FIFO overflow using PCIE 1429 on a windows 7 64bit system

[odedbd]

Hi all,

For a few years we've been using a PCIE 1429 framegrabber in conjunction with a VDS CMC1300-485N fast camera on a windows XP system, and it worked well most of the time.

We have now upgraded our lab computers to Windows 7 professional 64bit and have run into a FIFO overflow problem with the image aquisition.

The problem we have is as follows -

when setting number of lines (in the acquisition parameters panel only, not in the camera attributes) with up to 22 lines with the full 1280 pixels per line, everything works fine. Increasing the number of lines to 23 or higher causes an immediate FIFO Overflow error message and acquitision halts. The exact same behavior is realized when runnig acuisition from our IMAQ written stand alone VC concolse application. On our old system we had no problem achieving the full 1280X1024 acquisition at the full 485fps rate.

We have installed the latest MAX 4.7.2 with IMAQ 4.5.0. The frame grabber was installed without a glitch (MAX recognized it) and MAX reports that it managed to negotiate the required X4 PCIe lane width. We are using the same .dat and .icd files we have been using for our camera all this time (VDS don't have any updated versions on their website).

What should we do? Is there a known issue with the 1429 card on Windows 7 64bit machines? I did not manage to find anything on ni.com regarding such a problem.

Help please...

[odedbd]

update:

following an old discussion in the forum I found this suggestion-

• Disable "Runtime Power Management"
• Disable "SATA Power Management"
• Change "Idle Power Savings" from Extended to Normal

This improved the situation quite a bit, now the grabgber will capture for a few seconds in full frame mode 1280X1024 before issuing the FIFO overflow error. reducing the number of lines seems to prolong the time it takes for FIFo overflow (for 100 lines it's more than a few seconds, maybe 10-15sec).

I hope this information will help in finding the solution

[Brad]

One thing to make sure is that the 1429 negotiated a x4 PCIe bus with the OS. You can see what is negotiated by going to MAX, clicking on the 1429 and the right portion of the MAX window should display the negotiated bandwidth. If this is x4, here is a KB that describes other things to try:

http://digital.ni.com/public.nsf/allkb/32F3F5E8B65E9710862573D70075EED1

Not all motherboards handle PCIe devices the same. So even though your new Windows 7 machine may have better specs than your older machine, PCIe x4 devices may not get as much bandwidth (your new machine may have more peripherals that take away some of the bandwith or more processors that take away some of the available resources). The PCIe-1433 was released to reduce these kinds of problems by having a much larger FIFO on board (1429 had 18kB and 1433 has 512MB) to reduce the chances that the FIFO doesn't get serviced fast enough by the motherboard/OS. Unfortunately there's not a good way to know which computers will handle that much data well vs. others, but the 1433 does a good job of handling this so you don't have to worry about changing motherboards/computers.

Hope this helps,

Brad

[odedbd]

Hi Brad,

Thanks for the reply. As I've written in my original post, "MAX reports that it managed to negotiate the required X4 PCIe lane width", so that is not the issue. We have three 1429 cards in our lab that we would very much like to keep using (as they meet the specs of our camers nicely), so buying three new 1433 cards as replacement is a little costly for what should have been a performance upgrade.

Is there somewhere I may find a list of 1429-compatible motherboards/pc-assemblies  that are currently on market? It'd be cheaper to replace the computer than to buy the new card, I guess.

Thanks,

o

[Brad]

I don't have a list, but I'll get one of the application engineers in touch with you...they may have a list or at least recommendations.

Thanks.

[BlueCheese]

Another suggestion to try would be disabling the Aero interface in Windows 7. Depending on the graphics chipset, it may require lots of memory bandwidth from the system to update the display when Aero is enabled.

Eric

[odedbd]

Thanks, that sounds like an interesting suggestion. I will check it Sunday morning and see how it goes.

[Kristoff2]

odedbd,

Please let us know if disabling the Aero interface in Windows 7 helps.  If not I can look into motherboard requirements for the 1429 or other possible troubleshooting steps for you.

Regards,

Sam K

Applications Engineer

National Instruments

www.ni.com/support

[odedbd]

Sam,

Will do that first thing sunday morning (Israel time), wil update here.

Thanks!

[odedbd]

Hi Sam,

I've disabled all of windows 7 eye candy (set to maximum performance) and while it improved the situation (gave me a few more seconds of aquisition before halting on FIFO overflow) it did not solv ethe problem. Increasing the NI MAX process priorit in the task manager to "high" gave yet another improvement (it now sometimes only halts after ~15 seconds), but this is still far from enough.

I would be very grateful for further suggestsions, as well as for a list of current motherboards that are known to be compatible with the 1429 card.

Thanks,

o