1200 networking speed report.

[bbond007]

I have done some tests with various A1200 networking options and here are the results:

Hardware:

A1200 Ver 1D4
DCE Blizzard 1260 80MB RAM
Subway USB
SMC EZ Connect USB Fast Ethernet Adapter
Linksys WPC11 V3 Prism 2.5 PCMCIA 802.11b Wifi Adapter
Dlink DFE-670 10/100 PCMCIA Network Adapter
Netgear WNCE2001 universal RJ45 to WiFI adapter
Netgear XET104 Powerline ethernet port with 4 port switch(up to 85MB)

USB
GET / PUT  (KBPS)
131.45 / 124.05 (XET104)
104.73 / 78.21 (WNCE2001)

WPC11
329.83 / 273.55

DFE-670
468.55 / 334.16  (XET104)
373.05 / 332.29  (WNCE2001)

Tests were performed on :
  - OS 3.1
  - MiamiDX stack
  - MiamiFTP
  - SimpleFTP (server/windows)
  - Resolution 1024x768 HGFX 4BPP@16 Colors
  - 64MB test file
  - RAM:

[amyren]

Thanks for the test data.

By using capital letters KBPS you mean kilobytes (kBps) per seconds and not kilobits (kbps), right?

Normally network speeds specs are given in Mbps (megabit per second), so in your case the fastest rate would equal 3.66 Mbps and the slowest 0.6 Mbps.

The biggest bottleneck here seemes to be the USB option. From wikipedia even USB 1.0 lowspeed mode should manage 1.5 Mbps. But maybe the combination of the subway/SMC EZ/WNCE2001 is the problem.
It would be interesting to know what causes this slowdown.
Have you tested the Subway USB/SMC EZ with ethernet cable direct to your windows server og via your router?
If its still very slow, test the subway for USB transfer speeds (direct to a usb storage).
Then test the SMC EZ itself on another computer (windows/mac), just to see if the amiga drivers could be the problem.

[Zac67]

USB adapters are very prone to stalling in certain conditions. When a network packet is received by the adapter it's buffered locally, then the host needs to poll the adapter to find out there's data to collect. The data is read over a - supposedly - slow bus (don't think clockport offers any significant throughput) which empties the buffer and enables the next packet to be buffered. A larger buffer (=better, more expensive adapter design) provides for receiving packets while transferring previous ones to the host, but there's only so much you can do.

Even with a much faster USB 2.0 host adapter on a modern machine it's hard to reach Fast Ethernet limits, let alone Gigabit - there's simply too much overhead and all the latencies add up. Plus, USB doesn't really help missing overlapped I/O, decent interrupts, etc.

I remember my experiments with a Fast Ethernet ISA(!) card many years ago: it simply had no point.

A native Zorro II card (or Z III - I'm dreaming) should be able to get somewhat higher speeds by lowering the overhead and increasing bus speed. However, to get near-maximum throughputs you need to give your NIC direct memory access on a low-latency no-hassle basis (like PCI or PCIe).

[bbond007]

The biggest bottleneck here seemes to be the USB option. From wikipedia even USB 1.0 lowspeed mode should manage 1.5 Mbps. But maybe the combination of the subway/SMC EZ/WNCE2001 is the problem.
It would be interesting to know what causes this slowdown.
Have you tested the Subway USB/SMC EZ with ethernet cable direct to your windows server og via your router?C
If its still very slow, test the subway for USB transfer speeds (direct to a usb storage).
Then test the SMC EZ itself on another computer (windows/mac), just to see if the amiga drivers could be the problem.

USB mass storage on subway only goes 250k/sec max. Even though subway supports USB  2.0 protocol standard, it can't achieve even 1.0 speeds over the clock port. I doubt other USB adapters would do much better.

I could run the test with a long Ethernet cable and eliminate the power line adapter but the computer running the FTP server is also on a power line adapter. I only have one extra long cable,

[Cego]

i have to dig this one out. I've got the same Wifi card in my A1200 with Apollo 1260@80MHz, 32MB, Indi AGA MKII and my Wifi speed is about 100kb/s in download. :confused:
It is much faster on your config. When i'm downloading a huge file, the mouse pointer starts to lag and CPU goes up to 100% usage.
I have a basic MiamiDX install. On Genesis its the same, speed is a bit less around 90kb/s. I have tested downloading with browse and amitradecenter.

any ideas?

[Oldsmobile_Mike]

any ideas?

I love Miami for the GUI, but if you want speed, you need this:

http://roadshow.apc-tcp.de/index-en.php

[Cego]

Well, miami may be slower than other stacks, but not that slow. I should have something about 300kb/s. i'm nowhere near that. With Genesis its the same. btw, i need miami for routing my a4000 into the internet, which is connected via magplip cable to my a1200.

[Oldsmobile_Mike]

but not that slow

LOL. Better watch out, I said that exact same thing once in the forums, and dang near got my head bitten off by the fanatics.

Good luck, hope you're able to get it figured out!

[mechy]

Well, miami may be slower than other stacks, but not that slow. I should have something about 300kb/s. i'm nowhere near that. With Genesis its the same. btw, i need miami for routing my a4000 into the internet, which is connected via magplip cable to my a1200.

Miami is known to be much slower than roadshow. It is the slowest tcp/ip stack of all of them. Seeing how you are getting slower speeds with other stacks also, it sounds like you have a bottleneck somewhere else.

Check out some testing on hdzone here:

http://www.hd-zone.com/2011/07/ethernet-adapter-speeds-amigaos-3-9-and-amigaos-4-1-classic/

[danbeaver]

Download speed is linked to CPU speed and data through-put.  Hence the Z3 X-Surf 100 + AmiTCP 4.0 gets the best numbers, but getting there on an A1200 would require a tower set up, and I'd just go with an A3000 or A4000.  Sneaker-net may be best for non-surfing situations (large file downloads).

[Damion]

Well, Miami isn't always slower in every case, and it's a different story if you have a card with an MNI driver vs a poorly written SANA-II driver (like the Ameristar A4066 - in that situation, Miami and Roadshow trade blows, and Miami is overall the better choice because of the superior driver).

I wish I could help, I had the same accelerator (Apollo 1260 @ 80MHz) and the network cards ran well. You can tweak the caching mode of the PCMCIA port with Thor's MuTools for slightly better performance, but I don't think this is your problem.

[mechy]

Download speed is linked to CPU speed and data through-put.  Hence the Z3 X-Surf 100 + AmiTCP 4.0 gets the best numbers, but getting there on an A1200 would require a tower set up, and I'd just go with an A3000 or A4000.  Sneaker-net may be best for non-surfing situations (large file downloads).

Only there is no zorro3 for 1200 towers  well except that buggy one micronik? made
for the most part everyone with 1200 expansions would be limited to z2 mode of the x-surf 100.

[mechy]

Well, Miami isn't always slower in every case, and it's a different story if you have a card with an MNI driver vs a poorly written SANA-II driver (like the Ameristar A4066 - in that situation, Miami and Roadshow trade blows, and Miami is overall the better choice because of the superior driver).

I wish I could help, I had the same accelerator (Apollo 1260 @ 80MHz) and the network cards ran well. You can tweak the caching mode of the PCMCIA port with Thor's MuTools for slightly better performance, but I don't think this is your problem.

The A4066 may be the one exception since it was prob. one of the best old ethernet cards you could get speed wise and the only one i haven't owned. I have tested just about every other ethernet card out there with MNI drivers and with sanaII and  they come up slower in Miami than Roadshow. Roadshow actually triggered a bug in the x-surf drivers that got fixed if i recall.

So what speeds do you get on the A4066 and Miami?

[Damion]

The A4066 may be the one exception since it was prob. one of the best old ethernet cards you could get speed wise and the only one i haven't owned. I have tested just about every other ethernet card out there with MNI drivers and with sanaII and  they come up slower in Miami than Roadshow. Roadshow actually triggered a bug in the x-surf drivers that got fixed if i recall.

So what speeds do you get on the A4066 and Miami?

I have some netio and tcpspeed benchmarks saved somewhere, I'll try to dig them out. The impressive bit was that with Roadshow, the Ariadne (original model) was equal, or perhaps slightly better performing than the A4066 using MiamiDX+MNI. Measure both cards with MNI drivers and the 4066 is clearly better.

I don't recall any significant differences in the numbers between the two stacks with the A4066, but the A4066's poor SANA-II driver provides a bad overall experience, where it's seamless with Miami (especially web browsing).

Ultimately, I left the Ariadne in the machine because of the newer stack, similar performance, and the unfortunate fact that the A4066's SANA-II driver doesn't support multicast, so no Shapeshifter.

It's a shame, because with its better hardware, the A4066 could be king of the classic cards. Getting the best out of it means sticking with  Miami.