Depends on the implementation, but most systems provide nanosecond accuracy using for example RDTSC - of course, that is still subjec to the problems described elsewhere which is that it CAN vary between CPU's. There are various things done in the system to keep them in sync, but there is still some chance that two or more processors run out of sync, thus leading to minor differences in timing.Originally Posted by cyberfish
--
Mats
Compilers can produce warnings - make the compiler programmers happy: Use them!
Please don't PM me for help - and no, I don't do help over instant messengers.
Does it mean the accuracy is about 1ns, but in 2 consecutive calls to gettimeofday(), the second one may return a time earlier than that returned by the first call?
the timer tick resolution on a win system can be increased by call to timeBegin(End)Period
http://msdn.microsoft.com/en-us/libr...13(VS.85).aspx
high precision timing in XP on multicore is a big issue (google will give a ton of references). If there is one argument for windows vista regarding multimedia applications/simulations/games and so one than it is it's support for timing via cpu independent HPET chips.
And of course, on really modern processors, the TSC is constant and synchronized across cores and sockets, so there's not the same need to worry about it. But for older processors, the HPET is a good way to get it. But there's been problems with HPET on some models of chipsets too, unfortunately.
--
Mats
Compilers can produce warnings - make the compiler programmers happy: Use them!
Please don't PM me for help - and no, I don't do help over instant messengers.
