hello

how fast could a 6510 become if produced with/in todays technology?
1. would 1000 mhz be possible? or even more?

2. what's the most accelerated 6510 you've ever heard of?

thanks for your thoughts!

_________________
e2020

It would be limited by the memory speed, since a processor clock cycle and a bus cycle are the same thing; so if you put everything on-chip so you don't have to take the buses outboard, you could reach hundreds of MHz, and in fact one of WDC's customers does run a 6502 core at over 200MHz. The 6510, used on the Commodore 64, was a special case of 6502 which just had some I/O on it. Someone else can give details. It was never made in the CMOS version. If you want to get into 6502, I definitely recommend going with the CMOS version (65C02) for more instructions and addressing modes, NMOS 6502 bugs and quirks fixed, greater bus-driving capacity, etc.. The 65816 is a 6502 with 16-bit internal registers and instructions, and has a lot more instructions making it far more suitable for multitasking and more efficient at higher-level languages, and provides the signals for better supporting separate program and data chaches, etc.. [Edit, 5/15/14: I posted an article on simple methods of doing multitasking without a multitasking OS, at http://wilsonminesco.com/multitask/index.html.]

Edit, about a year later: I added the section to my website for the large look-up tables for super-fast, accurate, 16-bit math, in some cases making it nearly a thousand times as fast to look up a trig or log function for example than to actually calculate it, and all 16 bits will be correct. The EPROMs (which I can supply) can go on a board you build to plug into the back of the C64 on the right side. Imagine doing math on the C64 at the equivalent of one or two hundred MHz or more.

_________________
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?

for historic & idealistic reasons (c64) I'll stay with the 6510 ...

what would the dimensions of the die could be if produced today?

_________________
e2020

It is basically a 6502 with a serial connection. The 65812S can possibly reach 20 MHZ.

I think you could easily emulate the 6502 with a 16 bit 65812S.

I would like to have newly build 6510 run @ 1000 mhz and *more* !

is this possible?

_________________
e2020

With a high end FPGA, it might be possible to get close to that but I was told by Jeri that I don't know the power in those chips.

Have you seen the Gameduino? This is what is possible with FPGAs and the Natami won't have as many Sprites but it should surpass the Gameduino:

viewtopic.php?t=1774&start=0

http://www.kickstarter.com/projects/208 ... me-adapter

http://excamera.com/sphinx/gameduino/

Make sure you watch the video.

The 6502 had multiple clocks so trying to speed them all up at once may be a difficulty. I'm not a hardware person so I can't answer that. There really wasn't a lot of money spent on R&D until the purchase of Amiga and that was a costly venture which involved paying the banks back.

I also believe they slowed down the CPU to write to the screen. That is what I gathered from reading some of the messages here. What you need is a GPU to take over other taks.

It's certainly possible, but it costs... a custom silicon project is probably going to be six figures minimum. If you could make do with 500MHz then an FPGA project would be possible for four figures or maybe less. (This is complete guesswork)

You might have more fun for a lot less money by buying an HP-20b business calculator, which just happens to have a 30MHz ARM system-on-chip inside, with PCB pads available for several analog and digital I/Os, serial and JTAG ports. And there's a software development kit to download with circuit diagrams and so on.

In an odd twist, you can then load some GPL calculator software on it, and it's a calculator!

(The HP-30b has the same features but a better keyboard, at a higher price)

on http://en.wikipedia.org/wiki/Field-prog ... gate_array at the end of the artikle it says "1.5 GHz fabric speed" - so (you think) that would that be the hightest possible speed a (virtualised) 6502 could run?

I read once (some time ago) that fpga are slower than regular chips (with physical die etc) ...

_________________
e2020

Fabric speed is probably the fastest an individual flip-flop can toggle. In practice, you usually don't get close. The routing delays from one flip-flop to another are usually much bigger than the delays from the logic blocks/flip-flops themselves. Sometimes you can get higher, but that usually requires careful manual layout, and a very good design.

Also, with a 6502 design, you'll have to count on at least 10 levels of logic in the longest path. Even if you can get 1.5 GHz per level, the overall speed is already down to 150 MHz.

Good point, Achronix seem to have some very fast FPGA technology - if it's available then that will be cheaper than a custom project. (It might be rather expensive, and they might only be selling to major volume sales prospects. Somehow, I feel pessimistic. They're a startup, just got $45 million in funding and a new CEO. They won't particularly want to sell single parts. Having said which, XMOS are interested in the hobby market.)

There are no prices on that product. I can't find it listed at Digikey or Mouser. It doesn't look like it will be cheap and I don't know if it is a consumer product.

I asked others about it and here are their thoughts:

http://www.natami.net/knowledge.php?b=2&note=36544

If a chip ran that fast, how would it communicate over the system bus? Do you want the screen running at 500 MHZ?

thanks for asking there!

what would the dimension be (die etc) if build today?
how much 6502s could be place within 1 cm² ?

is there any way to calculate this with actual waver- production- data?

_________________
e2020

If you want to do something practical - if you're not just performing a thought experiment - you should be thinking in terms of FPGA implementations. (Unless you're able to raise serious money to run a custom silicon project)

ElEctric_EyE ran a survey of 6502 cores for FPGAs. You can compare the sizes of those with the sizes of FPGAs - which vary according to how much you want to pay, of course.