Finally: Broadband for the Commodore 64

GP writes "Now even die-hard Commodore 64 users are able to enjoy the benefits of broadband Internet connectivity. A newly announced Ethernet card together with the Contiki operating system lets you surf the web, send e-mail, host web sites with the built-in web server, and soon even play LAN games on your good old Commodore 64! All this with a computer that is old enough to drink."

Re:Uhmm..

[Thomas M Hughes]

It means the Commadore was released more then 21 years ago, which is that age at which its legal to drink alcohol in the US. I believe the original release was in 1982.

Re:My 486 DX/2 66mhz machine hardly push 200kbps

[magarity]

My 486 DX/2 66mhz machine hardly push 200kbps

Then you haven't configured the thing properly. I used a 386DX-20 as a firewall for a cable modem for a couple of years and it had no problem with >1024kbps.

Re:Uhmm..

[Nos.]

Well, in some countries there is a minimum age to purchase alcohol. In the US, its 21. According to: http://oldcomputers.net/c64.html [oldcomputers.net] the commodore was released Jan, 1982. Making it 21 years old, or, old enough to drink.

Announcement, when Slashdotted...

[Anonymous Coward]

(Anonymous non-karma w***e posting.)

13.09.2003: Retro Replay and RR-Net available
In time with the announced date, the new production run of the Retro Replay is finished. Compared to the old cartridges, only cosmetic changes have been made: The most significant change is the colour: Blue instead of black. After many requests by users, the jumpers are now mounted straight, not to the side. To ensure proper mounting in our new cases, the mounting hole has been moved and changed in diameter to perfectly fit the transparent cases.

At the same time, the networking card RR-Net is going on sale. The card is plugged to the expansion port of the Retro Replay, and allows connecting the C64 to an intranet. Although the operating system Contiki is freeware, we have an agreement with the author Adam Dunkels: He gets paid for every RR-Net unit that's sold. Contiki is an operating system that offers many features in very small space: A TCP/IP stack, a web browser, a webserver, a VNC-client and of course a graphical user interface. It is included on a 5,25 inch disk for the C64. To make use of all features of Contiki, an intranet with router should be available.
As an introductory offer, there's a network-bundle. It contains:

# Retro Replay
# RR-Net with Contiki
# transparent case
# worldwide shipment

together for only 100,- EUR!

Please use the contact form for your order. Unfortunately, our domain ami.ga does not work at the moment, because the republic of Gabon is currently migrating their internet connection from satellite to undersea cable. Even with our server in Germany, both the website and email addresses are affected, but the contact form works reliably!

Nothing all that new

[Anonymous Coward]

It's been done before: http://www.dunkels.com/adam/tfe/ [dunkels.com]

Re:But can it fill 10BaseT?

[AKAImBatman]

I think that question depends on whether the C64 had DMA or not. If yes, then it would have a chance in hell of doing burst transmissions that could flood a 10BaseT line. If no, then why the **** are people wasting time on this?

Lords of Conquest

[Morglum]

In a shocking fit of synchronicity, I stumbled upon a java aplet version of Lords of Conquest, and have been playing for the last hour...an hour before this story was posted...

Ah, the good ol days. Who said you could never go back?

Re:watch out

[ichimunki]

Forget that! I have code (if you read machine language, that is) for an update which allows for 80 columns. Oh, and GEOS, which is the One True GUI. ;)

Re:But can it fill 10BaseT?

[FrostedWheat]

I think that question depends on whether the C64 had DMA or not.

Kinda. All the chips (video, sound, etc.) had direct access to the memory. But they all have to take turns, when the video chip was reading from memory, the CPU couldn't and would pause. Some of the turbo-loaders (heh, load 64k in 15 seconds from floppy .. wheeeee) would blank the screen to make sure nothing interrupted the CPU and upset the timing.

Re:watch out

[mistered]

Yep, here's a screenshot [www.sics.se] of the C64 showing slashdot in a browser.

Re:New kind of bottle neck

[Oliver Wendell Jones]

Point of order, Mr. Chairman...

Modems measure speed in Kilobits, computer RAM is measured in KiloBytes, so a computer with 64K of RAM contains ~64,000 Bytes or 512,000 bits.

A 56K modem downloads (at a theoretical maximum) 56,000 bits per second.

Simple math reveals that with 512,000 bits of RAM, at 56,000 bps it will take 9.1 seconds to fill it's RAM capacity.

Re:useless

[Rkane]

Ah yes, it was the guy at humanclock.com [humanclock.com] that said he was using a tandy trs 80 as a webserver [humanclock.com]

Re:But can it fill 10BaseT?

[ewhac]

So...can it [flood a 10BaseT network]?

No. It can't.

If not, how much traffic do I have to send it to bring it to a crawl? :-)

Not much, I would think.

The C64 has a 1MHz 6510 8-bit CPU. The memory bus is also 1MHz. Moreover, the fastest instruction on the 6510 (which is a 6502 derivative) is two clocks. Thus, at four clocks per byte (two to read, two to write), the fastest data transfer rate you could conceivably get is 0.25 MBytes/second (in reality, it would be rather slower as the LDA and STA instructions take more than two clocks, but I don't have the timing chart in front of me).

The C64 does have DMA, but it's dedicated to video access and refresh and can't be redirected. Moreover, these DMA cycles completely take over the bus for 40 clocks every eight video lines. So your packet writes will likely hiccup from time to time. (Presumably they have big silos on the NIC.)

Even if the NIC did DMA itself, it would have to get out of video's way every eight lines, which means you couldn't flood the line indefinitely. Also, the C-64 has a mere -- surprise! -- 64K of RAM. At 1MByte/sec, you'd run out of RAM in 0.065536 seconds.

Schwab
C-64 Early Adopter

Re:My 486 DX/2 66mhz machine hardly push 200kbps

[DemoLiter3]

Most home computers of that era, like C64 or ZX Spectrum wrote their data on tape with ca 1500-2000 bit/s by default. Many games (originally packaged) used their own loading routines with speeds up to 4000 bps in order to load faster. You could go higher over 6000 bps, but the risk of getting read errors was too high.

The important thing is, the read/write routines were CPU-timed, they used short loops to precisely time the moment to switch between 1 and 0 on output. These loop have a natural precision limit of 1 loop, which would be x CPU cycles (7 on Z80??? or 10? I don't remember correctly). You could reach maybe 100000 bps (theoretically) with this principle, but practically you'd have too many errors.

Re:Liquid refreshment

[Ophidian P. Jones]

It means that, in order to run at a decent speed, you have to overclock the C64's 6502 [snip]

The C64 actually ran on a 6510. The disk drives were 6502.

Also notable device with 6502 CPU: original Nintendo.

Re:C=64 sucks, ATARI RULES!!!!!!!

[Wraithlyn]

C64's didn't use standard ASCII, they used "PETSCII" which had an ecleptic collection including everything from cursor control codes, colour change codes, function key codes, and a large assortment of graphic symbols.

Weird but kinda neat... how many computers do you know where you can write an upward-slanting diagonal, multi-coloured string with a single PRINT statement? :)

This reminds me of the only single-line animation program I've ever seen... It was basically (no pun intended):

10 A$="{cu}{cd}{cl}{cr}":PRINT " ";MID$(A$,INT(RND(1)*4),1);"O{cl}";:GOTO 10

Where {cX} are cursor codes for up, down, left, and right.

Ahh.. those were the days. Reaching for the power switch was the longest part of bootup time, and nary a bit was wasted. BASIC interpreter in 8KB, DOS in 8KB, and a complete graphical OS (GEOS) in 64KB. And a dozen games on a 170K floppy :)

I think it's time to dig up an emulator and play some Impossible Mission, Space Taxi, and Jumpman :)

Re:My 486 DX/2 66mhz machine hardly push 200kbps

[zakezuke]

I tend to disagree. If a 486 can't push that many K/s, it's not the processor's fault, but the OS (and the program you try to push with).
On my mac G4 450 Mhz (hardly a rocket), Camino browser easily d/l's @ 200 K/s, while MS Explorer can't go higher than 50 K/s, and then goes down in speed. It's terrible!

Actually, I think this misconception of the speed of the 486 is due to people who are

1. Using serial port communications
2. Have a crappy serial port

I got this faulty logic when I was hunting for a SCSI rom drive for a 486sx PS/2. I was told by the staff "oh, it's a 486, well they can't use anything but a 1x drive anyway" and it's like "oh really, so I guess I have to use a 1x hard disk cause modern ones are just too damn fast".

My 8bit experence is pretty limited to the Atari, but I did own a scsi controler and had a 1meg ramdisk and let me tell you there was a serious peformance increase. Given the fact that it's practicaly impossible to get replacement drives and such for these vintage computers, it makes sence to go ethernet.

Re:Uhmm..

[Zan Zu from Eridu]

The C64 had a 6510 actually, but the 1541 (diskdrive) used a 6502 for a disk controller. The way the 6502 was used in the drive was a cool hack btw., they had it running two threads (each having 128 bytes of stack).

I loved the C64, and 20 years later I still know some of the ROM routine addresses by heart (like the obvious $ffd2); but when I got down to studying the 1541 it was a revelation, I got the feeling for the first time there was code in there so clever I couldn't have written it myself. I was about 17 at the time, and it convinced me that I could actually learn something in CS. Ah, memories :)

I highly doubt it

[multipartmixed]

From a cursory glance at the board, it looks like it plugs into the user port. That means it has access to the data lines from the 6526 VIA, which yields a single memory-mapped address for I/O.

That means the fastest you could write a page would be something like this:

STA 56579, 255
LDX, #0
LOOP:
LDA $BUFFER,X
STA 56577
DEX
BEQ LOOP ...I think that adds up to 3 + 3 + 1 + 2 cycles per byte, and an overhead of at least 6 more cycles per page crossed. You could shave off two cycles out of the loop by using addresses in the zero page, but since some of those addresses are reserved, you wouldn't be able to use all 255 bytes.

In order to do DMA, the controller would need to plug into the expansion port, which gives you direct access to the address and data lines of the system bus. But as another poster pointed out, you have to blank the video during transfers to achieve maximal throughput due to the VIC-IIs habit of stealing cycles for itself.

Re:Uhmm..

[Anonymous Coward]

The zero page was available for the 6502, too, so this is not the difference - you might know this, but I doubt the former poster understood it.

The 6510 not even had bank switching capabilities. The difference was that it had a 6 bit I/O port, which was uses in the C64 in combination with the PLA to implement bank-switching.

It was only 6 bits because there were no more pins available in the DIL40 package.

The other differences were merely minor. The data sheet tells something about a revised timing, which should have causes problems if interfaces with some other parts from the 65xx series (like a 6522) - but it seems it did not do.

The reason for this was that the 6502 had a setup time > 0 for before Phi2; anyway, since the VIC 6567/6769 was to share the bus with the 6510, it was not feasible to have a setup time bigger 0, so this detail was changed.

Re:New kind of bottle neck

[Afrosheen]

The problem isn't the satellite concept itself, but in the methods used. In order to have lots of legal satellite sending units at people's homes, they have to be weak. Weak signals take longer to receive, hence latency.

You do know, of course, that most calls to overseas get bounced off lots of satellites before they reach the caller. Most international calls are still nearly realtime, not half as bad as a 1000ms ping which home satellite connections give you.