It's a great deal cheaper, if your chip is not too complex, to make it on an older technology. I think BDD or Garth has said what WDC's current technology is - it's possible that their licensed designs are on other technologies, as I think they license HDL these days.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

I'm always mindful of a quote I heard a while back: newer isn't necessarily better.

As you point out, the "new and improved" sales pitch is most often just that: a sales pitch. None of the above devices are new. Yes, they do get a die shrink once in a while, and hence the clock gets ramped up a bit. But the end result is only an incremental improvement. Of course, when the sales and marketing folks get done, they make it sound as though the slight die shrink and slight clock increase is revolutionary.

The interesting thing about the 65xx ISA, especially the 65C816, is that the performance has stayed consistently high for the die size and clock rate. That 200 MHz 65C02-on-an-FPGA that Garth mentioned has enormous throughput, despite being based on a 35 year old design. The ARM, MIPS and some others sound as though they are really fast when you read the manufacturers' descriptions. However, their throughput per clock cycle is actually no better than the 65C02/65C816 and in some cases, actually worse.

By the way, I first heard the "newer isn't necessarily better" adage 47 years ago when I was in the U.S. Navy. The ship I was on was engaging in gunfire support exercises on targets that were some 11 miles away. Our five inch guns were scoring better than 90 percent accuracy at this range, this despite being on a rolling and pitching destroyer at sea. Our fire control system consisted of radar, stereoscopic gun sights, gyroscopes and an analog ballistic computer, all technology that was mostly developed prior to World War II.

Also in our squadron was a relatively new ship fitted with the latest gadgets and firing on shore targets at the same range, with the newer 5 inch 54 caliber guns (ours were 5 inch 38 caliber). Their accuracy was around 75 percent, this despite the ship being some 20 years newer than ours. When I expressed surprise about this, the captain himself, who had seen combat during the Korean War, said, and I recall his exact words, "Son, newer isn't necessarily better. If it were, ships like ours would have been scrapped long ago."

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

I decided to use the sequence WDM, XCE to switch to 32 bit mode.


For a while they had some of the home Pc/business market, which is amazing for a product designed as a micro-controller. But as you say it's not really the same market.
Different markets are probably best served by different companies.

Service is important when the market is mature, and there aren't any real new "better" products. I think industry has a ways to go yet.

There is nothing stopping the hobbyist league from developing 65xx extensions. There are several "out there" and several in the works; they are not likely to be viable products though.

I currently have my RTF65002 running at 25 MHZ / 3CPI on an Atlys board; it's probably about as powerful as a 386 machine.

_________________
http://www.finitron.ca

Is there really still *hundreds of millions* (=10^8) of devices produced each year containing 6502 ?!
I'd be curious to know what those are.


It's not too complex nowadays to design a custom processor in HDL.


Yes, I won't deny any of that. But capitalism works in a way that relies on people always buying newer and better stuff (even if this is illusionary) to work. If you just go say your shareholders and customers that newer isn't better and your antique stuff is the best, then they won't belie you even if it's true. However, if you invest into each new stupid fad that will obviously only last a couple of years, they will respect you. That's how the world works, sadly.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

That makes sense to me. Or, you could do WDM -- CLC to go to 32 bits and WDM -- SEC to go to 16 bits.


Also, sometimes the market doesn't require a "new and improved" product. Frying skillets are largely as they were 30, 40 or 50 years ago. In fact, the only real improvements since the days of grandma's cast iron skillet have been Teflon coatings and hard anodized aluminum. Or, other than reduced water flow, toilets have changed little over the decades.

My point is that the market that WDC serves with the 65xx product line doesn't demand "new and improved." Yes, if WDC further shrinks the 65C02 or 65C816 die and makes it capable of running faster, the newer version will supplant the older, if it's at the same or lower price and the end user desires higher performance. However, there's really no demand for them to do so, so it's not likely to happen. Just how fast does that microcontroller in your Crock Pot® have to go, anyhow?

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

But I was under the impression that PIC, AVR and various 80C51 clones were sharing this kind of market. Also MSP430 series are specialized for battery-powered application where extremely low power consumption is a must. Sure there can be a few 65C02s, but aren't they really minority compared to those I just mentioned ?


Man this is the best argument ever. If a boss says I have to develop a new product, I'll say, "no, the man who invented the toilet did it perfectly on his first try, we should do the same".

Nice try, but Wikipedia tells us that Sir Thomas (19th century) made a rather impressive improvement to the work of Sir John (16th century). Still, 30 decades prior and still double digits since isn't bad.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

The present page is at http://www.westerndesigncenter.com/wdc/about_us.cfm - it's had similar statements on it since 2005, although they are now more vague.
2014: 6 billion sold, hundreds of millions per year, and multiple licensees
2005: over 5 billion sold, more than 200 million per year, and over 60 licensees.
2000: over 2 billion sold

More interesting perhaps, there's a table of core sizes in various technologies, and a timeline of technologies used by WDC, at https://web.archive.org/web/20001212060 ... nsing.html
"The W65C02 Hard Core has been manufactured in 3um (1981), 2um (1984), 1.5um (1986), 1.2um (1989), 0.8um (1992), 0.6um (1996), 0.5um (1997), and 0.35um technology (1998)."

I agree, I am a fan of it's instruction set, although this is purely down to personal preference.


That doesn't make any sense. Any Turing complete CPU can do anything provided there is enough memory.
If by "cannot do" you mean some PICs lacks memory to do things, then this can be solved by having a bigger PIC variant.


I'm no expert, but I'm pretty sure PIC, AVRs and various 80C51 clones can be licenced for being integreated as parts of ASICs as well. 80C51 is even in public domain I heard from a teacher - don't know at which point it's true.

One of the big segments for 6502 was in toys, and one of those was the little LCD photo album keyrings. They sold for a few dollars, so the chip with the 6502 core couldn't have cost more than a few cents. I can see why a chip manufacturer looking for minimal engineering cost would license a known and trusted and supported core from WDC or a partner instead of picking up an open source one. To make money at that end of the market you can't have many full time HDL experts on your staff: possibly as few as zero. The last thing you need is to make a chip and then find something amiss with the core. It's even likely that the application would be programmed in C - not because C is a good fit, but because it's cheaper to get a result in C. (See HP's 6502-based calculators as an example: not a good product as it turns out, but that's another story.)

None of this is to say that 6502 is an obvious choice - the availability of a reference design, and price, are likely to be deciding factors.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

PICs lack of addressing modes is partly due to its RISC origins. The 18F (high-performance) family has more addressing options than the base and mid-range families, while the new enchanced mid-range family offers a 'linear view' of the entire data and program address space if you need it.

The 16-bit 24F/30F/33F family has 16 16-bit of registers which tends to offset the lack of modes and the MIPS based PIC32 devices give you an almost pure load/store RISC archecture (with memory read delay!! You can't use a value read from memory in the next instruction) and no flag registers!. (In my PIC32 emulators I resorted to reading the whole of the target instruction in one 32-bit chunk and extracting the opcode and operands from it rather then trying to reading them individually byte by byte)

The trick with PICs and other microcontrollers is picking the right one for the job. The right perpherals, the right CPU and right amount of RAM and ROM for the job.

_________________
Andrew Jacobs
6502 & PIC Stuff - http://www.obelisk.me.uk/
Cross-Platform 6502/65C02/65816 Macro Assembler - http://www.obelisk.me.uk/dev65/
Open Source Projects - https://github.com/andrew-jacobs