P4C800 Deuxe and "IRQ", I look for info.
Fogar
I have just made a new PC for editing.
I have put the Canopus M1 card in stot 5 and the RT card in slot 3.
Have I made a good thing?
What have I to disable in the BIOS so that I will not have problems
with IRQ?
Thanks Fogar
bye Fogar
Paul
Google lists IRQ trouble with the M1 card, so slot 5 for M1 is good,
as long as there is no card inserted in PCI slot 1.
Slot 3 shares with USB2 and SATA. So, it would be hard to disable
the USB2 and SATA, in order for Slot 3 to not share the IRQ. So,
slot 3 is not a good choice.
Slot 4 is easier to work with for the RT, as you only have to disable
the Firewire 1394 chip in order to have an unshared IRQ. You can
always buy and place a Firewire PCI card in another slot, if you
need firewire.
Slot 2 is also a good candidate for the RT, as you only have to disable
the onboard LAN chip, in order to get an unshared IRQ. You can
always buy and place a LAN card in another slot.
So, here as some configurations you could build:
Disable onboard Firewire 1394 chip. Buy Firewire PCI card for slot 2.
AGP Video Card
Slot 1 (Do not use this slot)
Slot 2 Firewire PCI plugin card (buy one if you need Firewire)
Slot 3
Slot 4 RT Card
Slot 5 M1 Card
Disable onboard LAN chip. Buy PCI LAN card for slot 4.
AGP Video Card
Slot 1 (Do not use this slot)
Slot 2 RT Card
Slot 3
Slot 4 PCI LAN Card (buy one if you need a LAN connection)
Slot 5 M1 Card
HTH,
Paul
Fogar
exhaustive answer.
I think I will choose the first way (slot 4 and 5).
But if I will buy a new Firewire PCI and if I wll put it in another
slot, is there some IRQ problems in the other slot?
I don't understand the technique of the PC but if I remember well the
the technician who has built my PC has told that he has disabled "LAN
control" in the BIOS.
Has he done an error?
thanks again.
Fogar
R. Pazderski
Slot (5) as Paul said, is Non Shared (Soft IRQ 21) as long as you leave Slot
(1) Vacant AND also have NO Wi-Fi device in the Special Wi-Fi Slot next to
Slot (5).
Slot (2) is a bit more confusing.
The manual shows Slot (2) Sharing a Soft IRQ with the On-Board LAN BUT the
On-Board LAN actually shares Soft IRQ (18) with the ATA On-Board Controller
(on my board) THUS leaving Slot (2) on Soft IRQ (22) as Non-Shared.
Slot (4) may also be made Non-Shared if you Disable 1394 Fire Wire Devices
in the Bios.
The statement concerning PCI Slots on page 2-16 of the manual refers to
physical interference in lieu of electrical interference.
Regards,
Richard (Rick) A. Pazderski Sr.
============== END MESSAGE ==============>>>>>
"Fogar" <alv...@tin.it> wrote in message
news:mknkkvgrm4830tlc9...@4ax.com...
Paul
> Thanks very, very much for your
> exhaustive answer.
> I think I will choose the first way (slot 4 and 5).
> But if I will buy a new Firewire PCI and if I wll put it in another
> slot, is there some IRQ problems in the other slot?
>
> I don't understand the technique of the PC but if I remember well the
> the technician who has built my PC has told that he has disabled "LAN
> control" in the BIOS.
> Has he done an error?
> thanks again.
>
> Fogar
>
If you buy a Firewire PCI card and place it in slot 2, then it will
share with the onboard LAN. It won't affect the M1 or the RT, so
everything should work fine. Firewire cards with TI.com chips on
them are supposed to be good, if you need to buy one.
As for the onboard LAN chip, the setting can be found in the BIOS here-
Advanced : Onboard_Devices_Configuration : Onboard_LAN
If it is disabled, then no harm is done, unless you want to use the LAN
connection, in which case you can set it to [Enabled] again.
Regarding Ricks post, to clarify what is currently being used
for IRQ numbers, the information can be found in the Device Manager.
Have a look at this picture, then maybe you could post the contents
of what your Device Manager entry shows. The following picture shows
the properties of the entry "Computer" in the Device Manager:
http://www.wown.info/w_baeten/gifbasic/pcinic12.gif
HTH,
Paul
Paul
<raph...@fayelectric.com> wrote:
> Fogar,
>
> Slot (5) as Paul said, is Non Shared (Soft IRQ 21) as long as you leave Slot
> (1) Vacant AND also have NO Wi-Fi device in the Special Wi-Fi Slot next to
> Slot (5).
>
> Slot (2) is a bit more confusing.
>
> The manual shows Slot (2) Sharing a Soft IRQ with the On-Board LAN BUT the
> On-Board LAN actually shares Soft IRQ (18) with the ATA On-Board Controller
> (on my board) THUS leaving Slot (2) on Soft IRQ (22) as Non-Shared.
>
> Slot (4) may also be made Non-Shared if you Disable 1394 Fire Wire Devices
> in the Bios.
>
> The statement concerning PCI Slots on page 2-16 of the manual refers to
> physical interference in lieu of electrical interference.
>
> Regards,
>
> Richard (Rick) A. Pazderski Sr.
I have to admit, Rick, that when I give advice on this subject, I
haven't the foggiest clue exactly what I'm fixing.
Again tonight, I tried to trace how interrupts work in a Microsoft
OS, and I cannot seem to find all the bits and pieces necessary to
work it through.
At the PCI bus level, there are hardware wires that devices share.
In the case of the 875 chipset, the PIRQ[A-H] signals on the ICH5
are connected to the PCI bus, but the other entries Asus shows
in the manual, are for internal logic blocks that are also tied
to those signals (but probably not physically).
(ICH5 manual - Page 131 shows hard wiring of internal resources to PIRQ)
http://developer.intel.com/design/chipsets/datashts/25251601.pdf
The PCI spec lists an interrupt acknowledge cycle, that when a PCI
card is addressed, the card will spit out an 8 bit value, which is
presumably used somehow by the processor, as part of the response to
an interrupt service routine. I presume the "routing table" that the
OS builds during boot, is used to determine which hardware devices to
poll, for an interrupt signal that is shared (but I'm guessing at this,
because I cannot find a description of the process).
The ACPI spec defines a bunch of "methods" for handling various
aspects of interrupts, but it doesn't go into details (more of an
architecture spec).
So, unless the only problem with sharing, is one of excessive interrupt
service latency (i.e. when an IRQ number is shared, the interrupt
service routine is taking too long to find the actual device doing
the interrupting), I cannot see a reason for sharing to screw up.
The interrupt servicing hardware even has a null vector, for interrupt
requests that are too short, so the system is protected from
spurious interrupts.
So, for the sake of completeness, could you post the contents of
the "Computer" entry of your Device Manager? This page shows an
example:
http://www.wown.info/w_baeten/gifbasic/pcinic12.gif
There have been several motherboards, where the IRQ table in the
manual doesn't match the assignments seen in the Device Manager.
Now, either this is a measure of the incompetence of Asus tech
writers, or it implies that the assignments can be changed.
I was thinking, perhaps, that if you have set the BIOS entry
"PNP OS" to [Yes], that the OS could reassign IRQs, and that is
why the LAN entry is moved. Normally, people use [No], so that
the BIOS defines the IRQ table, and the OS is not supposed to
juggle standard stuff.
And, if you happen to know of any web pages or docs that describe
how an interrupt is processed, maybe I can finally understand what
is so evil about sharing.
Paul
Paul
<jae...@jaeger.com> wrote:
> In article <nospam-2508...@192.168.1.177>, nos...@needed.com
> says...
>
> > And, if you happen to know of any web pages or docs that describe
> > how an interrupt is processed, maybe I can finally understand what
> > is so evil about sharing.
>
> In short, nothing. The only time it might be an issue is where latency
> is a huge factor, such as with any sort of ASIO audio hardware. I'm not
> exactly sure what it is you're asking...One of the purposes of an APIC
> is to allow the OS to build the routing table itself rather than just
> mirroring the BIOS, but as you noted the control only goes so far; a
> shared IRQ will always be shared. The reality is that in 2003 any
> hardware that can't share an IRQ belongs in the trash can.
I'm looking for failure mechanisms. In other words, can a plugin card
really screw up interrupts - is it even possible ? With the stuff I've
read so far, there doesn't appear to be a mechanism where the card
can do it.
So, I assume the "legend of sharing" has to do totally with
interrupt service latency - and in this case, the software Fogar is
using does a check to see whether other hardware is sharing the IRQ.
That is why I've tried to give him a couple of hardware configs that
don't share for both the cards in the two card set, so the software
will stop complaining about the hardware config. The hardware is a
DVRex RT, an expensive real time video editor.
Paul
R. Pazderski
I appreciate your re-post and the link to the Intel ICH5 Hub Manual. Have to
admit that is 200 Pages plus of heavy electrical engineering that takes some
time to wade through. I tried to save it for reference but it appears to be
in a locked format so a desk top link will have to suffice.
Quickly let me explain how I determine the best solution for my systems on
my network. It is in no way the best solution for others but may be of
benefit. I'll try not to be too technical as practicality must be employed
as your very good last post implies.
*** Also this walk through is for a Win2K System but may be generally
applied to most all systems ******
1) Assume that Critical PCI Card Devices will ask for the PCI bus more
often, hold the bus longer and will need more band width than non-critical
device cards. Some PCI Devices such as H/W Faxmodem cards will not need the
bus until used but when used may become critical devices. The builder must
use good judgment in such cases as to the main function of their machine and
what really is a critical PCI device for the machine's intended purpose.
Obliviously, in a system built for the audiophile, the PCI Sound Card
becomes a critical device.
2) Building on assumption 1 above, it follows that Critical PCI Cards such
as Wi-Fi, Raid Controllers, Sound Cards, etc. must be in non-shared or less
shared interrupt slots to Best Perform as intended without interference.
With the new boards, this goal is becoming increasing difficult as there are
so many On-Board devices which have to also share a limited number of hard
interrupts, signal timings, etc, with the PCI slots.
3) As a first pass, I look at the Sharing Scheme as depicted in the manual
and insert the PCI cards into the slots per assumptions 1 & 2 above.
3) Not knowing exactly how the O/S HAL (Win2K-SP4) is going to handle the
hardware, I (ENABLE) every device and set O/S P&P to (NO) in the Bios.
EXCEPTION: FOR WIN2K, (DISABLE) HT before the installation or one will
install an MPU Setup with the HT processor. There is an easy fix if one
forgets but there is no use to incur this extra problem.
4) I then Clean Install the O/S ensuring that ALL H/W drivers are F6
preinstalled or installed after load-in according to normal procedure. There
should be NO device with a (Yellow Question) or (Red not installed).
5) After the first pass installation, I go into Device Manager and observe
what devices are actually sharing what IRQs. The assumption here is that the
machine slot IRQs are fixed as per the ICH5 HUB manual, Asus design scheme
and that the O/S is managing the system properly (IF) all O/S service packs
/ quick fixes are up to date.
*************As Paul indicates in his post, the manual tables are (NOT)
always correct and this step is where one finds what PCI devices are really
sharing what IRQs.
*************A Slipstream Installation is required in some cases. (For later
discussion)
6) Upon first pass examination of Device Manager, PCI cards may have to be
relocated, not needed devices DISABLED to obtain optimum performance per
assumptions 1 & 2 above and another Clean Install performed from scratch.
7) In no case should one just rearrange the PCI cards, use the original O/S
installation and not perform another Clean Install as the registry will
become a total mess with problems occurring down the road.
8) Lastly, after ascertaining which slots will always have which IRQs,
annotate the manual correctly if need be and continue with a good
installation.
As a note, most new PCI cards can readily share IRQs but it is my opinion to
take a little time to ensure optimum performance. After all, most have put
hard earned money into their equipment and take pride in being called a
Computer Builder.
Paul, will do Screen shot for you tomorrow as the hour is late.
Best Regards,
Richard (Rick) A. Pazderski Sr.
============== END MESSAGE ==============>>>>>
"Paul" <nos...@needed.com> wrote in message
news:nospam-2508...@192.168.1.177...
Fogar
I write to you again because someone told me this:
"If you have the ASUS P4C800-Deluxe and NOT the P4C800-E Deluxe then
you are cursed with the 3Com on-board NIC that shares with PCI slot 2.
If you disable the LAN then the M1 should go into PCI slot 2 and the
RT/MPEG in PCI 3 (shares with USB2). Thats where mine are on the
P4C800-E Deluxe and so far so good but I have not connected a USB
device yet and I have not put the DVRexRT into heavy load yet."
Is it the true? I'd like to tell you I don't remember if I have ASUS
P4C800-Deluxe or the P4C800-E Deluxe.
I have definitely put DVrex in slot 5 and Rt card in slot 2.
I have disabled "Onbord LAN". The slot 1 is free but when I make
editing, always I have problems. RexEdit very often jams and when I
open an avi file with RexVideo it jams.
I have unistalled RexEdit and its ini files and istalled it again but
the problem is the same.
Do I still have to disable anything?
Have I to replace Win XP with Win 2000?
yhanks again Fogar