My network is not reliable on my machine. When I stress the network under linux or OS4 update1 my network locks up after some time (half a minute).
I stress the network by ping flooding my A1 from another machine (ping -f) so I get a constant stream of packets to my A1. When the lock up appears the machine is dead on the network. I can't ping to or from the A1 and this behaviour is the same under linux or OS4.
If I only do the ping flood everything runs OK but the problems occur when also having heavy IDE e.g. when I'm copying large files from my slave IDE CDROM to my master IDE drive.
I have an A1SE with 128MB of RAM, Radeon 7000, Catweasel MKIII available, no sound. The interrupts in the BIOS are configure OK: all level and 9,10,11 and 7. On the motherboard I can see small green wires so I think the 3COM patch is applied to the motherboard.
Please help.

Staf.

Edit: mention IDE load and BIOS interrupts.

Last edited by Fats on 23-Oct-2004 at 10:04 AM.

@Fats
Take a look at _MCP 7450_ thread in this forum.
Snuffy may have found an answer to your problem.
Worth a try anyway.

Last edited by Mechanic on 23-Oct-2004 at 09:30 AM.

@Mechanic

I removed the Catweasel MKIII from my A1 but the problem still remains.
Staf.

@Fats

I think mech was talking about making sure your voltage setting were correct for your CPU.

kgrach

@kgrach

Can somebody give me the definite settings for the voltage regulator for a A1G3SE (750CX from /proc/cpuinfo)? (Maybe a picture would be nice also).
I always assumed that the settings that were delivered from ComputerCity for a SE would be OK. I contacted Ron from ComputerCity and he doesn't know anymore.

@Fats
Before going to the voltage settings;
I did not see a Silicon Image si680 card
listed in your first post. Do you have one.?

It's been determined that the ethernet problem
is really a problem with the IDE interface
( VIA bug ).

P.S. I have a G4.

Last edited by Mechanic on 14-Nov-2004 at 09:08 PM.

@Mechanic

I don't have a SiI card. But I thought that it would not happen on OS4 because the IDE driver is in PIO mode and doesn't use DMA. Or are problems also expected in PIO mode ?

@Fats

I also have a patched SE and get the similar problems; any major stress on the network (such as FTP-ing over the LAN) causes the A1's net system to die.

This is what I get, in detail:

* Internet traffic via my router (512k DSL) works fine, with downloads clocking at upto 50K/sec.
* Fetching files via FTP from another machine on the LAN causes the A1's ethernet system to die after a random period, usually between 5 and 20 seconds. Shutting down the OS4 net system and re-starting it restores connectivity temporarily.
* Throtting the FTP server to 50K/sec does not help the above issue.
* Downloads to Ram/VIA/Sil680 all have the same problems.
* Sending files via FTP over the LAN causes the A1 to freeze so completely a hardware reset is needed to bring it back.
* Ping-floods cause the net system to die too.

There's something wierd going on with the SE's ethernet port at a physical level, too. I recently changed the cat-5 cable that serves the A1 (it's bright blue and very ugly) for another one that is known to function just fine, yet the A1 just totally refuses to work with the new cable, giving a 'self-test failed' message at boot time.

I really doubt the ethernet port on the patched SEs is ever going to work okay. I'm pretty much resigned to the prospect of having to sacrifice my last free PCI slot for an RTL8110 card, once the driver for those chips is finished.

_________________
Who do you serve, and who do you trust? - Galen

@DrBombcrater

Your reply hit me like a cluestick, but the answers still elude me.

The only problem I have with Ethernet is that Amiga Explorer on Windows will lockout with a coms failure on very large files. Envoy from A4000 to AOne and reverse is fine with complete partitions being no problem (albeit at 10Mbit/s limited by the A4000).

Now, here is the odd bit. Firstly, I have an XE with all IDE on an sii card. My internet connection is via a router and is always on. I usually have IBrowse working throughout the sessions.

I was trying to install the latest GoldEd and, later, run AdPro. I got the usual GR type DSI on both occasions but AdPro continues to work perfectly well. However, the GR report cites a1ide.device as the culprit. There are no devices running on the VIA ide ports.

The GoldEd install and uninstall generate strings of GRs resulting in the machine being locked and un-rebootable. No problem, I installed on the A4000 and copied it across.

I suspect some interaction between the A1ide.device and Roadshow is causing this with the hardware being the thing they have in common, but proof is way beyond me. Are the two pieces of software causing the VIA bridge to generate a failure condition which can, in some cases, cause a complete freeze?

Regards
Roy Leith

Quote:

DrBombcrater wrote: @Fats I also have a patched SE and get the similar problems; any major stress on the network (such as FTP-ing over the LAN) causes the A1's net system to die. This is what I get, in detail: ...

I run my A1 most of the time in linux with the ethernet port switched to 10Mbit mode and this makes normal day usage possible without (almost any) network lockups.

Quote:

I really doubt the ethernet port on the patched SEs is ever going to work okay. I'm pretty much resigned to the prospect of having to sacrifice my last free PCI slot for an RTL8110 card, once the driver for those chips is finished.

So, I bought an expensive motherboard with onboard ethernet, UDMA IDE and sound. But to get these I have to add three PCI cards (together with my Radeon card I don't think this leaves room for the catweasel mkIII). Let me tell you that I'm dissatisfied with this. The A1SE (with the 3COM patch) was marketed as a fully functional board only without a finished OS. Only software problems needed to be solved in OS4 and linux. Now almost two years later it seems there are still problems in OS4 and linux is still a disaster.
I hope Eyetech will do something to correct their mistake(s); it would already be a start that they confirm there are still problems with ethernet on the SE.

Staf.

@royleith

>However, the GR report cites a1ide.device
>as the culprit. There are no devices
>running on the VIA ide ports.

Hmm,, I wonder if the GR is si680 aware or
if it contributes any IDE snafu to a1ide.?

@Mechanic

After my reply, I found the advice in another thread to comment out the a1ide.device in kicklayout if the VIA ports were not used. Having done so (twice! Only did the debug lines the first time) the a1ide GRs are gone.

I think that means that it was the a1ide.device module causing the problem. Still don't know what the problem is. I am hoping that the board patch allegedly being trialled will sort it all out and that I can put the sii board in my Amiga museum. I also hope that the SE users get their networking back to full force.

Regards
Roy Leith

@DrBombcrater
Quote:

DrBombcrater wrote: @Fats There's something wierd going on with the SE's ethernet port at a physical level, too. I recently changed the cat-5 cable that serves the A1 (it's bright blue and very ugly) for another one that is known to function just fine, yet the A1 just totally refuses to work with the new cable, giving a 'self-test failed' message at boot time.

Info tidbit.
Not all cat 5 cables are wired the same. Just like
serial cables, some of them cross pins, others do not.
Depends on the equipment being connected.

@Mechanic

I know about cross-wired cables, but that's not a factor here as the router I use auto-detects them and alters its pin-out to use cross-wired cables as normal ones. If I plug the cable into a PC (tried several) it works fine, plug it into the A1 and nothing. Dead link.

_________________
Who do you serve, and who do you trust? - Galen

@Fats

I'm sorry for you... i hope Eyetech will give a new MicroA1 to SE owners.

_________________
ikir

I finally could contact Adam Kowalczyk and he send me his latest internal test version of eth3com.device. This version seems to have solved the problem.
I'm happy with the quick help of Adam.

greets,
Staf.