Hello everyone. I just have what I hope is a simple question. When measuring the speed of my machine (A p3 500 Mhz), I usually get 498 Mhz or something close like that. My question is this, which is more accurate, the 500 Mhz that my processor is rated for, or the speed I measured. If it's the rated amount, it should be pretty simple to round the numbers to the nearest multiple of 33 or 100, right? Then my timings (RDTSC) should be accurate down to a few 500 million-ths of a second, right? Thanks for any information guys.
Posted on 2002-07-26 02:16:34 by AlexEiffel
RDTSC has been very accurate here with small loops (0.001 sec.) and my CPU timing matches with the settings in the BIOS and on the motherboard. This is both on an Athlon 1.334Ghz and 550Mhz Celeron.
Posted on 2002-07-26 02:50:39 by bitRAKE
I have always found that the advertised speed of a cpu is marginally different from its actual speed. My celeron 500's clock at 498 as well. They are advertised as 500MHz simply because they are so close, and it sounds better from a marketing perspective.
Posted on 2002-07-26 05:42:10 by sluggy
it's to do with the clockgenerator and the accuracy of the one used on the motherboard... you could probably slap the cpu in another one one and get results above 500... you'll never get 100% accuracy to the rated speed...
Posted on 2002-07-26 06:18:24 by NervGaz
I thought that the engineering of the silicon crystal was so perfected that if they advertise 500mhz then that is exactly what it is. The discrepancy must be in the rest of the system or in the software/method of checking the speed.
My system time on my computer creeps forward about ten minutes a month. Now that is huge when you consider the time interval of 500,000,000 pulses per second. But then again someone told me on this board once that the line/power frequency to the box (i.e. 120 volts 60hz) may have something to do with it. If its not exactly 60 hz then the crystal will generate a different frquency. I'm not sure i believe this. Doesn't the crystal just use the voltage potential to get itself oscillating. Line frequency would then have nothing to do with it. I dunno. Anybody know about this for sure?

:alright:
Posted on 2002-07-26 11:17:08 by IwasTitan
Hmm, now I'm a bit confused. Does the CPU actually run at what it's rated, or does it differ by a few Mhz? Thanks for the replies so far guys, the effort is always appreciated. :)
Posted on 2002-07-26 17:21:04 by AlexEiffel
The CPU is tested to run at a certain speed and sold with a guarantee to that effect, but there is nothing that says it can't run faster or slower. There is an external clock circuit that drives the CPU, and if the CPU can handle the speed of that circuit then it will run at that speed. My Dual Celeron BP6 board has two 450Mhz CPUs, they are running at 550Mhz since the day I got it.
Posted on 2002-07-26 21:11:21 by bitRAKE
Thanks a lot bitRAKE, that answered my question perfectly. :alright:
Posted on 2002-07-26 23:46:49 by AlexEiffel
The crystal is powered from a regulated power supply IWT. If AC in is marginally off and you have a half way decent supply the the crystal will not notice it.

TV tuners are voltage controlled oscillators and I've seen poor AC in do a number on them.

Never seen it happen to a PC.
Posted on 2002-07-27 01:20:18 by alpha