Quicksilver non-chronological PCI card order

[Lee Blaske]

In another thread, Knobturner wrote:

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:<HR>About the non-chronological order of PCI buss slots, an Apple tech told me it was a last minute problem before shipping. It had something to do with the busses not getting power so they swapped around the order in which the PCI slots enter into the master buss. This COULD cause a problem with setups which require a specific slot order. I asked DIGI about it.....still no reply. But my system is rocking now even with the "inhibit..." enabled.
<HR></BLOCKQUOTE>

There definitely is something weird here.

Since my Bit3 is still in the shop, I've been experimenting with a Mix Core, two Mix Farms, and an ATTO card directly in the G4.

Here's the card order (visually):

Slot 1 (AGP slot) ProMax DH Max monitor card
Slot 2 Mix Core
Slot 3 Mix Farm
Slot 4 Mix Farm
Slot 5 ATTO EPCI-DC

When I run DigiTest on this set-up, it tells me that the system is mis-configured (although this should be correct).

When I run Apple System Profiler, I see something different than what I'd expect.

Here's what ASP shows:

Slot-1 (AGP) >PCI Slot (Display)
Slot-3 >PCI Slot (6)
Slot-4 >PCI Slot (6)
Slot-5 >SCSI Card (2)
Slot-2 >PCI Slot (6)

The question is: Why does ASP jumble up the order in this manner? Does this actually mean that the PCI slot order from lowest to highest is 1 3 4 5 2? If that's the case, then maybe I should have the SCSI card in slot two, and my Digidesign cards in slots 3, 4 and 5.

When I go back to the expansion chassis, perhaps I should put the host card in slot three, and the SCSI card in slot 2 (rather than putting the host card in slot 2 and the SCSI card in slot 5 as the Digidesign expansion manual suggests).

Any thoughts?

Lee Blaske

[Lee Blaske]

BTW, anyone from Digidesign is welcome to hop on this thread. [img]images/icons/wink.gif[/img]

We all know the testing takes time, but it wouldn't hurt to jump in with a conjecture (a disclaimer to the effect of "this is not the final word" would be OK).

Lee Blaske

[Kurtstudio]

I too have the exact setup and no problem running so far, I'm 2 weeks deep 16 hours a day [img]images/icons/grin.gif[/img]

[Lee Blaske]

What card order are you using? Are you using an expansion chassis?

If you followed Digi's recommendations, did DigiTest tell you your system was misconfigured?

Lee Blaske

[Knobturner]

Lee:
Every digitest I do checks out OK. Maybe your config is different than mine.
Food for thought.....if the slots ARE out of order, the ribbon cable that connects the digi cards won't work, because it would be impossible to go from the card in slot 5 to the one in slot 3. Now if we had a special cable made up.....or if we could get digi to say if it is mandatory to have the cards in the "core-farm-farm-sczi" order......
Digi: I know you're doing your own tests, but I bet some of us could be a lot of help to you if you would just communicate with us. I would just like to know if the slot order is absolute, with ANY computer model config, and what happens if it is NOT in the right order. Then we might could determine if the symptoms are similar.
Thanks
E
(Also posted on the "Quicksilver success/failure" thread)

[Knobturner]

Here's my config:

Quicksilver 867/CD-RW
Memory cache 512k
1.5 gig ram
OS 9.2.1
P.T. 5.1.1
stock AGP card/slot 1
Mix core/slot 2
Mix farms/slot 3-4
Atto UL3D/slot 5 set to 20/10 32 pci burst (when using PT)
Rorke Data dual Kingston hotswap
Seagate Cheetah drives
Sony AIT-1 tape drive
Griffen G-port
USB Imation Superdisk floppy drive

Lee:
Can you post your config?
Good luck
E

[ August 24, 2001: Message edited by: Knobturner ]

[ppine]

Knob- My take on the TDM Ribbon BUS cable is that all cards connect to it the same way. So the the cards do not care which node of the cable it is on. Take a close look at it, you will see that all nodes make the same common connections to each card.

FWIW [img]images/icons/smile.gif[/img]

[Knobturner]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:<HR>Originally posted by ppine:
Knob- My take on the TDM Ribbon BUS cable is that all cards connect to it the same way. So the the cards do not care which node of the cable it is on. Take a close look at it, you will see that all nodes make the same common connections to each card.

FWIW [img]images/icons/smile.gif[/img]<HR></BLOCKQUOTE>

Ppine:
Good point.
However, is it the slot order or the ribbon cable order or both that's important.? If it is both AND the PCI busses are out of order AND we switch around the cards, then we would need a special cable to keep the core card before the farms on the ribbon cable. I'd like to make sure it doesn't do any damage to the cards before I try any further tests. This is why we need to hear from Digi.
Tech..........Please?

Thanks,
E

[Lee Blaske]

Actually, the way the slots are out of order, proper sequencing could still be done.

Apple system profiler lists the order as 1, 3, 4,5, 2 - so, since slot number two is actually highest, the SCSI card would go there. Slot number 3 is lowest (after the AGP card) so the Mix core would go there, and then the other DSP cards would be sequential in slots 4 and 5.

As it is, even though DigiTest says my cards are misconfigured (haven't tried the new beta DigiTest yet, BTW) things seem to be working fine with all my TDM cards directly in the Mac. My expansion chassis just came back from getting revisions (they also found some short in the works) so I'll test that today.

Lee Blaske