I noticed that QueryPerformanceCounter uses a signed LARGE_INTEGER to store the time value. Isn''t this a little odd? When a wrap-around occurs does QPC start returning negative values or does it catch this and wrap to zero? I''d test it myself but I''m not leaving my PC on that long.

A LARGER_INTEGER is 64 bits. A signed large integer is between 2^63-1 and -(2^63), since you are excluding the sign bit. This is 9,223,372,036,854,775,807 and -9,223,372,036,854,775,808.

On my machine, the performance counter frequency is 1,680,440,000 ticks per second.

Dividing 9,223,372,036,854,775,807 by 1,680,440,000 gives 5,488,664,895 (that's max upper limit in total seconds for a performance counter).

5,488,664,895 seconds is 91,477,748.25 minutes or 1,524,629.1375 hours or 63,526.2140625 days or 173.9253 years.

I don't think you will be around in 173.9253 years, so your question is pretty much moot.

botman

[edited by - botman on July 2, 2003 8:59:35 AM]

Thanks...I''d neved actually done the math to see how often a wrap would occur. I randomly guessed maybe a month or so (well within possible limits for a PC to be turned on)! :O

Interestingly enough, after your example I checked my frequency and my counter runs at about 3.5 million per second, so it will only wrap roughly every eighty thousand years ! Hopefully I''ll have a new PC by then.