Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

for bbc micro/electron hardware, peripherals & programming issues (NOT emulators!)
User avatar
jgharston
Posts: 3212
Joined: Thu Sep 24, 2009 11:22 am
Location: Whitby/Sheffield
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by jgharston » Wed Dec 06, 2017 1:42 am

The data is put "temporarily" in registers in the NMI handler. The only source of any timing mismatches would be because the NMI handler is running in main RAM at &0D00 and the 4MHz board lets anything in main RAM run at 4MHz, so the devices at the other side of the Tube may not be able to cope with the data being flung at them twice as fast.

The NMI code is essentially:
BIT status:Bxx quit
PHA:LDA data:STA &FEE5:PLA
.quit
RTI

so the inter-byte delay is exactly as fast as the NMIs from the disk/network system. As long as the 4MHz board doesn't change the clock speed to the FDC or the ALDC the NMIs shouldn't be coming faster than "normal".

Hard drive access is polled, so the data is flung at the Tube at exactly the speed the hard drive choses to make data available to the CPU. This is an issue as the ADFS hard drive transfer loop is very tight and depends almost entirely on the instructions themselves for the inter-byte delay, so if the instructions execute faster they're going to go faster than a slow Tube device can cope with. There was a similar issue in the early Retroclinic RAMFS which stopped CP/M working as the code it was based on depended on the actual instructions to create the inter-byte delay, and the slight change to use the RAM disks shortened the loop.

Code: Select all

$ bbcbasic
PDP11 BBC BASIC IV Version 0.25
(C) Copyright J.G.Harston 1989,2005-2015
>_

User avatar
fordp
Posts: 963
Joined: Sun Feb 12, 2012 9:08 pm
Location: Kent, England
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by fordp » Wed Dec 06, 2017 8:16 am

jgharston wrote:
Hard drive access is polled, so the data is flung at the Tube at exactly the speed the hard drive chooses to make data available to the CPU. This is an issue as the ADFS hard drive transfer loop is very tight and depends almost entirely on the instructions themselves for the inter-byte delay, so if the instructions execute faster they're going to go faster than a slow Tube device can cope with. There was a similar issue in the early Retroclinic RAMFS which stopped CP/M working as the code it was based on depended on the actual instructions to create the inter-byte delay, and the slight change to use the RAM disks shortened the loop.
Thanks for the clarification. I think there was a special version of ADFS for the 32016 to slow it down. In any case I am not sure why the 4M board breaks the tube but I thought I would pass on what I had read in case it helped.

It may well be a low level signal integrity issue. I wonder if BeebFPGA could get 4M support. Another idea that may be interesting to try is driving one of the expansion connectors on BeebFPGA and using Dave's excellent 6502 logic decoding on there. So many ideas and so little time :(
FordP (Simon Ellwood)
Time is an illusion. Lunchtime, doubly so!

duikkie
Posts: 2885
Joined: Fri Feb 07, 2014 3:28 pm
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by duikkie » Thu Dec 07, 2017 8:24 am

oke shoot me , but i have a question , and i am building a board signal converter with old pcb board

i found a drawing of pitube direct signal converter v1.1 somewhere

now i see signal converter v1.3 with b7 ta3 what in v1.1 b7 is to gnd ?

is ta3 used ?

or can i leave it to gnd ?

https://easyeda.com/sundbyk/PiTubeDirect-ske82kCTf

User avatar
hoglet
Posts: 7502
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by hoglet » Thu Dec 07, 2017 8:40 am

duikkie wrote: is ta3 used ?

or can i leave it to gnd ?
To allow for future developments, A3 should be connected as well.

It is not used in the current releases (e.g. Diamondback-RC2).

(but it is used in the older Atom VDU version that Roland is using).

Dave

duikkie
Posts: 2885
Joined: Fri Feb 07, 2014 3:28 pm
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by duikkie » Thu Dec 07, 2017 8:57 am

oke it is a test pcb board what i use first looking if pi2 is working on beeb b. so i soldered b7 to pin 10 and b7 to pin 19 (green wire) and so on so i can chance it in the future :)
hoglet wrote:
duikkie wrote: is ta3 used ?

or can i leave it to gnd ?
To allow for future developments, A3 should be connected as well.

It is not used in the current releases (e.g. Diamondback-RC2).

(but it is used in the older Atom VDU version that Roland is using).

Dave
Attachments
DSCF5115.JPG
DSCF5114.JPG

User avatar
hoglet
Posts: 7502
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by hoglet » Thu Dec 07, 2017 10:28 am

You'll need to add some decoupling capacitors, ideally two per level shifter (one on VCCA and one of VCCB).

Dave

User avatar
KenLowe
Posts: 365
Joined: Mon Oct 18, 2004 4:35 pm
Location: Scotland
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by KenLowe » Sat Dec 09, 2017 8:22 pm

The github home page indicates that the ARMnative CoPro should run at 1000Mhz, but the latest Diamondback RC2 reports it running at 700Mhz on my system. Is this correct?

Edit: Ah. I'm thinking this might be because I'm running it on a RPiZero?

User avatar
hoglet
Posts: 7502
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by hoglet » Sat Dec 09, 2017 8:33 pm

KenLowe wrote:The github home page indicates that the ARMnative CoPro should run at 1000Mhz, but the latest Diamondback RC2 reports it running at 700Mhz on my system. Is this correct?
It should run at the default speed of the Pi.

What Pi model do you have?

You can override this in the config.txt file by setting arm_freq:
https://github.com/hoglet67/PiTubeDirec ... ig.txt#L39

Dave

User avatar
KenLowe
Posts: 365
Joined: Mon Oct 18, 2004 4:35 pm
Location: Scotland
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by KenLowe » Sat Dec 09, 2017 8:38 pm

I update my post to show it was a Pi Zero. I guess I did this as you were responding. I need to stop editing my posts!

User avatar
hoglet
Posts: 7502
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by hoglet » Sat Dec 09, 2017 8:41 pm

KenLowe wrote:I update my post to show it was a Pi Zero. I guess I did this as you were responding. I need to stop editing my posts!
I'll check my Pi Zero when Strictly is finished!

dp11
Posts: 830
Joined: Sun Aug 12, 2012 8:47 pm
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by dp11 » Sat Dec 09, 2017 8:54 pm

Do you have a Pizero or PizeroW ?

If you have a ZeroW then in the config.txt you will need to set the speed to 1000MHz

User avatar
KenLowe
Posts: 365
Joined: Mon Oct 18, 2004 4:35 pm
Location: Scotland
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by KenLowe » Sat Dec 09, 2017 8:58 pm

dp11 wrote:Do you have a Pizero or PizeroW ?

If you have a ZeroW then in the config.txt you will need to set the speed to 1000MHz
I've got both Zero and ZeroW, but I'm not sure which RPi is in which machine! I'll need to check.

dp11
Posts: 830
Joined: Sun Aug 12, 2012 8:47 pm
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by dp11 » Sat Dec 09, 2017 9:01 pm

If you are just going to use pi zeros ( of either type) then you can just set the frequency to 1000MHz.

User avatar
KenLowe
Posts: 365
Joined: Mon Oct 18, 2004 4:35 pm
Location: Scotland
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by KenLowe » Sat Dec 09, 2017 9:39 pm

dp11 wrote:If you are just going to use pi zeros ( of either type) then you can just set the frequency to 1000MHz.
I'm a bit confused. I removed the # from the arm_freq line, and now the RPiZeroW just doesn't boot.

Code: Select all

# clock frequency override 
# all pis bar zero W default to the correct frequency uncomment for pi zero W
arm_freq=1000
I also tried changing 'arm_freq' to 'core_freq'. The RPiZeroW would boot with this setting, but was very unreliable. That was obviously the wrong thing to do! I'm currently back running at 700MHz with this line commented out.

Note that the RPiZero (non W version), works fine with the comment removed from this line. The ARMnative processor runs at 1000MHz with or without this line commented out.

sundbyk
Posts: 90
Joined: Thu Jun 16, 2016 10:03 am
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by sundbyk » Tue Dec 19, 2017 7:23 pm

There are some problems with snow effects on some BBC Master's using the internal PiTubeDirect Signal Converter for the Master.
The ones connected to the external tube works fine.
It seem like are different version of chips in the Master causing the problem, but this need to be investigated.
To see if you have this problem run the program below and see if you get the snow effect.

10 MODE 0
20 GCOL 0,129
30 CLG
40 GOTO 40
Will investigate and find a solution for this. There may be more that needs to be fixed.

All people that have bought the Master version and have this problem will get a new card.
Since this is a "hobby" I would just ask to get back the cost of shipping the replacement.
I estimate to have new cards ready some time in January for shipping.

Hope nobody gets any problems with their Masters after this.

Kjell S.
***************************
Kjell Sundby
Web: http://www.sundby.com
****************************

User avatar
marcusjambler
Posts: 424
Joined: Mon May 22, 2017 11:20 am
Location: Bradford
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by marcusjambler » Tue Dec 19, 2017 9:20 pm

Thanks Kjell

Please put me down for a new board.
No problem about the postage.

Marcus

User avatar
myelin
Posts: 443
Joined: Tue Apr 26, 2016 9:17 pm
Location: San Francisco, CA, USA
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by myelin » Wed Dec 20, 2017 8:24 am

Moving over from the open source logic analyzer thread, because this is more PiTubeDirect related now.

The board pictured at the other end of the link above connects a Raspberry Pi in parallel with the FX2 logic analyzer board, for convenience. The downside here is that the FX2 needs to see the CPU's data bus all the time, whereas the Pi needs to be able to drive it sometimes as well.

A question: what does PiTubeDirect's bus access timing look like around requests? I assume it tristates the data bus when /TUBE is inactive, but I imagine there's some sort of margin between /TUBE transitioning high and the Pi dropping the bus. Or does it also tristate during the low clock period?

AFAICT the code that handles the GPIOs is in tubevc.s; is this correct? It looks like it drives the bus as soon as it detects /TUBE=0 and RnW=1, and drops the bus as soon as the clock goes low. Is it safe to assume that PiTubeDirect will always behave this way? If so, it makes my life fairly easy... I just have the CPLD drop the bus when /TUBE=0 and RnW=1, and drive it at all other times.
SW/EE from New Zealand, now in Mountain View, CA, making BBC/Electron hardware projects for fun.
Most popular: fast serial port, FX2+PiTubeDirect Tube/Cartridge adapter, USB cart interface.

User avatar
BigEd
Posts: 2092
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by BigEd » Wed Dec 20, 2017 8:47 am

It might be helpful to distinguish PiTubeDirect, as a piece of software running on a Pi, from a level shifter, which connects the 3V Pi to the 5V Beeb.

That's because the usual and common design of level shifter has two parts: a control lines part which is (mostly?) unidirectional towards the Pi, and a databus part which is bidirectional and controlled from the Beeb side.

I think all the level shifter designs we've seen so far, until this latest round of observations about video snow on the Master, have set the direction of the level shifter on the databus using only the RnW line. And that works well in all cases except one, which is the internal (Turbo) connector in the Master. In that case, it would be better to set the direction according to both the clock phase and the RnW line.

So, there's the timing of the Pi's turnaround from reading the databus to writing it, which might be of interest, and then there's the timing of the level shifter's turnaround from reading the databus to writing it.

Dave has posted photos of scope traces which might help: see here
viewtopic.php?p=152451#p152451
but more importantly see here:
viewtopic.php?p=143194#p143194

(As a survey of level shifters: there's Dave's original one-off board, Dominic's passive shifter, John K's tube silencer, and then three(?) models of Kjell's shifter, each in several revisions. There's at least one other DIY solution, I think. Kjell's is probably the one most seen in the wild.)

User avatar
myelin
Posts: 443
Joined: Tue Apr 26, 2016 9:17 pm
Location: San Francisco, CA, USA
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by myelin » Wed Dec 20, 2017 9:18 am

BigEd wrote:It might be helpful to distinguish PiTubeDirect, as a piece of software running on a Pi, from a level shifter, which connects the 3V Pi to the 5V Beeb.
I should have explained my question a bit better :)

I've designed a board that is mainly intended to connect a cheap FX2 logic analyzer dev board to an Electron or Master cartridge slot, or a BBC Tube socket, but also functions as an adapter to allow a Pi running PiTubeDirect to connect to all of the above.

On the Electron, this means it uses a CPLD to generate the /TUBE signal, because the Electron doesn't have a Tube port of its own. On the BBC, it just level-shifts the provided /TUBE signal.

The Electron also has a screwy clock signal, so the CPLD cleans it up using the Electron's 16MHz clock, using a similar process to the AP5.

Because there's a CPLD, it can define fairly complex rules as to when it should drive the bus. I'm trying to figure out when the Pi running PiTubeDirect could possibly be trying to drive the bus, so I can get the CPLD to drop it then.
So, there's the timing of the Pi's turnaround from reading the databus to writing it, which might be of interest, and then there's the timing of the level shifter's turnaround from reading the databus to writing it.

Dave has posted photos of scope traces which might help: see here
viewtopic.php?p=152451#p152451
but more importantly see here:
viewtopic.php?p=143194#p143194
This is perfect, thank you! The first link answers my question very nicely. It looks like the Pi takes up to 150ns to respond to changes in the various signals, and is just barely done by the time /TUBE goes high again. So to be safe, I should have the CPLD drop the bus whenever /TUBE=0 and RnW=1, and not drive it again until the following rising clock edge. That should be absolutely safe. I've tested it and it works!
(As a survey of level shifters: there's Dave's original one-off board, Dominic's passive shifter, John K's tube silencer, and then three(?) models of Kjell's shifter, each in several revisions. There's at least one other DIY solution, I think. Kjell's is probably the one most seen in the wild.)
Mine may be the other DIY solution? I posted about it on this thread a few months back. It's an Electron cartridge that is basically a simpler version of the board I'm asking about today. I sent some to Dave Hitchins a while back, but I have a feeling the combo FX2 board is going to be more popular :)
SW/EE from New Zealand, now in Mountain View, CA, making BBC/Electron hardware projects for fun.
Most popular: fast serial port, FX2+PiTubeDirect Tube/Cartridge adapter, USB cart interface.

User avatar
BigEd
Posts: 2092
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by BigEd » Wed Dec 20, 2017 9:49 am

Excellent result - glad I could help (a little).

That's a very interesting multi-purpose board - worth a promotion in its own thread?

User avatar
BigEd
Posts: 2092
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by BigEd » Wed Dec 20, 2017 9:52 am

myelin wrote:It looks like the Pi takes up to 150ns to respond to changes in the various signals, and is just barely done by the time /TUBE goes high again. So to be safe, I should have the CPLD drop the bus whenever /TUBE=0 and RnW=1, and not drive it again until the following rising clock edge. That should be absolutely safe.
Just a thought - however you're making TUBE, beware of glitches, which is to say, false matches while the address lines settle from one value to another. Would that fail-safe in your case, or cause bother? Or can it be ruled out?

User avatar
myelin
Posts: 443
Joined: Tue Apr 26, 2016 9:17 pm
Location: San Francisco, CA, USA
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by myelin » Wed Dec 20, 2017 10:19 am

BigEd wrote:Excellent result - glad I could help (a little).

That's a very interesting multi-purpose board - worth a promotion in its own thread?
I've kind of hijacked Hoglet's thread already, on the assumption that all the discussion about FX2 logic analyzers belongs in the one place, but maybe let's create a new thread once my board is fully tested and safe for everyone to use...
BigEd wrote:
myelin wrote:It looks like the Pi takes up to 150ns to respond to changes in the various signals, and is just barely done by the time /TUBE goes high again. So to be safe, I should have the CPLD drop the bus whenever /TUBE=0 and RnW=1, and not drive it again until the following rising clock edge. That should be absolutely safe.
Just a thought - however you're making TUBE, beware of glitches, which is to say, false matches while the address lines settle from one value to another. Would that fail-safe in your case, or cause bother? Or can it be ruled out?
Good question. That'll cause a glitch in my /TUBE output, but the bus driving logic will fail safe; how I've implemented it is to look at /TUBE and RnW on the rising clock edge, and if /TUBE=0 and RnW=1, set a flag that tells us not to drive the bus for the next cycle, in addition to the usual conditions. So things would only get weird if /TUBE had a glitch right on a rising clock edge, which shouldn't happen.
SW/EE from New Zealand, now in Mountain View, CA, making BBC/Electron hardware projects for fun.
Most popular: fast serial port, FX2+PiTubeDirect Tube/Cartridge adapter, USB cart interface.

jregel
Posts: 92
Joined: Fri Dec 20, 2013 6:39 pm
Location: Gloucestershire
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by jregel » Thu Dec 28, 2017 10:50 pm

Thanks to Kjell, I am a happy member of the PiTubeDirect owners' club. A fantastic development, congratulations to all involved in creating this project!

Although I'm mostly interested in using it for the 6502 co-pro functionality, I'm intrigued by the native ARM mode. Looking through this thread (all 34 pages, so sorry if I missed the answer!), it looks like some people have got BASIC running on it. Is this ARM BASIC V, or an older BBC BASIC? Presumably the assembler is ARM?

Thanks!
BBC Master Turbo
Retroclinic External Datacentre
VideoNuLA
PiTubeDirect with Pi Zero

crj
Posts: 834
Joined: Thu May 02, 2013 4:58 pm
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by crj » Thu Dec 28, 2017 11:08 pm

Dumb question that's been bugging me for a while, by the way: how does native ARM mode work? I thought PiTubeDirect had to take the place of an OS so it could be locked into cache and thereby run consistently fast enough to perform bus emulation without any latency spikes ruining everything. That doesn't seem consistent with running other code natively.

Conversely, if it is possible to run other code natively, what's to stop a suitably tricked out Linux kernel from running on a PiTubeDirect machine?

User avatar
BigEd
Posts: 2092
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by BigEd » Thu Dec 28, 2017 11:12 pm

About max latency: see previously (upthread).

About the ARM Basics: yes they have ARM assemblers built in. For the native ARM, you need BAS135. I'm sure there's an SSD around somewhere nearby. (Edit: see here)

jregel
Posts: 92
Joined: Fri Dec 20, 2013 6:39 pm
Location: Gloucestershire
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by jregel » Thu Dec 28, 2017 11:15 pm

BigEd wrote:About max latency: see previously (upthread).

About the ARM Basics: yes they have ARM assemblers built in. For the native ARM, you need BAS135. I'm sure there's an SSD around somewhere nearby. (Edit: see here)
Thanks, BigEd!
BBC Master Turbo
Retroclinic External Datacentre
VideoNuLA
PiTubeDirect with Pi Zero

User avatar
hoglet
Posts: 7502
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by hoglet » Sat Dec 30, 2017 2:43 pm

Hello all,

At long last, here's the final Diamondback release, just in time for 2018!
https://github.com/hoglet67/PiTubeDirect/releases

Here's the full release notes, detailing all changes since Cobra:

Features:

- Implemented elk_mode configuration property (enables patching of 6502 code in Client ROMs for Elk Tube address)
- Updated 6809 Client ROM to 1.05 (8e9823ac)
- Added OPC5 Co Processor (in slot 5, experimental)
- Added OPC6 Co Processor (in slot 6, experimental)
- Added OPC7 Co Processor (in slot 7, experimental)
- Native ARM Co Pro: increased memory from 2MB to 16MB

Fixes:

- Fixed a GPU issue that might result in lost read messages if there is a glitch in a0
- Fixed a GPU issue that delayed the tube cycle following a nTube glitch causing random failures
- Fixed an issue with Mini UART initialization than sometimes meant receiving didn't work
- Tube ULA: make unused status bits read back as '1'
- Lib6502 Co Pro: Fix Bug in decimal flag in BRK instruction
- ARM Co Pro: Re-instate just the reset vector (word 0) on break
- Native ARM Co Pro: Restore probable missing else (command line args parsing)
- Native ARM Co Pro: fixed a hang with SWI &0C (GBPB) - thanks RobC
- Native ARM Co Pro: command parser less greedy
- Native ARM Co Pro: OS_Find, OS_File, OS_CLI now allow any ctrl char as string terminator
- Native ARM Co Pro: fixed bugs in copying command line into environment

For more details on the new OPC5/6/7 Co Processors, see: https://revaldinho.github.io/opc/

(Ed and Dominic, the current development branch is now egg-eater-dev, and I've just merged diamondback-dev -> master -> egg-eater-dev)

Dave
Last edited by hoglet on Sun Dec 31, 2017 10:39 am, edited 1 time in total.

User avatar
BigEd
Posts: 2092
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by BigEd » Sat Dec 30, 2017 2:58 pm

Well done Dave!

Revaldinho has been working on a BCPL compiler targeting the OPC6 and 7 - these are a 16 and a 32 bit version of a small simple CPU - and there are a couple of threads worth reading if you like this sort of thing: (Edit: now added to the OPC minisite.)

Edit: as background, the OPC6 and OPC7 are respectively a simple 16 bit and a simple 32 bit machine suitable to be targeted by a BCPL compiler. They are also quite fun to program in assembly, being somewhat 6502-like, somewhat RISCy and a little bit ARMish. In both cases they have 16 registers and are very regular. "OPC" stands for "one page computing" and refers to a goal of having the source code fit on one page of fanfold, a constraint which keeps the CPU simple. Up to OPC6 we imposed an additional constraint on ourselves, of fitting the implementation in 128 slices on Xilinx FPGA. For more on the evolution from OPC5 to OPC6, see here.

User avatar
BigEd
Posts: 2092
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by BigEd » Thu Jan 11, 2018 6:01 am

(I see PiTubeDirect has been posted on Hacker News (again))

sundbyk
Posts: 90
Joined: Thu Jun 16, 2016 10:03 am
Contact:

Re: Pi-based Co-Pro on the cheap - 100MHz 6502 for £10? (now 274MHz)

Post by sundbyk » Sat Jan 20, 2018 1:44 pm

For those of you with an internal Master level shifter who are experiencing video interference:

The new updated PiTubeDirect SignalConverter BBC Master version (1.4) in now ready to be shipped.

It have been test by a group of people in in Norway and UK and the interference is now gone.

If you have this problem and want your board(s) replaced I ask only for the cost of shipping.
The cost is £7 for the shipping to UK (Europe) for up to 10 board.
Please send it to kjell@sundby.com via PayPal with address, number of boards and I will ship as the boards get ready.

Thanks for the understanding.

Kjell S.
***************************
Kjell Sundby
Web: http://www.sundby.com
****************************

Post Reply