Roadshow 68K - Needs your support!

[Slayer]

Nothing is ever easy eh?

Keep it simple guys, that's actually how things move...

Yes, I'd buy it

[Olaf Barthel]

Originally Posted by Zetr0

For a reasonble price I am know to register almost anything =)


if Roadshow 68k really wants to make some pennies, then I would also suggest a USB stack - compatible with the Spider/Subway and Thylacine cards =D

Sounds like too much work to me. I usually tried to make software I would use myself (I believe in "eating your own dogfood" when it comes to software development), and a USB stack isn't exactly what I need for my Amiga...

[Olaf Barthel]

Originally Posted by alexh

Roadshow was written for PowerPC class processors with gigabytes of RAM.

Actually, it was written for the kind of Amiga hardware commonly available in 2001. I developed it on my trusty A3000UX, but it worked fine on a plain vanilla A1200HD running OS 3.1, with a PCMCIA Ethernet card installed. Of course, the more memory you can throw at the network I/O, the more throughput you can squeeze out of the stack (within limits, of course: the networking hardware will only go so fast).

Also what would the driver class structure be? All existing SANA-II drivers? If it is a new driver scheme then what hardware will have drivers?

I did not reinvent the wheel. It's SANA-II, and nothing else. Curiously, Roadshow outperforms Miami on the same hardware, even with Miami using its custom MNI drivers.

What would be the advantages over Genesis (aka AmiTCP) & Miami? IPV6? ZeroConf? VPN? Firewall? IPSec?

Compared to AmiTCP it's a more modern TCP/IP stack. AmiTCP is based upon an early 1987/1988 4.3BSD TCP/IP stack, whereas Roadshow uses code that was released around 1994. Roadshow has a built-in DHCP client, which AmiTCP has not. Roadshow also supports ZeroConf for network interface address assignment in the absence of a DHCP server (neither AmiTCP nor Miami does that). Roadshow also supports a firewall, specifically Darren Reed's ip filter package.

Compared to Miami and AmiTCP, Roadshow features significantly higher data throughput using the same drivers. For example, with Holger's own ppp.device Roadshow runs faster than Miami does. Roadshow comes with its own PPP and PPPoE drivers, by the way.

Unlike Miami, Roadshow does not have a GUI. You would be editing configuration files, just like with AmiTCP. But unlike AmiTCP, Roadshow detects changes made to the configuration files at runtime and will reread them automatically. You will not need to restart the TCP/IP stack after making changes. Roadshow also does not come with its own SSL implementation, but AmiSSL works just fine (this is what OS4 uses, but a plain 68k version of AmiSSL has been available for a while, too).

As for IPv6, VPN and IPSec, that's way out of our league on the Amiga, and this is how things are probably going to stay for a while. Roadshow implements none of these, but it doesn't make prevent eager developers from developing or deploying VPN or IPSec. There's an SDK, which has been around for a while.

Miami is a bit of a pain to set up but once working it has relatively low CPU overhead and things such as NAT and DHCP.

The NAT requires a bit of ip filter config file wrangling, and DHCP works out of the box with Roadshow.

CPU usage should be better than with AmiTCP, because of how Roadshow internally separates the kernel operations from the user code operations. AmiTCP has an elaborate set of procedures for that which mimick the original BSD interrupt handling. Roadshow has none of that and never drops into Forbid() at all when dealing with I/O.

[cosmicfrog]

great post Olaf

so what ideas on price have you got ??

[kriz]

Its really interesting, i will for sure support if it will be released ! About time for some new serious software in 2010 !!

[Olaf Barthel]

Originally Posted by cosmicfrog

great post Olaf

so what ideas on price have you got ??

I don't have any. That's one of the problems that need to be solved.

What price would you suggest, given the facts about the product I outlined in the previous posting?

[ppill]

I'd be ready to pay somewhere around 15-25 euros. And it sounds great. Is there a demo available?

[Zetr0]

@Olaf

Welcome to EAB, as you can see theres plenty of mischef around here that we normally get upto =)

I admit your project is most certainly of interest to me and since you are here, I have a few questions (sorry if they are awkward ones)

1. What sort of price point / target price would you be looking for this software?
2. When you say significantly higher data throughput, how much higher (10% or 20% perhaps 80% etc) ?
3. Will you use a resgistering key (like WHDLoad) i like that software =D

Have a chat with Wepl, to discuss how he feels his registration method works for him, it may work for you or atleast provide you a model to develop in distrubting you work.

One of the reasons I registerd WHDLoad (Perhaps the Biggest) is the tie-to the Amiga community it gives me. Your TCP/IP stack Olaf, could also be a big part of this since I hope that you will release an SDK, I am sure some GadToolBox guru's will love to make GUI's for some of the work and help improve it over time.

Like I said yours is a very interesting project, its got to face off to and prove itself against some tough established free alternatives, but theres a lot of support in the community for the right projects, I am sure that this could do very well indeed =)

I wish you only the very best in this project, I can promise I will be watching this very VERY keenly.

--- Sorry for the slight OT bit guys ---
--- Yeah it was my fault and the need for a Spider USB STACK ---

--- I am a bad Zetty =( ---

@Camstah'

Alas I have not tried Anaiis, however I did read up on your work with it, I was quite impressed =), Now I have lots of equipment I could probably help Giles out with, if you could hit me up with his contact details in PM I will see what I can do.


@Alexh

Indeed, for the moment it does seem a little bewildering to know where to begin to rip the sources... its one of those.... if only i had the time... its not one you can pull - out in 30 minutes (or 3 hours it seems) and think about implementation for the rest of the day and have a rough up by the weekend... no.... this will take a GOOD couple of months.... alas dont think I have any of those yet...

I will ask / beg / plead with the AROS guys to see if they can help me in this endovor when I get time.... when...... LOL

[Zetr0]

Originally Posted by Olaf Barthel

I don't have any. That's one of the problems that need to be solved.

What price would you suggest, given the facts about the product I outlined in the previous posting?

LOL this will teach me not to make long posts as I miss the discourse that happens lol.


hmm... realistically, I think $10 or slightly higher (transfer / banking fees etc) for registration, - similar to say the way WHDLoad works - it could have a 4 second splash screen or text saying

"Roadshow 68k Advanced TCP/IP Stack, registerd to XXXXXXXXX"

That would be nice =D

[Olaf Barthel]

Originally Posted by Zetr0

@Olaf

Welcome to EAB, as you can see theres plenty of mischef around here that we normally get upto =)

I admit your project is most certainly of interest to me and since you are here, I have a few questions (sorry if they are awkward ones)

1. What sort of price point / target price would you be looking for this software?

I haven't got any idea yet. Back in 2002 I still had hopes to sell this, with a GUI, a manual and a CD-ROM for less than about 40€. But that isn't a remotely realistic figure any more, especially since there's no GUI and the Amiga market has taken several turns for the worse in the last decade.

2. When you say significantly higher data throughput, how much higher (10% or 20% perhaps 80% etc) ?

The first rule of applied statistics is to make up some figures which look good enough Here are some of the numbers I took note of during development. These are data throughput figures, created during a sustained download in the LAN which lasted several minutes. The respective test setup (ftp server hardware and server software involved, test duration, etc.) was always identical, only the Amiga hardware and software changed. The "Ariadne I", "Ariadne II" and "A2065" are the respective Amiga networking cards. As you can see, performance improved as Roadshow evolved over time.

Here is the list:

Ariadne I, Roadshow 4.69, A3000T (040, 40 MHz)
891 KBytes/s

Ariadne II, Roadshow 4.69, A3000T (040, 40 MHz)
845 KBytes/s

A2065, Roadshow 4.69, A3000T (040, 40 MHz)
781 KBytes/s

Ariadne I, Roadshow 4.69, A3000UX (060, 50 MHz)
891 KBytes/s

Ariadne I, Miami 3.2b SANA-II, A3000UX (060, 50 MHz)
669 KBytes/s

Ariadne I, Miami 3.2b MNI, A3000UX (060, 50 MHz)
882 KBytes/s

Ariadne I, Miami Deluxe 1.0c SANA-II, A3000UX (060, 50 MHz)
633 KBytes/s

Ariadne I, Miami Deluxe 1.0c MNI, A3000UX (060, 50 MHz)
860 KBytes/s

Ariadne I, AmiTCP/IP Genesis 4.6, A3000UX (060, 50 MHz)
812 KBytes/s

Ariadne I, Roadshow 4.71, A3000UX (060, 50 MHz)
916 KBytes/s

Ariadne I, Roadshow 4.162, A3000UX (060, 50 MHz)
941 KBytes/s

Ariadne I, Roadshow 4.205, A3000T (040, 40 MHz)
977 KBytes/s

As you can see, Roadshow goes about as fast as the 10 MBit/s Ethernet hardware will permit. None of the alternatives (Miami, AmiTCP) come as close to the hardware limit as Roadshow does.

3. Will you use a resgistering key (like WHDLoad) i like that software =D

I hate having to hand out key files. This approach is proven to be both a nuisance for the guy who makes them, and is then naturally disappointed when unauthorized copies are being made, and to be practically useless.

[Olaf Barthel]

Originally Posted by Zetr0

LOL this will teach me not to make long posts as I miss the discourse that happens lol.


hmm... realistically, I think $10 or slightly higher (transfer / banking fees etc) for registration, - similar to say the way WHDLoad works - it could have a 4 second splash screen or text saying

"Roadshow 68k Advanced TCP/IP Stack, registerd to XXXXXXXXX"

That would be nice =D

As for splash screens, one of my reasons for writing Roadshow was that I found Miami's splash screen to be overly intrusive and annoying. With Roadshow you can boot your system and have the network come up before Workbench opens, and you will never ever even have to see a splash screen. I hope to keep things this way.

[vulture]

I too prefer it without a splash screen or window of any kind. Just boot n go. BTW the transfer results you posted are very very good. I hope they're about the same on A1200 with a pcmcia net card.

[Skope]

Originally Posted by Olaf Barthel

As for splash screens, one of my reasons for writing Roadshow was that I found Miami's splash screen to be overly intrusive and annoying. With Roadshow you can boot your system and have the network come up before Workbench opens, and you will never ever even have to see a splash screen. I hope to keep things this way.

My cup of tea right there. Good luck with the project, I'm definately supporting it.

[chiark]

This sounds excellent, and I'd be all for it! To be honest, even with no improvements in performance or features it would be worth registering to have a supported stack on the amiga...

Pricewise, I'd be very happy with €15-25, and if there was a GUI tool in the future I think you could definitely be at the bigger end of that range. Perhaps €10-15 for the stack and €5-10 for the GUI if it gets written?

It's great to see this name about again

[kriz]

This sounds really nice, but we need some sort of gui... I will pay about 20-30 Euros for something like this.

[Zetr0]

@Olaf

my thanks for your answers and some impressive metrics =)

would you have any tests results for those on lesser 68k based machines like the humble 68k moto or 020 / 030? (would you expect them to be about the same?)

I can see there is a limmitations of the adapters and this (pending on the stack used) swings quite wildly... especially with your latest "Roadshow 4.205" core achieving almost 37% more on an Adriane 1 equipped 040@40mhz [977KB] compared to "Miami Deluxe 1.0c SANA-II" but with an 060@50 [633KB]

Thats quite impressive - not to mention 977KB is has probably maxed out the Adriane, what do you think this stack could achieve on a PCI based 10/100 card?

(forgive my ignorance its been a while, but this would likely be limmited to the bus-board speed? - Z2 being about 3.58MB a second and Z3 being about 9MB - of course the CPU will make a difference.)

Some great stuff Olaf =)

[ppill]

A GUI for a stack that supports DHCP and Zero conf from the get-go? Surely most of the people nowadays have some modem/router thingy that supports DHCP.

Some impressive speeds but like others have mentioned a 1200 + PCMCIA card would be of interest to a greater crowd.

A legal TCP/IP stack that support autoconfiguration (fire and forget style), faster that the rest and for an affordable price? Hell yes!

And how's the memory footprint?

[Olaf Barthel]

Originally Posted by vulture

I too prefer it without a splash screen or window of any kind. Just boot n go. BTW the transfer results you posted are very very good. I hope they're about the same on A1200 with a pcmcia net card.

The transfer rates on the A1200 very likely won't be quite as good. Here's why.

The tests would use a modified version of the "wget" utility which, while it would read the data received, would not write it to disk. This modification neatly excludes any delays caused by disk I/O. While the test was running, the machine did practically nothing else that would interfere with network I/O operations.

For the top end of the performance figures, the "Ariadne I" card was involved. That card featured a large (32K) on-board transmit buffer which in combination with the AMD chip and the hardware design helped to attain the high transmission speed. I have yet to see an Ethernet card for 68k Amigas which comes close to the performance of the "Ariadne I" (well, there was the Ethernet interface for the DKB Wildfire, which could do 32 bit DMA, but you can't compare this to the "Ariadne I" design).

Finally, there's the machine's RAM configuration. I conducted my tests on an A3000UX and an A3000T. Both machines would use phase 5 CPU hardware, but the memory interface of the A3000T is slightly faster than the the one used in the A3000UX. Because network I/O on the Amiga involves copying data to/from main memory and hardware buffers very heavily, a system with high performance main memory will score highly in terms of network I/O performance.

Sorry to disappoint you, but an A1200 is unlikely to reach the same level of performance as these older Amigas.

[vulture]

Still, it'll be faster than Miami so, it's ok for me. On my 060 A1200 w 3com I ftp dl from my pc with amitradecenter at about 500k/s via MiamiDX. If it gets to 600-700k I'll be a happy camper! If not, well, I'll still have a lightweight tcp-ip stack starting on boot. Not bad! Also another question, right now we have to disable amitcp or miami to be able to run whdload games. Will roadshow have the same issue? Is this something the whdload devs should fix or is it simply one of those things that can't be fixed because of the way things work? Thx!

[Olaf Barthel]

Originally Posted by Zetr0

@Olaf

my thanks for your answers and some impressive metrics =)

would you have any tests results for those on lesser 68k based machines like the humble 68k moto or 020 / 030? (would you expect them to be about the same?)

I only tested this on my development machines, and it's been a while since I did those tests (Roadshow 4.205 was in development in 2003). I'm not sure I can muster a plain 68k machine to test with. I have an A600HD in storage (acquired for research purposes: I wanted to figure out why the scsi.device driver in the A600 OS 3.1 ROM was version 39 and not version 40), but if I remember correctly, the driver required for the only PCMCIA networking hardware I have does not support the 68000 processor.

The overall performance depends upon several factors: memory performance, networking card transmit buffer size, CPU performance. If you don't hit the high notes on memory performance and networking card transmit buffer size then your performance figures will be more on the disappointing side, I'm afraid.

I can see there is a limmitations of the adapters and this (pending on the stack used) swings quite wildly... especially with your latest "Roadshow 4.205" core achieving almost 37% more on an Adriane 1 equipped 040@40mhz [977KB] compared to "Miami Deluxe 1.0c SANA-II" but with an 060@50 [633KB]

Thats quite impressive - not to mention 977KB is has probably maxed out the Adriane, what do you think this stack could achieve on a PCI based 10/100 card?

As I wrote in a previous posting to illustrate how these figures came about, the transfer figures are not entirely realistic as they merely exercised sustained data transmission without doing anything useful with the data. The 977 KByte figure, although it was repeatable, almost seems to be physically impossible, given the limitations of the Zorro II bus.

On a PCI based system with a faster bus this TCP/IP stack probably won't go that much faster. The reason is in the SANA-II driver design which does not mesh at all with how PCI networking hardware does DMA. Hence, you will end up copying to/from the networking hardware buffer to local Amiga memory a lot, and this is where the performance hits a wall.

(forgive my ignorance its been a while, but this would likely be limmited to the bus-board speed? - Z2 being about 3.58MB a second and Z3 being about 9MB - of course the CPU will make a difference.)

Some great stuff Olaf =)

The bus speed is one of the limiting factors. But it also matters how the networking hardware offers up the data it receives. If you have a small receive buffer, like apparently all of the PCMCIA cards do, then the consequences for the I/O performance should be obvious. But it's not just the buffer size alone. If I remember correctly, the "Ariadne I" and the "Ariadne II" both have 32 KBytes of on-board transmit memory, but the "Ariadne II" performs slower than the "Ariadne I" under otherwise identical circumstances.