Sign in with
Sign up | Sign in
Your question

why does core 2 performs so well?

Last response: in CPUs
Share
August 26, 2006 5:10:05 PM

plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?

More about : core performs

a b à CPUs
August 26, 2006 5:19:07 PM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?


Lol... to make a long story short.. to understand the potential of Core 2, well one only needs to look at the architectural specs to understand that it is far more advanced then AMD K8/Intel Netburst. Intel has moved from a 31 stage pipelined architecture (Prescott) to a shorter 14 stage pipeline with the Core 2 architecture. In addition to just reverting to a shorter pipeline, Core 2 processors also have a 33% wider execution, deeper buffers, 4 wide-decode to execute, 4 wide-micro-op execution, micro and macro fusion and enhanced ALU’s for better overall instructions-per-clock performance. Then add to the fact that Core 2 can execute 128bit wide SSE, SSE2, SSE3 intructions in a single clock cycle (vs 2 clock cycles in the past).. this makes the architecture VERY VERY efficient.

So per MHz the Core 2 is the most efficient x86 architecture available.
August 26, 2006 5:29:37 PM

pipe lines, shared cache and the design of the prossessor.

in the same way you could have a 4L desiel engine which might only have a max BHP of 200 but you could get a 2l tuned petrol to pump out 300+ if its well designed. the 4 lieter might consume more fuel but that does not make it better.
Related resources
August 26, 2006 5:42:33 PM

Black magic and witchcraft.
August 26, 2006 5:52:53 PM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?


core 2 has super powers, p4s don't :lol: 
August 26, 2006 6:04:37 PM

Its a newer better design/a new generation....
August 26, 2006 6:05:01 PM

If the EE965 had a shared cache it would've performed faster too. I think one of the glitches in the 965 is having to use the fsb to synchronize the data in the separate 2MB caches thus not using it effectively enough to feed the cpu with data...or something like that+)
August 26, 2006 6:15:26 PM

The shared cache wouldn't have done much since it still had that ungodly long pipeline. That and the architecture left much to be desired.
a b à CPUs
August 26, 2006 6:22:45 PM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?


It's a new architecture which means it's ability to execute more instructions per clock cycle has improved --- significantly as it would appear.
http://www.intel.com/technology/architecture/coremicro/...

Edit:
I have gone through this a few times through the forums, it's ok to go through it again.

The computational performance that you and I observe, that is how fast does a software algorithm take to complete it's function, is not just the clockspeed of the processor alone. It is actually a function of how much work can be done in one clock tick and the speed of the clock ticks. Mathematically it can be expressed as:

Observed Computation Performance (instructions/sec) = IPC (instructions/clock tick) X Clock speed ( clock tick/sec)

For example, and these are just wild number out of the air to illustrate the concept, say the P4's have a horrid IPC, that is on average they can complete 1.5 instructions per clock tick, the AMD chips are good at IPC efficiency and complete, on average 3 clock ticks per second. So AMD chips could perform as well as a P4 at hald the clocks speed because they complete twice as many instructions in one duty cycle.

So here you go:

P4: Observed computational performance = 1.5 X clock speed.
K8: Observed computational performance = 3.0 X clock speed.
C2D: Observed computational performance = 4.5 X clock speed.

In short order one architecture can leap frog over the other without necessarily pushing clocks to the max. C2D> K8 > P4.

This was actually the reason for the in the difference in approach, Intel had always had success in process technology and getting clock speeds to go way up, AMD always lagged intel in process technology and to compete performance wise worked to design better IPC efficiency. As a result, Intel designed the P4 which was intended to allow clock speeds to scale very very high (which, of course, physics did not let them do) where as AMD focused on getting IPC up and released the k8 -- great product and great timing.

Jack

Lol.. I PM'd him that info..:p 
August 27, 2006 1:25:19 AM

Quote:
Black magic and witchcraft.

:lol: 



A real simplified explanation would be to say that every time that high speed P4 tries to process a load of data, it gets some of it wrong and has to redo it. The Core 2 duo gets a lot less wrong, so doesn't have to redo it.

Basically it just does the job quicker, despite moving slower.


But that's a really, really simplified explanation.
August 27, 2006 1:31:54 AM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?



Dont listen to any of these guys, they have no clue what they're talking about, especially that Jack guy and the Elmo dude.

Its the wee tiny leprechauns in the chip that make make Core 2 work so well :wink:
August 27, 2006 1:39:05 AM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?



Dont listen to any of these guys, they have no clue what they're talking about, especially that Jack guy and the Elmo dude.

Its the wee tiny leprechauns in the chip that make make Core 2 work so well :wink:

The leprechauns are only in C2Ds from Fab24 in Ireland. All other C2Ds have special hamsters (not the reported gerbils from trade magazines/websites) that were breed at 60nm, but with special shrinking properties to fit into the smaller spaces. It's those unique creatures that make the C2Ds more efficent than the mistake of using 75nm possums in the previous P4 CPUs.
August 27, 2006 2:18:28 AM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?



Dont listen to any of these guys, they have no clue what they're talking about, especially that Jack guy and the Elmo dude.

Its the wee tiny leprechauns in the chip that make make Core 2 work so well :wink:

The leprechauns are only in C2Ds from Fab24 in Ireland. All other C2Ds have special hamsters (not the reported gerbils from trade magazines/websites) that were breed at 60nm, but with special shrinking properties to fit into the smaller spaces. It's those unique creatures that make the C2Ds more efficent than the mistake of using 75nm possums in the previous P4 CPUs.

I thought it was the 65nm oompa-loompas?

That or the flux capacitor. My core 2 runs consumes 1.2 jigawatts @ 88 miles per hour!
August 27, 2006 2:27:55 AM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?



Dont listen to any of these guys, they have no clue what they're talking about, especially that Jack guy and the Elmo dude.

Its the wee tiny leprechauns in the chip that make make Core 2 work so well :wink:

The leprechauns are only in C2Ds from Fab24 in Ireland. All other C2Ds have special hamsters (not the reported gerbils from trade magazines/websites) that were breed at 60nm, but with special shrinking properties to fit into the smaller spaces. It's those unique creatures that make the C2Ds more efficent than the mistake of using 75nm possums in the previous P4 CPUs.

I thought it was the 65nm oompa-loompas?

That or the flux capacitor. My core 2 runs consumes 1.2 jigawatts @ 88 miles per hour!

I thought the oompa loompas were the AMD dual core solution.
August 27, 2006 4:00:25 AM

all your base are belong to conroe.
August 27, 2006 5:42:34 AM

excellent analogy! =)

Quote:
pipe lines, shared cache and the design of the prossessor.

in the same way you could have a 4L desiel engine which might only have a max BHP of 200 but you could get a 2l tuned petrol to pump out 300+ if its well designed. the 4 lieter might consume more fuel but that does not make it better.
August 27, 2006 5:55:50 AM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?



Dont listen to any of these guys, they have no clue what they're talking about, especially that Jack guy and the Elmo dude.

Its the wee tiny leprechauns in the chip that make make Core 2 work so well :wink:

The leprechauns are only in C2Ds from Fab24 in Ireland. All other C2Ds have special hamsters (not the reported gerbils from trade magazines/websites) that were breed at 60nm, but with special shrinking properties to fit into the smaller spaces. It's those unique creatures that make the C2Ds more efficent than the mistake of using 75nm possums in the previous P4 CPUs.

I thought it was the 65nm oompa-loompas?

That or the flux capacitor. My core 2 runs consumes 1.2 jigawatts @ 88 miles per hour!

I thought the oompa loompas were the AMD dual core solution.

Nah, AMD uses munchkins. Same as oompa-loompas, but they're not unionized so they can pay 'em less.
August 27, 2006 3:16:03 PM

Quote:
plz tell me why does Intel Core 2 Extreme X6800 with 65nm architecture, 4mb l2 cache, dual core 2.93 GHz and 1066mhz FSB outperform a seemingly better Inter Pentium Extreme 965 with 65nm architecture, 2x2mb l2 cache dual core at 3.73GHz and 1066mhz FSB ?


Mainly due to the shorter pipeline.

MHz/GHz dont measure processing speed, they measure the number of 'steps' the CPU takes per second.

Compare this with an olympic sprinter, lets say they take 10 massive strides in 10 seconds.

Lets now say that I shuffle along with 100 litle tiny steps in the same 10 second period.

If we measure speed in 'Steps/minute' then I seem 10x faster, but that doesnt consider all the factors - In actual fact the Olympic sprinter would cover a hell of alot more ground. (and not die of a cardiac trying to keep up, but thats another story)
!