PCI-GPIB performance slow on Windows XP?

[jeli]

I am working on moving ISA-GPIB+WinNT to PCI-GPIB+WinXP. GPIB card is connected to DMM to gather 1000 data points, testing application is the same. My testing steps are the followings: 1. Reboot computer with ISA-GPIB+WinNT and gather 1000 data points: 365 seconds 2. Reboot computer with PCI-GPIB+WinXP (NI488.2 version 2.6) and gather 1000 data points: 455 seconds 3. Reboot computer with PCI-GPIB+WinXP (NI488.2 version 2.6), open Measurement&Automation OR Ni spy, then close it, and gather 1000 data points: 305 seconds Only ibdev, ibclr, ibrd and ibwrt are called in my tesing program, I have two questions: 1. Why does PCI-GPIB+WinXP perform slower than ISA-GPIB+WinNT? 2. Why does PCI-GPIB+WinXP perform faster than ISA-GPIB+WinNT after Measurement&Automation OR Ni spy was run? Thanks, jeli

[JasonSmith]

This is an interesting situation you are explaining. Only one explanation stands out in my mind. That would be that with Windows XP you are able to log into the machine and start working long before the system has really finished starting up all of its services. This means that if you immediately run your GPIB application it is contending with services that are still starting up. I think that Windows NT did most of this work prior to giving you a login prompt (and just had fewer services), so you may not see the same contention on NT. The reason I could see starting NI-Spy or Measurement & Automation Explorer (MAX) appearing to fix things, is that they can take a while to load and cause you to wait before you run your app.

What happens if you restart the XP computer, then let is sit idle a few minutes before you start your test? You should wait until after the hard disk LED and the CPU meter have pretty much settled down, to indicate that your system is idle. I would like to know if this allows your test to run quickly, even without starting NI-Spy or MAX first.

-Jason S.

[jeli]

Thanks very much for your reply.

I've tested on three industrial computers and got the same result. It doesn't matter how long (even several hours) the test was started after computer/WinXP was reboot, the problem is still there. As I mentioned in my last message, the test procedure is very simple. I'd like to know what NI-SPY or MAX does, change some values in computer registery? or change some configurations of GPIB driver? or initialize something of the GPIB driver or GPIB board?

Thanks

Jeli

[Steven_T]

Hello,

We've tried quite a bit to reproduce your issue, but we have been unable to do so.  The problem that we are facing is that we are unable to ask for a capture.  If we had a capture, there is a possibility that we could find something, however opening Spy causes the problem to go away. 

Have you been able to reproduce these problems on non-industrial computers?  It would be nice to know what computers you have seen this problem and on what computers the performance is good.

Thanks,

Steven T.