65c02 clock speed (WDC's version)

[piemmm]

Hi!

Has anyone run a QFP wdc 65c02 at 20MHz? - looking at the datasheets(and for supporting chips), this would seem to 'imply' that it's possible (seems to stop at about 19MHz in the sheet).

My bus length is about 3cm, so this is quite interesting after looking at the datasheet - I remember on here reading something a good while ago on here that it should run fine at 16MHz in the right conditions, it's also amusing that it would technically put it ahead of an ARM2 (instructions per cycle wise, though the old ARM2 has more registers to do 'stuff' with so I suppose it will always beat the 6502 in 'getting stuff done' from that perspective)

73

[GARTHWILSON]

piemmm wrote:

Hi!

Has anyone run a QFP wdc 65c02 at 20MHz? - looking at the datasheets(and for supporting chips), this would seem to 'imply' that it's possible (seems to stop at about 19MHz in the sheet).

My bus length is about 3cm, so this is quite interesting after looking at the datasheet - I remember on here reading something a good while ago on here that it should run fine at 16MHz in the right conditions, it's also amusing that it would technically put it ahead of an ARM2 (instructions per cycle wise, though the old ARM2 has more registers to do 'stuff' with so I suppose it will always beat the 6502 in 'getting stuff done' from that perspective)

73

Windfall got even a 65816 going at 24MHz even at 3.3V. That's three times what the data sheet guarantees.
viewtopic.php?p=50721#p50721
viewtopic.php?p=51696#p51696
(Now I wonder what it could do at 5V!)

[Chromatix]

Well, the ARM2 not only has more registers, but each register and the ALU are much wider, and the arithmetic instructions are much more powerful especially if you make good use of predication. For most applications, you'd have to crank your 6502 up an awful long way before it caught up with even the earliest ARMs.

But yes, I've heard of someone here running a W65C02S at 20MHz, though I'm not sure which package was involved. WDC's specs are pretty conservative.

There's also the SuperCPU for the C64, which includes a W65C816S running at 20MHz. I expect it has to deal with a lot of wait-states when talking to the original C64 hardware.

[Dr Jefyll]

piemmm wrote:

I remember on here reading something a good while ago on here that it should run fine at 16MHz in the right conditions

As Garth noted, Windfall reports achieving 24 Mhz. Of course the conditions are important, as you say. His machine is a Co-Processor for the Acorn BBC Master, and its glue logic is perhaps substantially simpler than that in a more conventional system.

-- Jeff

[jds]

Rockwell's 65C02 based microcontrollers ended up running at 35MHz in their final standalone versions. After that everything was integrated into a single chip modem so they're not that useful for other tasks, but for quite a while they had a two-chip modem chipset where one chip was a standard microcontroller and therefore could be used for other uses. They are not in production any more but you can still find some on eBay. Documentation is good for some parts, and impossible to find for others, but they all appear to be quite similar in implementation. The 35MHz parts are the L2800 series which are a bit hard to find, but if you did want a fast 65C02 this could be an option. They have a multiply instruction too.

[drogon]

piemmm wrote:

Hi!

Has anyone run a QFP wdc 65c02 at 20MHz? - looking at the datasheets(and for supporting chips), this would seem to 'imply' that it's possible (seems to stop at about 19MHz in the sheet).

My bus length is about 3cm, so this is quite interesting after looking at the datasheet - I remember on here reading something a good while ago on here that it should run fine at 16MHz in the right conditions, it's also amusing that it would technically put it ahead of an ARM2 (instructions per cycle wise, though the old ARM2 has more registers to do 'stuff' with so I suppose it will always beat the 6502 in 'getting stuff done' from that perspective)

73

Not quite as fast as you're after, but I'm running both 6502 and 65816 at 16Mhz (DIL versions) on my Ruby boards without doing anything too special. I've not considered even trying to go faster - mostly because "retro" and all that.

Ruby project at: https://projects.drogon.net/ruby/


-Gordon

[piemmm]

Thanks for the replies - I'm going to have to see how fast my new board will run at once it's built!

[BillO]

I have a rather simple design running at 19MHz using the DIP40 package. Bus line lengths assorted but between 10cm and 14cm, no voltage planes and hence no impedance control. Using a more compact chip and layout with better bus length consistency, a 4 layer board and impedance control I imagine you could scoot along much faster.