The 100 MHz 6502!

[RobertBe]

And the 65F02 works with a Commodore PET 8032! Think of it working with a VIC-20! See

http://www.e-basteln.de/computing/65f02 ... st-systems

Truly,
Robert Bernardo
Fresno Commodore User Group - http://www.dickestel.com/fcug.htm
Southern California Commodore & Amiga Network - http://www.portcommodore.com/sccan
Nov. 6-7 Commodore Los Angeles Super Show 2021 - http://www.portcommodore.com/class

[mathom]

Neat project! Sounds like the VIC would be a better candidate than the C64 because of the memory swapping limitation. Any of you hardware gurus see any show stoppers for the VIC?

[chysn]

It's fun to think about. It seems like you'd be able to do cool graphical things on the VIC-20 with such speed. But the best VIC-20 graphics achievements rely on the 6522s' timers, and at 100MHz the 16-bit timers are no longer wide enough to express a reasonable amount of time.

So what kind of 8-bit computing tasks would benefit from 100MHz? Speed without memory gets us what? Maybe some interesting sound creation possibilities?

It all makes me think of Robin Williams's Genie in Aladdin: "Phenomenal cosmic powers! Itty bitty living space."

[Mike]

Any of you hardware gurus see any show stoppers for the VIC?

Not strictly showstoppers, but there are some timing loops inside the KERNAL which would need to be adjusted.

[...] the best VIC-20 graphics achievements rely on the 6522s' timers, and at 100MHz the 16-bit timers are no longer wide enough to express a reasonable amount of time.

The VIAs are supposed to remain clocked at 1 MHz and the 65F02 has provisions to sync I/O accesses to that slower clock. However, any video effects that rely on cycle exact code after they've been triggered by a timer IRQ synchronised to a raster line will lose their timing at that point.

It should be noted that the internal RAM must be considered I/O for the purposes of the 65F02, as the VIC reads the video data from it. At least the write accesses to $1000..$1FFF need to be special cased (the 65F02 can do the reads at full speed). $0000..$03FF are also accessible by VIC, but slowing writes on zeropage and stack would throw away a lot of any prospective speed gain. We're at less than 100% compatibility anyhow, so excluding $0000..$03FF from VIC access is the lesser loss. However, VFLI modders might want to to add the $0400..$0FFF range, so $0400..$1FFF is fully regarded as VRAM.

So what kind of 8-bit computing tasks would benefit from 100MHz? Speed without memory gets us what?

Fast vector graphics and fractals spring to my mind here. Or improved strength of chess programs.

...

I am not affiliated with the project and just found out about it 2 days ago when a new thread mentioned it on 6502.org

[RobertBe]

(Juergen has given me permission to post his responses to me.)

Fri., Oct. 15

Many thanks for reaching out! Yes, I have been thinking about supporting the
VIC-20, which should be a more realistic proposition than the C64 (with its
6510 and memory banking trickery...)

But I will probably look into an upgraded version of the 65F02 first, most
likely based on the Spartan-7: As I add more host computers, I am running
into timing bottlenecks in the FPGA. The address decoder (which has to
decide when to route a bus access to the fast internal RAM or to the
external bus) sits on a timing-critical path, and starts to get in the way
of 100 MHz internal operation. I could also do with the larger RAM available
in the Spartan-7, to properly support bank-switched host designs.

Since the Spartan-7 pinout is quite different from the Spartan-6, and it
also needs different supply voltages (and one more), this means a complete
redesign of the PCB. So it will take a while, unfortunately. But looks like
a good winter project, so I hope to get it done eventually!

Sat. Oct. 16

To avoid creating false expectations, it is probably worth adding that the
chances of making the 65F02 available to a broader audience are
unfortunately pretty slim. I have explored options last year to have it
produced in quantities of 100 or so, and found that

(a) The numbers don't work out; the 65F02 would need to be more expensive
than people expect.
(b) Since I am based in Europe, the compliance hurdles are a pain in the
neck. (EMC measurements, anyone?)
(c) Managing the production and distribution would not be fun at all --
certainly not my idea of a hobby, and a job I already have...

Hence, while I very much appreciate the interest and the feedback, I don't
want to create expectations that this will be available on Tindie or
elsewhere any time soon. And on the other hand, the required BGA soldering
will unfortunately mean that only very few hobbyists could build their own
65F02.

Best regards,
Juergen

[chysn]