CVF 6.6 and Windows XP

[lienert]

Regarding my problem with Windows XP dying, which others suggested was a resource/memory problem (I have 512 MB RAM). Here is a simple program that will "kill" Windows XP when compiled and run using CVF 6.6. I invite other users to try it.

do while (.true.)
end do
end

Once it is running, try doing ANYTHING with the operating system, e.g., just using the editor to open another file. An Intel engineer on this forum suggested using call sleepqq(1000) inside the loop. This does indeed return the system responsiveness. But is it reasonable to insist that users must insert sleepqq in all their calculation loops? Each sleepqq call wastes 1 second, meaning that calculations will take forever! The solution to this problem is apparently "don't try and multi-task continuous computation". What?? There must be a more satisfactory solution to this serious problem. This is is not a multi-tasking compiler/OS by any reasonable person's definition!

[Intel_C_Intel]

I am writing this reply on XP whilst running your infinite loop program. The program slows the computer down (a little) but I am getting plenty of operating system priority to be able to continue working.

I can open and edit files, read web pages, etc, etc, with only minor delays.

All the best,
Eddie

[james1]

Works fine on my XP system. You could try setting the priority of the application manually to see what affect this has on your system, in case for some reason you are running the application at a higher priority the normal foreground applications.

James

[Steven_L_Intel1]

The advice about putting in a call to SLEEPQQ is specific to QuickWin applications that consist of a main program that kicks off the user interface and then just sits at the end waiting for the user to close the window. The way a lot of the samples are coded, there is an infinite loop at the end of the main program with nothing inside it. While this "works", it does diminish system response. So the recommendation is to insert a SLEEPQQ inside this "busy wait" loop, where it does no harm.

In the case of code that is doing active computation, you certainly don't want to be putting SLEEPQQ calls in. However, as I and others here note, Windows XP typically deals with CPU-bound computational code in a reasonable fashion.

I don't know what is causing the problem on your particular system. I don't see any evidence it is a compiler problem. Perhaps you have some other software running on the system that is causing a decrease in responsiveness as CPU load goes up. You might try booting into "safe mode" to see if the symptom persists - this will disable non-standard drivers and startup applications.

If there were somehow an endemic problem with the combination of CVF and Windows XP, we certainly would have heard about it by now, with examples, from hundreds if not thousands of customers. We haven't. Your message yesterday to vf-support was, I think, the first to even express such a complaint, and you provided no sample program for us to try out.

I wish I could explain and fix the problem you are seeing. But I can't - it does not appear to be a general CVF-related problem. Does Windows XP handle thread scheduling different from previous versions of Windows? Yes, we have seen evidence of that - but in the direction of being more responsive, not less.

I don't know what else to say here...

Steve