-200621 bus bandwidth limitation

[GregJ]

I have two PC's, namely Dell GX270's.  They both have the exact same hardware in them (brand, and slot location in PCI bus, memory, cpu, hard drive etc.

Both PC's have the following:

2 NI-DIO32HS cards

1 RS232 Startech Serial Card

1 NI-GPIB card

I have DAQmx 8.3.0 installed in both PC's, as well as the GPIB software 488.2.

Both PC's are running Windows 2000, and have the exact same updates applied to them.

My problem is that one PC has a -200621 error (bus bandwidth limitation), and the other does not show any problems whatsover at any time.   However, if I change the location of the DIO32-HS card (swap it with my gpib card in the PCI slots), the newly swapped 32HS card will work temporarily, then fail after about a minute with the same error.

I am performing buffered digital I/O on this card.

Any suggestions on what I can try to resolve this problem would be greatly appreciated.  (I tried reinstalling the drivers already though)

Message Edited by GregJ on 03-19-2007 10:48 AM

[Nicholas_B]

Hi Greg,

For the PC that always gives the error, does re-arranging the 4 PCI cards have any effect? Are they in the same order as the machine that only gives the error when you swap the NI 6533 and the GPIB card?

When your computer boots up, it recognizes the PCI cards and allocates resources to them.  By placing the 6533 cards first, we might be able to portion more resources.  Do all the devices show up properly in the Device Manager as well?  The PCI bus is spec'd at 33MHz and 133MB/s.  Since this card has a maximum clock rate of 2MHz, there shouldnt be a problem with your buffered I/O application.  I'm also going to assume these machines have a sufficiently large amount of RAM and CPU speed... Can you describe this application in more detail, though?

Regards,

Nicholas B, National Instruments

[GregJ]

Hi Nicholas,

Thank you for the prompt response.   To answer your questions:

Re-arranging the 4 PCI-cards "seems" to have an effect sometimes, but inconsistently, and unrepeatably unfortunately.  I tried doing this about 10 times to see if I could resolve some sort of pattern, but the problem presented itself at random, almost immediately on all PCI slots of the Dell GX270, and once, it ran for about 2 minutes before generating the error (I was programmatically looping a finite sample).

To reduce  the possiblity that I'm breeching the PCI bus bandwidth limitation, I reduced my sampling rate and quantity of samples, to minimize my bus usage.  In addition, (to answer another question of yours), I was orgiinally using digital buffered input and output simultaneously, by connecting DIOA2 to DIOC3, and calling change detection on DIOC3.  The ports were configured as two 16 bit ports, (port A and B, using tradional NIDAQ terminology) A&B were grouped together, and C&D were grouped together.   I forced samples to occur on DIOC3, and issued basicaly a "clock" signal embedded in my digital output on the first group.  Anyway, I modified that completely, so that all I was doing is digital output (not input) only, to minimize bus usage as well.

To describe the PC's, they are both identical Dell GX270's, same DIO32HS cards in the same slots, as well as an RS232 card and NI GPIB card.  

Today, to update the world on this forum =), I swapped hard drives from the good PC with the "bad PC" drive.   The "good PC" running the "bad PC" hard drive worked!!! No errors!   This makes me believe that there is some sort of hardware issue with either the DIO32HS cards, or the motherboard.  I looked at the motherboards and there is a difference in the revision between the two!  One has Rev A00, and the other, has Rev A02.    My first hypothesis is that the revisions made to this motherboard relate to the problems I am having here.  I'm going to address Dell with the problem, but have you heard of anything similiar occuring, or could you suggest something else I could try?  I updated the BIOS and chipset drivers to the lateset that I could find on Dells website anyway.

The PC by the way, is a Pentium 4, with 512 RAM (DDR2 I believe).   Its the Dell model GX270 anyway (both of them are).

Message Edited by GregJ on 03-21-2007 10:29 AM

Message Edited by GregJ on 03-21-2007 10:30 AM

Message Edited by GregJ on 03-21-2007 10:31 AM

[GregJ]

I just checked both PC's, they actually have 1 gigabyte of RAM in each of them.

(couldn't edit my post on here again so had to add this reply)

[Nicholas_B]

Hi Greg,

I noticed that you have been working with another Applications Engineer on this issue through our email support system.  Jarrod and I discussed at length our ideas on the issue and determined that more troubleshooting has been performed through the email channel, so I am going to defer this support request to him.  This is the most efficient allocation of our support resources in Applications Engineering.  Please work with Jarrod through your email service request, and I will try  to continue to assist him with the issue.  Once we have determined a resolution, I will post the information to this thread for the benefit of our community.  Thank you and best of luck...

Regards,

Nicholas B

National Instruments