I am writing you because I am desperate. I have upgraded my MDD dual G4-1.25 with a SONNET 4+4 SATA card (for use with internal SATA disks) (I asked for more info on the sonnet card - I can't find a "4+4" model at sonnet's site currently.-Mike) and also an ADDONICS eSATA to connect two 5-slot each external SATA multiplier enclosures. (Ref: www.addonics.com/products/host_controller/ads3gx4r5-em.asp and www.addonics.com/products/host_controller/ad5sapm-e.asp)
(Just for the record, what SATA drives are used?)

Since performing these upgrades my G4 was perfectly stable - I can leave it running for days in row without any problems. It never gives me problems when I use Safrai or iTunes or iWeb or any other application. I can use iMovie, VisualHub - it is rock solid

Only when performing heavy IO and I run several applications/tasks at the same time, then my computer hangs. For example I cannot perform any two of these three tasks at the same time:

1) save 25GB DV file in iMovie (this takes about 50 minutes)
2) encode another 23GB DV to MP4 using VisualHub (this takes about 4 hours)
3) copy 50GB of DV files from one disk to another (this takes about 25 minutes)

Yet, I can perfectly perform these very same tasks - one at a time!
(FYI - see reader reply below)

When I perform two of these tasks concurrently, my computer always hangs solid with the mouse pointer frozen. Only once I got a heap corruption message writing black/white text over the desktop. Because it is frozen and I have to power off; there is no bug report to apple and nothing is visible in the logs (no crash dump report). Usually the first sign is that the LED of my Apple-mouse does not light any more. Corresponding, I find entries in the system log which show the USB timed out and was disconnected. Sometimes I am able to cancel the tasks running and restart - I can avoid a pending freeze up like this (but there is always loss of data/time to restart the tasks). On my G4's I have the mouse plugged into the keyboard; and these two are my only USB devices.

Things I have done in desperate attempts to resolve the problem:
1) I have bought new RAM (The MDD needs PC 2700 but I bought PC3200 just to be sure)
2) I have tried removing the two disks on the SONNET controller (leaving only the internal IDE boot disk)
(Does removing the Addonics card help? (and/or not using the port multipler) - i.e. can you test w/SATA drives connected to the Sonnet card alone? I thought there was some past issue with some cards and some Port Multipliers but can't recall now - Michael Bean of AMUG may have mentioned this in the past IIRC-Mike)
3) I have added additional fans to my MDD as described on some over clocking pages (it is not over clocked !)
4) In my basement I kept / and found the original power supply (it was replaced by Apple because it is too noisy); and I am now running with the original - albeit very noisy power supply
5) I have bought the BELKIN USB 2.0 card and connected my keyboard to this card
(I assume he did this due to the USB watchdog log errors below, but I'd try to minimize the addons-Mike)
6) I have swapped the original Apple keyboard and mouse with other (originals) from my old G4/500
7) I have tried using an externally powered USB hub and connected keyboard and mouse to the hub
8) I have flashed the SONNET card with the new 2.0.3 firmware (it had 2.0 when I bought it); there does not seem to be a driver
9) I have loaded the latest driver for the ADDONICS card, which is SIL-3124 v2.0.3u (there does not seem to be firmware update for this card)

I have been researching and the only article which describes a similar problem with the USB getting disconnected is with ATA RAID controllers on your site (links below).
Can you maybe publish this article and/or suggest anything?
thanks, Thomas
http://forums.macosxhints.com/archive/index.php/t-17603.html
Notes/Info on USB Audio Problems with Acard RAID PCI card in OS X (old April 2003 xlr8yourmac news post)
http://www.xlr8yourmac.com/IDE/acard_ata133_raid/acard_ata133_raid_update.html (old post)

"Personal conclusion: Maxing out one data bus (PCI) should not cause problems for another bus (USB). This should be regulated by the kernel or OS. OS 9 can handle it. The bug has been reported to appear with various PCI and USB hardware (Griffin iMic, Soundsticks, Raid controllers, TV-cards) in the above-mentioned discussion group. To me, this does not indicate a vendor- or hardware-specific problem. It points solely to OS X as the culprit. So I expect Apple to fix this."

(log entries)
May 13 19:04:38 G4-1250 kernel[0]: USBF: 51551.917 AppleUSBOHCI[0x1c6e800] Watchdog detected dead controller (hcca #: 19594, hc #: 19787)
May 13 19:04:39 G4-1250 kernel[0]: USBF: 51552.917 AppleUSBOHCI[0x1c6e800] Watchdog detected dead controller (hcca #: 19791, hc #: 20765)
May 13 19:04:40 G4-1250 kernel[0]: USBF: 51553.917 AppleUSBOHCI[0x1c6e800] Watchdog detected dead controller (hcca #: 20768, hc #: 21742)
May 13 19:04:41 G4-1250 kernel[0]: USBF: 51554.917 AppleUSBOHCI[0x1c6e800] Watchdog detected dead controller (hcca #: 21745, hc #: 22719)
-Thomas B.

(If anyone else has a suggestion, let me know. Thanks.-Mike)

A reader reply to this post:

" Thomas B. says...

1) save 25GB DV file in iMovie (this takes about 50 minutes)
2) encode another 23GB DV to MP4 using VisualHub (this takes about 4 hours)
3) copy 50GB of DV files from one disk to another (this takes about 25 minutes)

As a professional video producer, I would never consider doing any of the above concurrently. You're talking about some pretty processor intensive tasks. It's not like surfing the web and checking email at the same time! If I have multiple tasks (particularly encoding or decoding) that absolutely need to be run at the same time, I divide them up over multiple machines.

That said

"When I perform two of these tasks concurrently, my computer always hangs solid with the mouse pointer frozen."

I have at times encountered what appeared to be an unresponsive machine that was just tied up with it's processing. I always keep the seconds visible in the menu bar clock to help diagnose a true hang, but have encountered times when what appeared to be a true hang wasn't. I always let a machine run for another couple of hours before doing a forced shut down. 95% of the time when I return it's back to a normal state.

My recommendation... if he can get two of those processes to run concurrently, don't add the third. Enjoy your site, it's been helpful in the past.
TK
Digital Design & Production "