Explain

[logox]

Can someone please explain (in depth) why the intel I5 2300 performs better than the AMD phenom ii x4 955 ? I have a friend that insisted an AMD phenom ii x4 955 would perform a slight bit better than the I5 2300 because the AMD was clocked slightly faster. I showed him some benchmark results and he saw the I5 won most the time. I don't know ENOUGH to explain why this is.

 

[Anonymous_26]

Because of the micro architecture. Sandy Bridges is a faster micro aritecture than the older Phenom II's.

 

[neon neophyte]

for a variety of reasons. one of which is the amount of calculations done per clock cycle

a good illustration of this point was the battle between amd and intels p4. the p4 clocked quite a bit higher but amd destroyed it all the same. it was the mhz race. the magic number that sells computers to people like your friend. it was false and most people quickly realized it

edit: you can rest assured that you are quite correct and your friend is wrong. youve already proven it so good on you. mhz are only directly comparable with 2 of the same cpu types. yes, more mhz is faster but its relative to the cpu itself, not other architectures

 

[esrever]

the efficiency of the cpu per clock cycle is better in the i5.

 

[vic20]

Most people make the mistake of thinking Megahertz is a standardized measure of power, like horsepower. It's not like that all.

It is a measure of cycles, and more cycles is better, but the cycles only mean something to another CPU of identical design because the work performed by different CPU designs is different. Its kind of similar to engine rpm. Up to a point, as you increase the rpms the engine can produce more power, but the rpms on a inline 6 BMW engine mean nothing to a 5.0 mustang V8. It could take 8000rpm in the BMW to produce the same horsepower in the Mustang at 6000rpm because the design is very different. In short, like CPUs cycles, the rpm number alone only means something to another identical engine.

Anyway back to CPUs. An Intel 3.0GHz i3-2xxx CPU is faster than a 2.0 i3-2xxx CPU, but not faster than a 2.9 i7-2xxx CPU because the i7 is a more powerful design. In this example, the i7 has same core design but double the cores and double the threads.

In case you are wondering, the number of cores of threads isn't comparable to another CPU design or brand either. AMD's old 6 core CPU and new 8 core generally perform worse than Intel's current 4 core models.

Unfornately its difficult to point out exactly what part of a CPU architecture makes it better, because on paper a more advanced looking design may end up performing worse.

In the end, the only you can do is show someone the results in real world scenarios and testing, just like with engines.

 

[logox]

Thanks... Also is there a diagram (or something close to it) of both processors ?

 

[logox]

Most people make the mistake of thinking Megahertz is a standardized measure of power, like horsepower. It's not like that all.

It is a measure of cycles, and more cycles is better, but the cycles only mean something to another CPU of identical design because the work performed by different CPU designs is different. Its kind of similar to engine rpm. Up to a point, as you increase the rpms the engine can produce more power, but the rpms on a inline 6 BMW engine mean nothing to a 5.0 mustang V8. It could take 8000rpm in the BMW to produce the same horsepower in the Mustang at 6000rpm because the design is very different. In short, like CPUs cycles, the rpm number alone only means something to another identical engine.

Anyway back to CPUs. An Intel 3.0GHz i3-2xxx CPU is faster than a 2.0 i3-2xxx CPU, but not faster than a 2.9 i7-2xxx CPU because the i7 is a more powerful design. In this example, the i7 has same core design but double the cores and double the threads.

In case you are wondering, the number of cores of threads isn't comparable to another CPU design or brand either. AMD's old 6 core CPU and new 8 core generally perform worse than Intel's current 4 core models.

Unfornately its difficult to point out exactly what part of a CPU architecture makes it better, because on paper a more advanced looking design may end up performing worse.

In the end, the only you can do is show someone the results in real world scenarios and testing, just like with engines.

VERY well explained!

 

[logox]

Best answer selected by logox.