CTPCI development files released

[mikro]

I'm happy to announce that Rodolphe Czuba has kindly agreed to release everything he had on the CTPCI. That means not only PCB drawings but also ABL sources, bill of materials, datasheets, timing diagrams, even price calculation for the whole thing.

In his own words:

CTPCI logic is very complex.

The interface was designed for Single transfers (1 32 bit LONG word with CPU and PCI DMA) and the BURST transfers (4 longs with 060 and from PCI DMA) where planned but the logic inside CPLD is not finished and may need a bigger CPLD to be really functional. Thjis design is very difficult to elaborate and my ABL language is not enough powerfull to do that easely...Somebody with strong hardware & VHDL knowledge could dev that... Nobody else !

Actually with SINGLE transfers , it si not sure all is OK... because USB and Ethernet maybe not work well , but for me the hardware is OK (I spent hundreds hours to scope the CPU and PCI bus to find a bug I never found with some Didier transfers routs he did for me ! Didier always claimed it was not his fault , but I still he had some bugs to find... I really do not think somebody will find something inside code from Didier because no sources available and PCI is complex !

So in case there is a brave soul out there... https://mikrosk.github.io/ct60tos/ctpci/english/dev.htm :) I have cleaned it up a bit, removed redundant files and sensitive information like invoices and made that crappy website (English version only, sorry ;)).

[marss]

Great

[TheNameOfTheGame]

Here's hoping someone will take up the reins.

[Kroll]

Great, I am a big fan of CTPCI and a user for ten years

[Cyprian]

great news

[ragnar76]

WOW!!! Is it christmas already? Thank you so much Rodolphe and mikro!

[tksm]

CTPCI is a great device, if someone improves the software and adds new functionalities, it will be a real revelation.

[wongck]

Great that the CTPCI stuff are all available.... Thanks both Czuba and Mikro

but to get someone who can improve it will be very difficult.
I was hoping that Czuba would some day continue.... may be during winter when he has less sunshine to do the solarpanel thing.
But I already know from his emails years ago that he has given up.... chasing the bug/ghost as he puts it.
Nevertheless.... Thanks Czuba for the great work. I am still using CTPCi with graphic/network and USB (little bit) as they all works !!!

Also if it is not for the CTPCI that gave me the large high colour video estate and networking.... I would not have as such fun programming the various nonsensical software that I have done....

[rmahlert]

AWESOME!
Good work mikro!

[viking272]

Brilliant news, nice one Mikro and Rodophe!

[Badwolf]

Whilst I doubt I'll ever need to even look at this, I'd also like to thank Rodolphe and Miro for making this happen.

The more information that's out there the absolute better, in my opinion!

BW

[Ektus]

THANK YOU!

[foft]

This is very exciting, thanks Rodolple for releasing the files and thanks @mikro for sorting/posting them etc etc.

I've sent an order for a few boards from pcbway. Lets see if they work... They asked for some of the files to be adjusted (mirroring/offsets), if the pcbs work ok then I'll post the adjusted ones to be added here.

[Estrayk]

GREAT news !

[Perdrix24]

Thank you.

[viking272]

This is very exciting, thanks Rodolple for releasing the files and thanks @mikro for sorting/posting them etc etc.

I've sent an order for a few boards from pcbway. Lets see if they work...

Great initiative foft, let us know how it goes

[mikro]

For those who do not watch Exxos forum: https://exxosforum.co.uk/forum/viewtopi ... =29&t=6171

foft really did it, a new CTPCI built & booting after ~12 years! (however don't get too excited, except Radeon its usability remains questionable as it contains all the hardware bugs as the original).

[wongck]

Well works at least for mine, the original CTPCI for VGA, network and USB- but since my USB card is very shaky when plugged in, I do not use it as it is scary to plug/unplug a usb device. So I use only networking & VGA.

[alexh]

Should anyone serously think about picking this up. Get in touch and I will help convert from ABL to SystemVerilog (or VHDL if you prefer).

I don't have the time to go at this full on (and I only have ct60e which I'm not sure supports CTPCI?) but I can provide 22 years of HDL design experience with FPGAs and ASICs. PCI was before my time but what I do now (PCIe) has it's roots in PCI, how hard can it be?

[mikro]

CT60e's issue with CTPCI is that it is not very friendly to CTPCI's cabling as it is supposed to go under the CT60/63 but CT60e is too wide.

The saddest thing in all of this is that even if we team up and somehow manage to debug the CTPCI DMA issues (see https://exxosforum.co.uk/forum/viewtopi ... 065#p98065) we can't fix / extend the source code because Didier kept the 2.01 TOS part closed (in comparison to 1.01 Driver part).

There are is some (FireTOS) source code floating around but I don't have it.

[Kroll]

I don't have the time to go at this full on (and I only have ct60e which I'm not sure supports CTPCI?) but I can provide 22 years of HDL design experience with FPGAs and ASICs. PCI was before my time but what I do now (PCIe) has it's roots in PCI, how hard can it be?

It can be run, I do (I only use it with a Radeon) and my colleague uses it not only with a graphics card, but also with a network or USB.
From what I remember, a colleague did it for me, that for full compatibility it is necessary to replace the Flash in the CT60e with the one originally used in the CT60/3 by Rodolphe.
I applied a slight increase via goldpins, and still managed to close the Falcon case.

[alexh]

The saddest thing in all of this is that even if we team up and somehow manage to debug the CTPCI DMA issues (see https://exxosforum.co.uk/forum/viewtopi ... 065#p98065) we can't fix / extend the source code because Didier kept the 2.01 TOS part closed (in comparison to 1.01 Driver part).

I imagine that it is not particularly complicated? It is after all a low-level driver. It should be possible to use Ghidra to disassemble to 68k, work out the functionality, label functions, recode in C. If there are enough variants it might be possible to see the differences, understand the evolution.

For example recently Ghidra was used to disassemble Ambermoon for the Amiga. The entire game has been, disassembled, functions given meaniningful labels and parameter names, it is now almost as editable as the original code. (And it re-assembles back to 1:1 the original binary)

[foft]

Should anyone serously think about picking this up. Get in touch and I will help convert from ABL to SystemVerilog (or VHDL if you prefer).

I don't have the time to go at this full on (and I only have ct60e which I'm not sure supports CTPCI?) but I can provide 22 years of HDL design experience with FPGAs and ASICs. PCI was before my time but what I do now (PCIe) has it's roots in PCI, how hard can it be?

It seems to me that we should target:
CT63 with fpga replacing both cplds. With super videl included in the fpga and a few pins used for a pci express expansion. Perhaps with a pci express bridge chip to simplicity the fpga logic.

[mikro]

I imagine that it is not particularly complicated? It is after all a low-level driver. It should be possible to use Ghidra to disassemble to 68k, work out the functionality, label functions, recode in C. If there are enough variants it might be possible to see the differences, understand the evolution.

Of course, nothing is impossible. I even have a rough idea how it's structured from previous versions. The 'heretic' question would be whether this is all worth it - for graphics we have SuperVidel (I know, I know but the authors are here and responsive so it's always possible to license it or release to the public); for USB we have Ethernat, NetUSBee, Lightning; for networking we have NetUSBee, Svethlana (included in SuperVidel), Ethernec and all the others.

So I don't know... is it worth the effort except 'we can' ? Don't get me wrong, I like CTPCI, my towered Falcon is still ready for action again (I had CTPCI for about 3 years in there) but seeing the amount of work...

Anyway, if you are serious about it, if @foft makes CTPCI available to you, I'll gladly borrow you my CT6x and Radeon for development / debugging and if you find something wrong on the software side, I'll try to fix it. ;)

[tksm]

CT60E PLUS CTPCI.
In CT60E You must replace flash for full CTPCI Functionality, with old flash only Radeon is working.
In one of pictures below is proper flash chip.