Faster/wider CPU...

Nagging hardware related question? Post here!
User avatar
Dave
SandySuperQDave
Posts: 2420
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Faster/wider CPU...

Postby Dave » Wed Feb 15, 2012 10:02 pm

I think, sadly, this is the comment that broke the camel's back, and I am just going to quietly bow out of the thread in the safe knowledge that no matter what you do, someone will tell you you're wrong.


Brane2
Trump Card
Posts: 179
Joined: Fri Dec 30, 2011 5:42 pm
Location: Ljubljana, Slovenia

Re: Faster/wider CPU...

Postby Brane2 » Wed Feb 15, 2012 10:16 pm

I don't understand why my POV offends you.

I just offered _my_ perspective.

If you think it's wrong, well, you are always perfectly free to ignore it.

It is you that wanted response on the matter and it is you that opened new topic to that end.

I just offered my opinion.

Sulking is IMHO totally counterproductive. You can't force people to give you an answer you want to hear and there is no real benefit in insisting that any negative response to your question will offend you.

Any honest question implies uncertainty about the answer - you question stuff that you don't know yet, so the answer is unknown.

If you make a question and then force the answer, then question itself was fundamentally dishonest.

All this is IMHO, ofcourse.


On the journey of life I chose the psycho path...
Brane2
Trump Card
Posts: 179
Joined: Fri Dec 30, 2011 5:42 pm
Location: Ljubljana, Slovenia

Re: Faster/wider CPU...

Postby Brane2 » Thu Feb 16, 2012 3:31 am

OTOH...

With updated starting premises it seems that my basic FPGA/CPLD suggestions would work for you even with 74... logic chips. IOW, "your camel has lousy back"... :mrgreen:

IOW:

- you throw out ULA1 and CPU and all the original RAMs.

Get yourself 68SEC000, a couple of fast FLASH chips like 64Kx8 or even better, 128Kx8. Or more.
Get yourself a couple of fast SRAM chips. 8MBit ones are nice. If you don't want to buy them just use cache RAMs from ooold 486 and Pentium boards. But I recommend new ones, since they are big ( 1MB) and quite cheap. And really fast ( 15 ns or so)

Do the ULA1 with your 74HC chips. If you take away all tricks that ULA1 had to do because it has to work on slow 8 bit bus, your version should be trivial.

One pint of complication would be communication with the rest of the old system ( ULA2&SPC, expansion port).

With ULA2& SPC you have two basic choices:

1. You can make 8--16 bit mux/demux logic and use it for expansion port also.

2. You can throw away ULA2 and SPC and use PIC with SMALL amount of programming to recreate needed functions. Basically only really crucial thing is keybaord read ( rows and columns), everything else is not essential ( sound, COMports). With sound, you could easily miss 30% with timing and an octave or two and not many would notice and even less would care... :mrgreen:


That being said, you do realize that modern programmable logic does not really require programming ( as in code development) and that you can "program" it by drawing schematics with 74LS style elements, just like you would do with your totally discrete approach ? :)


On the journey of life I chose the psycho path...
Brane2
Trump Card
Posts: 179
Joined: Fri Dec 30, 2011 5:42 pm
Location: Ljubljana, Slovenia

Re: Faster/wider CPU...

Postby Brane2 » Thu Feb 16, 2012 6:52 am

I've just tried Schematic entry in Xilinx Webpack.

Works just like any other classic program for making PCBs.

And with quite rich libraries- they contain much more than what you'd usually get within 74xx series.

Making the circuit is matter of picking "elements" from library, placing them in schematics and "wiring" them together.

After that, instead of making the PCB as you usually would, you press the button to generate bitstream and then program the CPLD while in circuit.

I've checked the prices and availability also. Xilinx most vanilla, bog-down, cheap series XC9500Xl is exactly what you need. It is not that old, since it works on 3.3V, but it is 5V tolerant, so you could juice up the 68SEC000 to 5V with the rest working on 3.3V.

It goes to 288 macrocells and biggest SMD packing is PQFP160 with 116 I/O pins. Farnell doesn't have the price for that one, but it has listed more expensive version in BGA256- €21 per piece , for qty of 1.

If you are satisfied with 144 macrocell model, you can get it in TQFP144 for less than €8 for qty of one.

Even this small one is more than you really need. IIWY I'd use several pieces. I'd stuck smart decoder etc in one and ULA1 in another. And maybe leave place for some cool dsPIC ( they are fast and have plenty of periphery- all that for €2-3 a piece !) if some day you change your mind wrt to programming your own...


Is this thing you have with CPLDs something akin to e.g. Islam's aversity to pork- ie are you sure this is not religious thing ? :mrgreen:


On the journey of life I chose the psycho path...
User avatar
vanpeebles
Commissario Pebbli
Posts: 2152
Joined: Sat Nov 20, 2010 7:13 pm
Location: North East UK

Re: Faster/wider CPU...

Postby vanpeebles » Thu Feb 16, 2012 9:28 am

Maybe you should start a thread with your ideal setup and leave Dave with his setup which works for him? :)


Brane2
Trump Card
Posts: 179
Joined: Fri Dec 30, 2011 5:42 pm
Location: Ljubljana, Slovenia

Re: Faster/wider CPU...

Postby Brane2 » Thu Feb 16, 2012 9:48 am

To recap:

He said that he wanted solution that he can solder up from 74xx and not program.

I thought that his eyesight is poor so he is avoiding small-pitch housings.

BUt then he mentioned 68HC and obviously had no problem with using even EC or SEC, which are to my knowledge SMD, so that meant that SMD elements are no longer a limit.

Then he said that he ruled out programmable logic because of price. But its price is neglible and even lower than with all-discrete solution, so this was again, non-issue.

Then the argument was that he just won't do any programming. And again, since e.g. Xilinx Webpack has schematics and allows designing with it and it has quite rich library of elements ( and it's free of charge), this, again seems like optimal way according to his requests and limitations.

Then I thought that he might value pristine state of the machine and keep as close to the original as possible. BUt even there CPLD/FPGA wins. It is small and powerful so it can be physically smallest, with the rest intact.

Then he mentioned that he doesn't care that much for existing SERial ports and microdrives, so CPLD solution again increased its lead...

To summarize- this solution is cheap, functionally equivalent ( but much faster) than his all-TTL, requires practically nothing he wouldn't have to do when designing the PCB ( ie drawing schematics ) and it enables him to check many bonus points to boot.

One is for example that he has all he needs to implement by-the-way complete QBIDE, only this time with full 16-bit width !

- cheap
- FAST
- no programming required

It marks check in all three boxes.
Besides that, it is:

- immensely flexible. Many corrections that would require board redesign with TTL are doable within minutes here
- testing is simple and fast, so any ever-so-crazy test costs you nothing vesides 30 seconds for reprogram.
- propagation times are negligable, compared to layers of TTL chips, so this is much less of a worry

Am I missing some _rational_ downside ? :)
Last edited by Brane2 on Thu Feb 16, 2012 10:08 am, edited 1 time in total.


On the journey of life I chose the psycho path...
MemoryLaneComputing
ROM Dongle
Posts: 9
Joined: Sat Mar 19, 2011 10:51 pm

Re: Faster/wider CPU...

Postby MemoryLaneComputing » Thu Feb 16, 2012 10:01 am

Brane2,

No, you're not missing anything, except a little bit of courtesy and restraint.

Why not just cool it down a bit? Everyone can see that you are "The Man". This is an interesting thread but you are really making it uncomfortable to read with the way in which you respond to people. There is absolutely no need to be so argumentative.


Adrian


Brane2
Trump Card
Posts: 179
Joined: Fri Dec 30, 2011 5:42 pm
Location: Ljubljana, Slovenia

Re: Faster/wider CPU...

Postby Brane2 » Thu Feb 16, 2012 10:15 am

I am "not the man" nor is this some kind of aspiration for me.

I was just trying to score the solution that would meet requested needs.

At each point I requested feedback, so that I could understand where have I missed and try again.

I invested quite some "brain cycles" into this and I resent the view that I do it just for the fun of it.

Try getting decent, authoritative answer that demanded some headscratching e.g. from a lawyer and then spit on his work this way.

If you don't want my brainwork, fine. Use your own.


On the journey of life I chose the psycho path...
User avatar
vanpeebles
Commissario Pebbli
Posts: 2152
Joined: Sat Nov 20, 2010 7:13 pm
Location: North East UK

Re: Faster/wider CPU...

Postby vanpeebles » Thu Feb 16, 2012 10:42 am

Brane I think you really need to take a look at your posts and how you react to other posters. This a friendly, supportive community with a very easy going nature. Any QL project will be supported regardless of how it gets made or achieves it's goals.

Dave has his way of working and uses the tools/methods that are right for him, as we all do.

The forum has been going great lately so I don't want to get heavy handed.



Who is online

Users browsing this forum: No registered users and 1 guest