Ok I think I understand the ratio think where basically you can push the cpu clock higher thus giving higher clock speed, but keep the memory bus lower so the system is stable.
I've allways thought memory and cpu bus speed in sync was the best because no clock cycles are wasted.
But for example I see some people running their P4 2.4c at 300fsb at 3:2 which gives a clock speed of 12x300 = 3.6ghz cpu with memory at 400mhz
And others are running 225fsb at 1:1 which gives a 2.7 ghz cpu speed, but in sync with 450mhz memory.
These numbers may not be exact but you get the idea. At 1:1 you get lower cpu clock but higher memory mhz and sync.
At 5:4 or even higher 3:2 you get higher cpu clock, but slower memory that is out of sync.
Once you get a stable machine at each of these settings how do you decide which is the best performance for your application? For example I do MPEG1 video encoding that runs the cpu at 100% for hours. Do I need more memory FSB or higher CPU or what?
I'm pretty confused and don't know how to apply these things to achieve the best real world performance.
Thanks Guys!
I've allways thought memory and cpu bus speed in sync was the best because no clock cycles are wasted.
But for example I see some people running their P4 2.4c at 300fsb at 3:2 which gives a clock speed of 12x300 = 3.6ghz cpu with memory at 400mhz
And others are running 225fsb at 1:1 which gives a 2.7 ghz cpu speed, but in sync with 450mhz memory.
These numbers may not be exact but you get the idea. At 1:1 you get lower cpu clock but higher memory mhz and sync.
At 5:4 or even higher 3:2 you get higher cpu clock, but slower memory that is out of sync.
Once you get a stable machine at each of these settings how do you decide which is the best performance for your application? For example I do MPEG1 video encoding that runs the cpu at 100% for hours. Do I need more memory FSB or higher CPU or what?
I'm pretty confused and don't know how to apply these things to achieve the best real world performance.
Thanks Guys!