I get occasional crashes when trying to open IB 2.4 under OS4 Final. The crashes are always exactly the same place and DSI.

Would you like a crash log? Here or privately?

@acefnq

I used to have a sheet of paper from back in the dreamtime when I first got my A1. I made up a matrix of the 16 possible switch combinations and the CPU multipliers that you got from each.

Although I can't find the damn thing now, I do remember multipliers like 8x (x133=1066), 8.5x, even an 11 or 11.5.
The multipliers were all over the place, binarily speaking, so you couldn't set up the switches without knowing the code.

Now 11x133 = 1466, so it should possible to get speeds of that order out of it. I'm sure I remember 1.3-1.4 GHz in the table somewhere.

I'm certainly not running my A1 at those speeds. With the dearth of hardware around, I'm not going to risk losing it and I'm running it a very conservative 800 MHz. Yes, it runs fine at 1066 with good cooling, but its life would be reduced (like any hardware that is run harder).

We've reorganised the house since the kids moved out and I'm sure the sheet of paper has been "filed" somewhere, but I've been unable to find it.

Jack, did you get on to PowerAmiga? He runs his machine at a good clip and might have the data you need. Also Excalibur (aw.net) might have the same data. If you can't get in contact with either of them, PM me and I'll email them.

@acefnq

If you ever do break the U-Boot chip, I have here plenty of them, and the means to program new ones, so just yell.

@Helge

This doesn't answer your question, but here is their site:
http://www.acube-systems.biz/

Any $amount that is quoted now is sure to be changed when they get into production.

@LiveForIt

TD_EJECT is not supported by the floppy driver. The internal "identify disk" function determines if a disk is present, then reads it in HD and DD format until it gets an error-free result. It then calls all connected filesystems if there is any change.

The DiskChange command does not issue any requests to the driver, only to the filesystems, so the driver is not told that a disk has been changed.

@abalaban

What about printing from Multiview? You should be able to print text and graphics from there. If that runs without crashing (as it should), then I'd blame AWeb.

If you can't print from Multiview without crashing, then I'd suspect an installation problem.

To prove it, use SGrab to capture the AWeb page (save to RAM:), then use Multiview to view the captured image and print it.

@LiveForIt

Please don't tell people to use "Diskchange". It does not tell the driver anything, it only tells the filesystems and they can't tell the driver that a new disk might be there.

To get the driver to check the disk formatting, you must use the FDTool utility, either in immediate mode or auto.

@hnl_dk

Tried the version of coreutils from 2006-10-29, but no better.

The newer "config.sub" doesn't get very far. It claims that gcc doesn't work and fails. Still trying to find out why.

[edit] Silly me, it IS working now with the newer config.sub and config.guess. It's now failing later in the configure script, so I think that solves the current problem.

Thanks for your help.

[I hate Loonix]

[/edit]

@hnl_dk

I'm using sh 52.1 (2007-01-07), coreutils 6.7.2 (2007-01-04).

I'll try those later versions, but if abc-shell "can't find" the file under its nose, it can't be a bug in the file contents.

I have a problem executing Loonix scripts under some conditions.

I am trying to configure GMP in its own drawer. I can configure Kaffe in its drawer, so the problem is not universal.

What happens is that when the script tries to run another instance of 'sh' to run a sub-script, the sub-script can not be found. If I try to run that sub-script from the command line, it also says it can not be found.

eg:

/Java/GMP/gmp-4.2.1> ./configure --build=ppc-amigaos
configure: creating cache /dev/null
configure: error: cannot run /bin/sh ./config.sub

/Java/GMP/gmp-4.2.1> ./config.sub
"config.sub" not found

The permission bits for the files are:

/Java/GMP/gmp-4.2.1> ls -l config*.*
-rwxrwxr-x 1 root wheel 23065 2006-05-05 02:36 config.guess
-rw-rw-r-- 1 root wheel 15353 2006-05-05 02:50 config.in
-rw------- 1 root wheel 4437 2007-01-14 20:44 config.log
-rwxrwxr-x 1 root wheel 4127 2006-05-05 02:37 config.sub
-rw------- 1 root wheel 1966 2007-01-07 15:41 config.sub.info
-rwxrwxr-x 1 root wheel 43636 2006-03-15 01:57 configfsf.guess
-rwxrwxr-x 1 root wheel 31164 2006-03-15 01:57 configfsf.sub
-rw-rw-r-- 1 root wheel 111778 2006-04-30 09:31 configure.in
-rwxrwxr-x 1 root wheel 1073487 2007-01-14 19:20 configure.old

Since this process runs quite well in another directory using other scripts, I suspect it is something to do with the permission bits or sh's treatment of them.

Stack is 5 MB.

Any ideas?

@Curty

Unlikely that you tightened it too much, but since it's only a nylon spacer, it will "creep" with time and gradually get even shorter.

If you can see that the module is crooked, perhaps it might be an idea to make a couple of simple washers out of some thin plastic (like a Chinese food take-away container or soft drink bottle) and slide them under the existing spacers without removing the module.

Don't try to use anything metallic, though. Too much breakable stuff in that area.

@Curty

When you have it out, sight along the board and check that the CPU module is sitting parallel to the board. The spacers supplied with the XE were slightly too short and really need to be about 0.5 mm longer to make the socket sit straight.

At least one friend has to press his CPU module back in every time he moves his machine - refuses to fix it properly :shaking head in dismay:

@samwel

Rich has asked us to post this reply on his behalf:

Quote:

@Billsey

I have (thankfully) few friends, and neither of them uses *Outlook* (they prefer Eudora or Pegasus, etc).

@afxgroup

I find that killing everything that comes from "X-Mailer=Outlook" gets rid of most of it. A large part of the remainder is addressed to a fake "To" name/address, so that goes in the bin, too.

Not much left after that. YAM does a good job.

@Jack

Sorry, I can't find the table I drew up some years ago. I can't even remember where I got the info from, but I do remember changing the switches and reading the CPU speed from U-Boot output.

If I find it in the near future, I'll send it to you. It's not the sort of thing to post in public, because some people might try it just for fun and break something.

@aldur

So what are the modes listed in the driver tooltypes? Which of those have you selected?

Do you have an old P96Settings file in Devs:? If so, get rid of it or rename it (even if it is commented out in the tooltypes).

Are the Max/Min scan rates in the tooltypes correct for your monitor? Are the modes in the tooltypes recommended by your monitor manual? Especially make sure that the refresh rate is not too fast (----x----@RR). Many TFTs give shifted, dark pictures if the RR is too fast.

@Jack

That's not right, you can set the switches to get multipliers like 11/11.5 (can't find the piece of paper at the moment). 1066 MHz is not the fastest.

You can certainly set a G4 to run at (IIRC) 1.3-1.4 GHz. I'll try to find the data and get back here.

You could try asking Excalibur or PowerAmiga, they should have the data.

@Grumps

I'd guess that it's nothing to do with the Radeon driver, since it only appears a long way down in the stack trace.

What DOES appear (recently) in the stack trace is icon.library calling DOS for something or other.

Have you got any funny icons and/or icon hacks running?
Why not remove them and see if it is any better?

@lazi

Not all the sii#? modules are SATA, the sii0680 is of course the original PATA driver.

Each mounted partition occupies a slice of memory. On my system I have about 25 separate partitions and each of those takes up memory for buffers.

If you want to minimise memory usage, you could disable some partitions in the ESM or unplug the drives you don't want to use today.

Also every running commodity takes up memory, so disabling screen savers and other stuff from WBStartup will save some more.

Re the WB memory figure:

The title bar uses Avail() to calculate free memory. Since some unused memory is always tied up in memory pools, the figure you get from Avail() is pessimistic. Just how pessimistic varies every time you call it. "Avail Flush" will free unused libraries etc, but not the pooled memory.