PIC32MZ interrupt speed

[BillDee]

Does anyone have an idea of whether the PIC32MZ (200 Mhz)
can process interrupts at 4 Mhz?
That's minimum. But what about 20 MHz?
With some "minimal (small-ish) amount of processing" for each interrupt sequence.

I want to fully emulate a CDP1802 microprocessor at chip-pin-signal level,
and I was thinking about running the XTAL clock pin
for the emulated chip as an interrupt of the PIC32MZ.

That way I could run it as a really fast 1802 (hopefully)
and also have it be plug-compatible so that it could be
used in any 1802 computer. (these mostly run at < 4 MHz)

I do need one to run VERY fast, at at least 20 MHz,
but that could be a mode -- "Run All Out".

If anyone has a suggestion for a better microcontroller to use,
PIC or other, I'm open to whatever.

TIA

 

[granddad]

Hi Bill , I started with 1802 in the 70's had some enjoyable coding ,had a 'DOS' running in 2k mem... actual instruction time was around 8 micro sec, 1802 @ 2Mhz, some opcodes are 2 machine cycles, PIC32 at 200mhz may be a bit over the top, PIC24 will do 70MIPS and easier to program in my opinion.
Edit , still have 1802 bible!

 

[Ian Rogers]

Maybe you could get some ideas from the ZX spectrum emulator... The pic32 emulates the Z80 and the ULA..

 

[Mike - K8LH]

Hi Bill.. Welcome to the Forum.

Interesting question. If it weren't for your enhanced speed requirement I'd simply suggest augmenting a "real" CDP1802 with a PIC, AVR, or other microcontroller which can provide the clock, reset, loader, RAM, and some memory mapped I/O. I did something like this using a PIC last year and it worked quite well...

Cheerful regards, Mike

 

[BillDee]

PIC32 at 200mhz may be a bit over the top, PIC24 will do 70MIPS and easier to program in my opinion.

Thanks. I was looking at that, but thought 70 MIPS would be too slow. Someone already did this with a PIC (PIC16F886 IIRC),
which I haven't completely looked at yet, that could run at 40 MIPS, and he said was pushing it to make it effectively run at
1.7 MHz "output". (to emulate the 1861 Pixie video chip, too) I think he was running it at either 7 or 14 MHz, and then
multiplying it by 4 using the PLL. (not completely sure why about that last)

It's the NMI that I'm also concerned about. Not only emulating the 1802 instruction set, but also the hardware signals,
all within an Interrupt Service Routine, at up to 4 MHz and ideally up to 20 MHz.

 

[BillDee]

Maybe you could get some ideas from the ZX spectrum emulator... The pic32 emulates the Z80 and the ULA..

Thanks. I've look at that one a little. Afaict, that one is not a "hardware-level" emulation. (chip-pin-signal)
"Just" a complete in-software emulation/simulation.

 

[BillDee]

I did something like this using a PIC last year and it worked quite well...

Thanks. Yes, I've looked at your stuff and would like to mess with that, as well.
But IIRC you don't have a website with the code and such available, right?
Just those nice photos that show how you did it hardware-wise?
Are those good enough to completely allow someone to make the hardware?
And what about the software?

 

[granddad]

Bill. a quick pencil to paper + calculator ... a 20Mhz 1802 will have a machine cycle ( PIC interrupt) of 400 ns , a PIC24 @ 70MIPS has instruction cycle of 14 ns , 400/14 = 28.57 PIC instructions.... not many although a good many 1802 instructions are simple bit manipulation , interrupt over head i guess is not many , the 1802 instruction 'look up ' may be difficult , what MIPS for a 32MZ ?

 

[BillDee]

{snip}

That's the stuff I have trouble with. And then I wonder if I'm doing it right.
I'm not even sure how to convert the 200 MHz the way you did.
But I guess it should be a little less than 3 times your number.

 

[granddad]

32MZ quote DMIPS , ( scratch head ) , actually a 16 bit PIC will be a sinch for the 1802s 16 x 16 bit data registers, you need 38 IO something like a 53 I/O dsPIC33EV256GM106 ( think right) will do 5v as well + 16k ram, put one on a breakout board and give it a whirl..

Edit could you write some 1802 instructions in PIC asm ....

 

[BillDee]

32MZ quote DMIPS , ( scratch head ) , actually a 16 bit PIC will be a sinch for the 1802s 16 x 16 bit data registers, you need 38 IO something like a 53 I/O dsPIC33EV256GM106 ( think right) will do 5v as well + 16k ram, put one on a breakout board and give it a whirl..

Edit could you write some 1802 instructions in PIC asm ....

Yeah.
It does, ideally, also need to be 5V. I don't want to mess with 3V <==> 5V interfacing.
They're cheap enough that I can breadboard one, hook up a freq gen to the NMI,
and write a routine to count how many instruction cycles can execute before
the next interrupt. That will give me absolute real-world values.

I'm also looking at GHz microcontrollers. Or even microprocessors.
It's just a matter of how much external logic you need for support / i/o / memory.

I need something fast enough that you could, for example, make a hardware
"debugger" with full data displays (LEDs) and be able to monitor a 4 MHz 1802's
databus and be able to monitor and interpret code execution and accumulator/register
values and such. And I want the program that does all that to also be 1802 code.
(instead of using the native microcontroller and its code to do it, which would obviously be easier)

 

[granddad]

The PIC 'V' series need a vcap 10uF tant low esr , MPLAB X (3.10) stopwatch will give number of instructions between break points. most 1802 instructions...as previous post.. have fetch and execute cycles. 140 Mhz PIC OSC... FRC+PLL follow the data sheet coding. But you knew all that... De-bug could use a PIC com port .
Edit
If I get time , may try test myself... see my post
www.electro-tech-online.com/threads/testing-the-waters-whos-interested.144055/page-2#post-1215268