Re: Terrabites of storage

On Tue, 11 Apr 2006 22:49:38 +0100, Neil Barnes wrote:

On Mon, 10 Apr 2006 19:21:20 -0500, Ric Locke wrote:

The 6502 is an interesting chip. Most of the instructions have four or
five "T-states" per clock; the designer exploited delays in the
transistors to make it work. Race conditions are integral to its
functioning. So yes, a 1 MHz 6502 is roughly equivalent in what it can
do to a 4 MHz Z80.

Not disputing, Ric, but that surprises me. The multi-t-states, not the
speed equivalence; I used to match a 750kHz 6502 against a 4MHz z80 and
found them remarkably close in performance.

I had occasion to dig into the 6502 as a 'black box' design recently, and
found it reasonably simple to replicate all the states (and improve on one
or two) using straight forward single phase latching - data available on
one half of the clock, latch on the other half. Effectively one t-state
per clock.

Well, I'd call that two by Z80 or 8080 standards, but I get your point.

Even the arithmetic instructions? The model I saw depended on the delays in
the clock buffers to latch input-result-out-store, and it varied according
to the delays through the ALU. But maybe analysis techniques have improved.
I haven't looked at the subject in a couple of decades.

At the moment I'm using a lot of the AVR range but I'd be happy if they
implemented the 6502 capabilities at the same speed, as a single-chip with
flash memory :)

OMG yes. Better yet, a cross between a 6502 and a PIC16 -- on-chip
peripherals, a couple K of memory, and the aforementioned flash block. A
whole 256-byte block of indirect memory access instead of one measly
register, and real arithmetic instructions. Heaven.

I don't expect it. Last I heard the IP was tied up in the MOS Technology
debacle, which is another reason there aren't any 6502-derived chips
around.

I use a lot of PICs (well, comparatively) and Z80/Z180s. One of the reasons
I stay with the old Zilog is the interrupts. I never quite understood why
it was that programmers and chip designers both don't scream for good
interrupt structures; it's by far the easiest and most elegant way to do
I/O. PICs are cheap, and useful development tools for them are cheap or
free, so they end up going into stuff almost by default.

Regards,
Ric
[neep neep neep. no more, I promise]
*** Free account sponsored by SecureIX.com ***
*** Encrypt your Internet usage with a free VPN account from http://www.SecureIX.com ***
.

Relevant Pages

  • Re: Yet more A3000 questions.
    ... So, if I re-route 24/2 to the memory I get a 12mhz bus, right? ... Clksys* can be found on the genlock connector LK28 pin 2. ... MEMC splits this up into Phi1 and Phi2, a two-phase clock formed by Clksys ...
    (comp.sys.acorn.hardware)
  • Re: A7V333 questions
    ... Board is an early version 1.01 with bios updated to 1017. ... I also recently found a good deal on Crucial 1 gig PC 2700 memory, ... the PCI clock divider switches to 1/5th at ... PCI clock, and will be a safe value as well. ...
    (alt.comp.periphs.mainboard.asus)
  • Re: A7V333 questions
    ... Board is an early version 1.01 with bios updated to 1017. ... I also recently found a good deal on Crucial 1 gig PC 2700 memory, ... the PCI clock divider switches to 1/5th at ... PCI clock, and will be a safe value as well. ...
    (alt.comp.periphs.mainboard.asus)
  • Re: Question about enable on DRAM module.
    ... always @(clk1, master) ... Very often in memory subsystem ... Most asynch memories use the edge of en to latch the address and rwb, ... everything in one clock cycle. ...
    (comp.lang.verilog)
  • Re: ASUS GeForce 6600 (256 MB) question
    ... "VGA's voltage is out of safe bound. ... Original Clock 3000 MHz ... Memory Bus Properties: ... Bus Type Dual DDR SDRAM ...
    (alt.comp.periphs.mainboard.asus)