Tech Support banner

Status
Not open for further replies.
1 - 8 of 8 Posts

·
Registered
Joined
·
6 Posts
Discussion Starter #1 (Edited)
Hi, I'm new here, and it looks like I'm not alone with my Promise problems...

Since May 2004, I've got an ASUS P4C800-E deluxe with 4 HDs and 1 DVD burner attached to the following channels:
* SATA1: WD740GD (Raptor) = system drive
* SATA2: unused
* PRI_IDE: NEC DVD_RW ND-2500A
* SEC_IDE: Maxtor 7Y250 P0
* SATA_RAID1: Maxtor 7Y250 M0 \ / Raid 0 Array
* SATA_RAID2: Maxtor 7Y250 M0 / \ on Promise 20378

Besides my Radeon 9800pro, there is no PCI card inserted, only mainboard features are used.

Since long, FastTrak sporadically reports IRQ problems und then hangs, complaining about being unable to detect proper interrupts. Initially, this happened just sporadically, but since more recently this is nearly standard. Pressing the reset button sometimes helps, but normally I have to disconnect power physically and wait a minute or two before starting again. If the systems happens to come up, there's a lot of bluescreens (several per hour).

Yesterday I tried to dig more into the IRQ stuff (haven't touched that since I grabbed this PC from my dealer): All influencible IRQs were set to "available for PCI": 3,4,5,7,9,10,11,14,15. Since I read somewhere that Promise 20378 can't share IRQ's 14 and 15, I decided to clean up the IRQs completely:
* disable COM1, COM2 and the parallel port (freeing up IRQ 3,4 and 7)
* reduce the number of USB connections from 8 to 4
* declare those IRQs "reserved" which are used by other (ISA) devices:
9 (ACPI compliant system)
14 (Primary IDE)
15 (Secondary IDE)
(all others are left "available")

Since I did that, I have the feeling that the conflicts happen less often, but they definitely did not disappear completely. I should mention that in cases where the system does come up, the 20378 finds itself using IRQ 23, which is always shared with "standard extended PCI-to-USB universal host controller" (this is my own translation of the German entry in XP's device manager).


Here's now my question: Is there a way to force the 20378 to an IRQ it can use *exclusively* ? I should now have plenty of these (IRQ4, 5, 7, 10, 11). Or move the USB controller away from IRQ 23 ?

Can anybody help me ? Unfortunately, my dealer can't - he's gone bancrupt meanwhile...


Thx
 

·
Guru
Joined
·
1,315 Posts
Hi bitcruncher, and welcome to the forum!

I could be wrong but I do not think you can manually force IRQ settings on this mobo. I run almost the same configuration on a system here, and never have this problem. Your problem sounds like it might be something else, and the interrupt error message might be just a coincidence.

Thanks for all the hardware attachment info, that is very helpful. I see you have a lot of devices attached to your PC. But if the power supply is not strong enough you could have these kinds of sporadic failures. Memory can cause it also, but I am thinking supply, so
??please tell us about your power supply: the wattage, and also the brand & model number if you know it? Also what CPU you have installed?

About the IRQ's, you should not need to force any IRQ settings with this mobo; everything running as defaults should work fine. I will go out on a limb here and say that all properly-written drivers today can share IRQ's. If you expect a lot of intensive interrupt activity on specific devices, such as Firewire it can sometimes make sense to try to move add-on cards to different slots to distribute your load. But in your case, with no addon devices, IMO there is little you can do.

Regarding IRQ 23 and the sharing you saw: Besides the standard 16 interrupt lines numbered 0-15, the ICH5R Southbridge chip in this mobo offers 8 more lines, lettered A-H. Since most O/S's deal only in numbers, you see this as 16-23. If you look in your mobo manual section 2.6.2 there is a helpful table called "Interrupt assignments for this motherboard". Here in column H you can see that the "Onboard USR 2.0 Controller" indeed shares H with the "Onboard SATA (optional)" which is your Promise chip, and H is where that number comes from. It also shares with PCI slot 3 (the middle slot), but as you said, there are no addon cards. Either way, some sharing is necessary because no Southbridge chip provides enough lines for every device to have its own private interrupt. One reason I like this mobo is for slots 1-5 which share only between themselves.


-clintfan
 

·
Registered
Joined
·
6 Posts
Discussion Starter #3 (Edited)
clintfan,

first of all thanks a lot for taking the time for such an excellent analysis of my problems. I'll try to give you some more information below, as you requested:

I see you have a lot of devices attached to your PC.
Yeah, because I use it for video editing (but not only).

But if the power supply is not strong enough you could have these kinds of sporadic failures. Please tell us about your power supply: the wattage, and also the brand & model number if you know it?
The power supply is a Silentmaxx proSilence DN 420 LNC. When I bought the PC (with only 3 HDs at that time) I told the dealer that I'm planning to add plenty of further HDs lateron and therefore requested 550W, but he assured me that 420W is more than enough.

Also what CPU you have installed?
P4 (Northwood) 3.2GHz

Memory can cause it also
This may be an issue. I asked for a slightly overclockable RAM, but the dealer proposed (and sold me) 2x512MB TwinMos Twister PC400, again argueing that this would be a best buy for my purposes. When I had that PC back home, I of course was curious about overclocking and carefully started playing around with some BIOS setting for that. The result was that *no* overclocking at all was possible. Consequently, I stayed away from any overclocking setting from that moment. Nevertheless however, one of the sporadic issues I'm facing about every couple of weeks is that BIOS during boot-up reports an overclocking failure (although I'm not doing any). When I then enter BIOS and leave it without any change, boot-up normally works.

About the IRQ's, you should not need to force any IRQ settings with this mobo; everything running as defaults should work fine. I will go out on a limb here and say that all properly-written drivers today can share IRQ's.
I think I agree to you - in particular since the only two devices sharing IRQ23 are onboard devices. And I chose the somewhat more expensive ASUS product because of its stability reputation.

If you expect a lot of intensive interrupt activity on specific devices, such as Firewire it can sometimes make sense to try to move add-on cards to different slots to distribute your load.
Indeed I'm using Firewire for video capturing, but that again is onboard and only used during capturing. I haven't done capturing since months (because of my wife having adjusted my priorities slightly :heartlove), but I still have these problems. And the IRQ conflict normally happens during first HD detection - well before XP and Firewire even dream of starting.

-----------

I should mention that since about 3 days I have disabled the Promise controller in BIOS (I only need those 2 drives for video editing and currently have no time for that). Since then I think I had no problem at all. These 2 drives are nevertheless still running - so I don't think that the power supply has an easier life now.

Coming back to the sporadic overclocking issue, I think - and you raised my attention to this - something must still be running at it's very limit. Would it be a good approach to start playing around with some "underclocking" ? I wouldn't mind to run that beast with only 2.8G or even less for a couple of weeks, if that helps isolating the fault. Or should I start with modifying the RAM settings first? Could you (or someone else) tell me what's the best strategy for this exercise ? We have to bear in mind that verification of one changed parameter may need several days of stable operation before I can be sure to be on the right track. So I think a good strategy is crucial.
 

·
Registered
Joined
·
6 Posts
Discussion Starter #4 (Edited)
bitcruncher said:
Would it be a good approach to start playing around with some "underclocking" ?
Hi again - I was so thrilled by my own idea :laugh: that I couldn't wait for any strategic advices to come. So I tried already:
  • changed the CPU external frequency from 200 to 150 MHz (=2.4 GHz CPU clock) => FastTrak couldn't find the drives 2 times. Only complete power-off helped so that I could get in to the BIOS at least. Restored value back to 200 MHz immediately.
  • set control of DRAM parameters from SPD to manual. Changed DRAM CAS# latency from 2.5 to 3.0. All other values were already set to their maximum value, so I didn't change them:
    RAS# Precharge = 4
    RAS# to CAS# delay = 4
    Precharge delay = 8
    DRAM burst length = 8

With this modified CAS# latency setting, the system booted a few times ok (although I know it can take days before the problem happens). I'll keep you posted on further results.
 

·
Registered
Joined
·
6 Posts
Discussion Starter #5 (Edited)
... and here are some other results:
  • reducing the external DRAM frequency from 400MHz to even 266 MHz does not help (FastTrack hangs completely several times)
  • disconnecting unused drives (7Y250P0 and ND2500A) does not help. (+5V Voltage changes from +5.12V to +5.16V). I know that doesn't tell anything about the dynamic behaviour, but I don't have a scope to check that thoroughly
  • BIOS > Jumperfree Config > Performance Mode: changed from [Auto] to [Standard] seems to help ! So far, I had 10 restarts without a problem. Does anyone know which parameters are affected by this setting, so that I may conclude on where the weak link may be ?

However, I haven't yet mentioned one strange behaviour. I hear odd beep sequences (from the built-in speaker) at different stages of the boot process. It's a repeating sequence of 3 frequencies f1, f2, f3, starting with the lowest (f1) up to the highest (f3). Each tone lasts maybe 100ms, because the entire sequence (normally 3 times f1+f2+f3) is finished after about 1 sec. Are this the AMI BIOS beep codes described in the manual (ch. 3.1) ? But the strange thing is: If I count right, in most cases I hear those 9 tones, but the manual lists only 8 or 10. And sometimes these tones are loud and clear while sometimes being very weak.

What the heck is that all about ?
 

·
Registered
Joined
·
6 Posts
Discussion Starter #6
yes, "Performance Mode" (not to be confused with Performance Acceleration Mode) definitely was the issue. Since I set that to "Standard" rather than Auto, things look much nicer now. Since then I'm able to overclock the CPU by 14.5% while DRAM is still running with 2.0-3-3-8 (without tweaking any voltages). I also had no single IRQ conflict / drive detection failure even at the limits of overclocking.

I don't think that I'm converting into the overclocker's camp now, but it's nice to see how large a margin my system has all of a sudden.

clintfan, thanks for your help. Your ideas kicked me into the right direction.

Regards bitcruncher.
 

·
Guru
Joined
·
1,315 Posts
Nice job of problem analysis. Glad you got it stabilized.

Sorry, nobody knows what internal settings Asus decides to change between Performance Mode= "Auto" and "Standard". Here's hoping your system remains stable under the "Standard" setting. I would not change any other fields on that screen though, if you are working well.

-clintfan
 

·
Registered
Joined
·
6 Posts
Discussion Starter #8
I googled "performance mode" meanwhile and found pretty much what you mentioned above: It's a mystery to everyone, many have problems with that, and the only advice you can get is "If Turbo works - fine, use it. If not - use Standard". So I'll follow the latter and stay with Standard.

Thanks for your help, clintfan.
 
1 - 8 of 8 Posts
Status
Not open for further replies.
Top