16-bit 6502 vs. ARM or MIPS?

[Arlet]

Somebody who's 50 years old may still have to work for another 20 years. That's a bit early to get stuck in a rut, especially if the rut itself is already 30-40 years old.

[satpro]

Arlet wrote:

Somebody who's 50 years old may still have to work for another 20 years. That's a bit early to get stuck in a rut, especially if the rut itself is already 30-40 years old.

True, and it always pays to learn new things no matter the age, but he might also be the guy who knows things about the technology no one else does because they haven't had the first-hand experience, especially in the case of something like 6502. If you're over 50 it's a sure bet you probably know a few obscure 6502 tricks or quirks in a way someone in their 30s or even 40s won't. Back in those days we literally coded by hand until decent tools came out a decade later. I didn't even own an assembler until my late 20s, and that was a type-in.

[GARTHWILSON]

jmp(FFFA) wrote:

Even among unbiased users (e.g. they don't work for Microsoft or one of its competitors) you will find that most people aren't yearning to switch to another operating system.

Again, there's a familiarity that keeps people there. But look how MS got there. Every time a court ruled against MS, MS would volunteer their "punishment" to be that they have to give a lot of free software to schools, which (by design) only made sure more people got familiar with MS first so they'd stay with it for life. It's fraudulent.

More on topic, our own Samuel Falvo, (kc5tja here on the forum, although he seldom checks in anymore), professional programmer and software engineer, who wrote the "Software Survivalism" and "Neo-Retro Computing" articles I linked to earlier, said here that he wanted to make his own PC with an '816 (even though he is quite expert at x86), because maintaining his Windows machines took too much time and emotional energy. (I think Linux has come of age since then.) If anyone here could do it, it's probably him. I've seen him type source code at 100wpm while talking about the background at the same time, and the code ran correctly on first try. His discussions here were often way above the heads of nearly all of us.

I have plenty more to say, but I'm very close to locking this topic.

Quote:

There may be a missing element here, which is age. An engineer in their 20s should be outward looking and flexible - it's far too young to get stuck in a rut. But an engineer in their 50s or 60s might be in a position to say "this is what I know, this is what I do, and these are the people and projects that can vouch for me" and choose not to upgrade their skillset. (Or, they might feel they can't take that bet, or they don't want to!)

So true. A young one is still getting to know himself, professionally and in every other way. Look at all the graduation speeches. They're all about the big world out there just waiting to be tackled by the young, inexperienced graduate who thinks he's Superman and is not yet aware of what things he will come to despise later on, or what things will defeat him. The boss may wish for designer employees; but at the end of the day he has to fit the job to the available people. Older ones who know more what they do or don't want to do, and are more established and have the security to take the "else," just might do that.

Edit, Oct 2022: Last year I had to make a special tester for in-house use. I used a 40-pin Microchip PIC16 microcontroller, using all the I/O pins and most of the peripheral modules, including the UART. I had a lot of trouble getting the UART going for the conditions we needed. Microchip has a lot of ap. notes, but none on this. Finally I got it going. Now I have code that I can drop into the next application that needs it (and I assume it's the same UART Microchip puts in higher-range microcontrollers), meaning there's experience to make the next application's development much faster. Similarly, I use particular analog ICs in our products that have been out for decades, because I've learned their nuances, things that make them perform better that don't come through in the data sheets. If I go to newer parts, I have to start over in that process.


In response to satpro: Agreed. It is my observation that the younger ones who have been handed all the tools that "do all that stuff for you" (and think it's only an advantage) don't have a good understanding of what goes on underneath, and it causes them, in time, to run into problems. In that first 6502 class in '82, there was a reason the teacher had us hand-assemble our code, even though the computer had a rudimentary assembler onboard. Also, as a person gains age and experience, what's of most interest on their resume is not their education or how young and invincible they are, but their past accomplishments, knowledge of the industry, connections, etc..

[Arlet]

Quote:

If you're over 50 it's a sure bet you probably know a few obscure 6502 tricks or quirks in a way someone in their 30s or even 40s won't

Probably, but if maintaining old 6502 code takes a considerable amount of time, it's probably a good move to switch to something more modern. And if maintenance only takes a little time, that means there's plenty left to learn something new

Just a few days ago, I was working on some code that needed a square root during a 1 kHz interrupt. Maybe some 6502 expert would come up with an incredibly clever way to avoid doing that. I figured that I was working with a fast ARM with built-in FPU, so I could just call the double precision floating point sqrt() function, and not waste any further effort on it.

[sark02]

BigEd wrote:

There may be a missing element here, which is age. An engineer in their 20s should be outward looking and flexible - it's far too young to get stuck in a rut. But an engineer in their 50s or 60s might be in a position to say "this is what I know, this is what I do, and these are the people and projects that can vouch for me" and choose not to upgrade their skillset. (Or, they might feel they can't take that bet, or they don't want to!)

I'm 47, which puts me (just) out of that range, but what I will say that a CPU is just a component of a design. I may get all misty eyed about the 6502, but that's pure nostalgia. It's from a time when everything was new and I was finally "getting it" in terms of assembly language. My first love, if you will.

I enjoy the opportunity to lean a new CPU architecture. It's a rare thing to get to do (and be paid for it), and sometimes I'll discover something myself (like the Apollo Guidance Computer - the specs of which are on the web... and it's fascinating). I've never met an assembly language programmer who can't quickly transfer onto a new architecture. Unlike the Java-script trained kids coming up through college, who don't seem to know anything, assembly programmers have a deep and fundamental (and practical) understanding of computer architecture. Registers, ALU, memory, interrupts. It's the foundation of all CPUs. They're all the same... really. Except the PIC.. It's the "Internet Troll" of CPU designs.

If this is your hobby then, as a fellow nerd, let me tell you that other CPUs are fun to discover. It's coming up to Christmas... put an ARM board under the tree and spend Christmas day ignoring your family and getting some LEDs to flash (that might not be the best advice). If this is your work, then I will say that there are different and easier ways to do things... and easier isn't necessarily bad.

This is a 6502 board. PHA; TYA; PHA; TXA; PHA forever!

[Arlet]

Quote:

Except the PIC.. It's the "Internet Troll" of CPU designs.

The PIC32 has a MIPS core that's pretty decent, although I can't understand why they didn't just get an ARM license instead.

[satpro]

Quote:

I've never met an assembly language programmer who can't quickly transfer onto a new architecture.

Right on!

Who says we can't learn new technology?
PHA:PHX:PHY

[jmp(FFFA)]

sark02 wrote:

Except the PIC.. It's the "Internet Troll" of CPU designs.

That's a strange statement to make. There are at least three different CPU architectures used in PIC chips -- the 8-bit, 16-bit, and 32-bit architectures. The 32-bit architecture is a 32-bit MIPS core (exactly which one depends on which version of the PIC32 you are talking about). Both of the other cores are RISC-based Harvard architectures, just like most other MCU families use. What makes you think any of these are "Internet Trolls?"

Edit: Fixed the incorrect quote attribution. Sorry BigEd...

[sark02]

Arlet wrote:

Quote:

Except the PIC.. It's the "Internet Troll" of CPU designs.

The PIC32 has a MIPS core that's pretty decent, although I can't understand why they didn't just get an ARM license instead.

The PIC32 doesn't count as it's a departure from "PIC Classic" - a 3rd party MIPS design. I suspect they chose MIPS due to both license cost (Imagination needs their MIPS to be relevant so I suspect they're almost giving them away), and to differentiate themselves in a crowded market.

[jmp(FFFA)]

Arlet wrote:

The PIC32 has a MIPS core that's pretty decent, although I can't understand why they didn't just get an ARM license instead.

I for one am glad they didn't. Nothing against ARM per se, but without competition the ARM wouldn't be half the architecture it is today. On the desktop, AMD keeps Intel honest (or at least they did in the past). I hope MIPS continues to fight the good fight with ARM so we can all benefit from it.

At 330 DMIPS(*) with an on-board FPU, the PIC32MZ series offers the Cortex M4F some strong competition if you need to run fast.

(*) According to Microchip -- I haven't verified this

[Arlet]

I just wish they would get their competition without me having to suffer trying to decipher MIPS assembly

[sark02]

jmp(FFFA) wrote:

Nothing against ARM per se, but without competition the ARM wouldn't be half the architecture it is today.

I agree. I've worked on ARM and MIPS, and I find them fairly equivalent. I wrote elsewhere that I work on ASICs with embedded MIPS cores, and we continue to select MIPS over generations of ASIC as they continue to perform well.

[joe7]

I tend to view the 65816 (and 6502) in a learning context. The lack of hand-holding forces one to actually learn how a multiplication algorithm works, or how to do a signed shift, etc. You have to either fill in the missing blanks yourself, or research the matter. Both are valuable skills.

These days I spend most of my time in C++, but found that working with the 6502, 65816, and other "easy" processors helped me understand what is going on "under the hood". (At least from a conceptual standpoint.)

It also occurred to me that older processors are known-constants, which may be important where safety/predictability are paramount.

[BigDumbDinosaur]

joe7 wrote:

It also occurred to me that older processors are known-constants, which may be important where safety/predictability are paramount.

I understand that that was a significant factor in the development of 65C02-powered pacemakers and defibrillators. The 65C02's extremely lower power consumption when WAIting or SToPped would be attractive as well.

Newer technology isn't automatically better technology.

[GARTHWILSON]

In another topic, ChuckT just posted this link (below) to an audio interview from this last August with Bill Mensch on the 6502 and '816, and comparing them to ARM, 68000, x86, 6800, 6501, etc. and his business model, and his goals. He obviously has a very clear vision, and he's accomplishing what he wants. Still. Today.
http://ataripodcast.libsyn.com/antic-in ... -6502-chip
He figures that if they went to 20nm geometry for the 65c02 like the newest Intel processors use, and put the memory and I/O onboard, they should be able to run at 10GHz. [Edit, Oct 2022: Now they're down below 5nm! What would that do to the speed?]

In the interview with David Cramer (in March of this year), he says there are hundreds of different products being made today with 65xx processors in them.
http://ataripodcast.libsyn.com/antic-in ... ign-center