age: 52 Posts: 1,210	well.. it is timeconsuming yes.. but well done maybe someone can take that thing and improve stuff etc..

13 January 2016, 23:15	#14
billt Registered User Join Date: Sep 2010 Location: Maryland, USA Posts: 82	As hard as it is to get genuine 68060 parts, I'd skip that. Do an fpga only, using tg68, wf68k30, Apollo etc. Maybe in a zync to get memory controller and some soc peripherals from arm, the arm itself for fun, and fpga soft-periphs. The I'd myself make this new accelerator a carrier for comexpress module, which could then be an accelerator on an accelerator for ppc, but that's a but crazy.

13 January 2016, 23:22	#15
Chucky Registered User Join Date: Mar 2015 Location: Karlstad / Sweden Age: 52 Posts: 1,210	well.. with 040/060 socket and jumpers.. so you can choose between 040 and 060. and hey.. rev 1 060s are quite easy to get hold of.. so I will think about this.

13 January 2016, 23:57	#16
strim NetBSD developer Join Date: May 2012 Location: Warsaw, Poland Posts: 411	Guys. I don't want to discourage anyone. Actually, the opposite. But it's damn difficult to do something this complex. @Chucky you might want to take a look at this thread (basically, a clone of A3630 + some stuff added = Decelerator 4030).

14 January 2016, 05:09	#17
TjLaZer Registered User Join Date: Sep 2004 Location: Tacoma, WA USA Age: 52 Posts: 1,918	All these new '030 accelerators being produced now, I just don't get it. Why not 040/060??? While nice, they are just too slow. But I guess perfect for WHDload

14 January 2016, 07:17	#18
Chucky Registered User Join Date: Mar 2015 Location: Karlstad / Sweden Age: 52 Posts: 1,210	Problem with 040/060 is the logic around it that is required.

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