why is rdtsc reporting wrong cpu cycles?

why is rdtsc reporting wrong cpu cycles?

[Marco Correia]

Hi,

I'm using the following function for getting the number of CPU cycles elapsed 
since I rebooted:

/* Read Timer Stamp Counter */
unsigned long long int rdtsc(void)
{
   unsigned long long int x;

   __asm__ volatile(".byte 0x0f,0x31" : "=A" (x));
   return x;
}

The problem is that the value I get dividing by my processor frequency (my 
computer is a 650Mhz toshiba satellite pro 4340) is giving me about 1 hour 
and I'm using the computer for about 4 fours.

Am I forgetting something obvious? I already googled around but I came to the 
conclusion that only new P4 M processors are able to change speed. 

I'm sorry if this is not the place to ask.

Thanks
Marco

ps: Please CC me as I'm not subscribed

-- 
Marco Correia <mvc@netcabo.pt>

Re: why is rdtsc reporting wrong cpu cycles?

[Willy Tarreau]

Hi Marco,

On Sat, Jan 03, 2004 at 03:48:39PM +0000, Marco Correia wrote:
 
> The problem is that the value I get dividing by my processor frequency (my 
> computer is a 650Mhz toshiba satellite pro 4340) is giving me about 1 hour 
> and I'm using the computer for about 4 fours.
> 
> Am I forgetting something obvious? I already googled around but I came to the 
> conclusion that only new P4 M processors are able to change speed. 

Perhaps it supports throttling. My mobile athlon supports this : 10000 times
a second, it is stopped for a certain amount of time, then resumed. Depending
on the throttling factor, I can even here it !

Cheers,
Willy