Pentium D 960 @4.16 GHz vs QUAD CORE Q6700 @ 2.66 GHz in games

[TocM]

I have now a Pentium D 960 clocked at 4.16 GHz.
I'm looking for a new quad core CPU. The Q6700 or the Q6600 are interesting. But are these two CPU's better in game performance than an overclocked Pentium D 960 at 4.16 Ghz. The Q6700 runs at 2.66 Ghz and the Q6600 at 2.4 GHz. That's a big different with my D 960.

 

[roadrunner197069]

LoL are you serious? Why dont you check what your cpu scores in 3dmark06 and then check what a stock q6600 or q6700 scores. Besides your OC what you got so odds are you wont be running a quad under 3.5.

 

[jsc]

Yes. The C2 architecture is so much better than the Pentium D that there's no comparison.

Most of posters replying, however, will tell you that since most games still do not support C2Q, you will get more gaming performance out of something like an E8400 which can also reach 4 GHz. Based on some benchmarks I saw in an issue of PC Pro (British computer magazine) when the C2D's first came out, an OC'd E8400 is at least 1 1/2 to 2 times faster than a D960.

 

[WR2]

This should give you a general idea. You should be able to estimate the correct scaling of a 4.1Ghz D960.

What you didnt mention was what games are imporant to you, video card and monitor resolution you game at [:wr2:4]
For example for if you have a NV7950GT or HD2900Pro and a 20" 16x12 res monitor you most games will be GPU bound and you'll get the same performance between a stock D960, your OC'd CPU and a Q6700 (QX6700 in the chart).
On the other hand if you game @ 1280x1024 and play UT2004 that chart could be accurate [:wr2:5]

 

[TocM]

This should give you a general idea. You should be able to estimate the correct scaling of a 4.1Ghz D960.http://img527.imageshack.us/img527/7006/21tl1.jpg
What you didnt mention was what games are imporant to you, video card and monitor resolution you game at [:wr2:4]
For example for if you have a NV7950GT or HD2900Pro and a 20" 16x12 res monitor you most games will be GPU bound and you'll get the same performance between a stock D960, your OC'd CPU and a Q6700 (QX6700 in the chart).
On the other hand if you game @ 1280x1024 and play UT2004 that chart could be accurate [:wr2:5]

I have an Nvidia Geforce 9800 GTX and i play mostly @ 2560x1024 (2 screens) or 1280x1024 in games like Crysis, Lost Planet, ...

 

[WR2]

You're not making this easy - are you? [:wr2:5]
Harder to guess-timate how much better a stock 2.66Ghz Q6700 will be in those different conditions.
With your 15% overclock on the D960 and the Q6700's ability to take an easy 25% overclock I'd make a SWAG that the chart above would hold up in terms of % improvement (rather than FPS) for 12x10 gaming. At 25x10 you'll be GPU bound and maybe not see a major improvement.

 

[grieve]

At 25x10 you'll be GPU bound and maybe not see a major improvement.

I agree.

Generally speaking however the quad will be a fair amount faster then a Pentium D

 

[sailer]

I think an overclocked Q6700 would wax the floor with a Pentium D. The newer and more complex games would make the difference even greater. As already mentioned, an overclocked e8400 would probably make an even greater difference. It should be remembered that its not just the clock speed, but the amount of work done per clock. I personally prefer quads, as I think the coming games will take more advantage of the multiple cores than the present games do, but that's an opinion and nothing more. Of course, if a person does lots of multitasking, then a quad gives good advantage.

 

[USA Mike]

I agree.

Generally speaking however the quad will be a fair amount faster then a Pentium D

i don't think so guy sorry !
a simple Quad@2.6 is NOT better than a D960@3.6 in a MONOTHREAT task.

If the application/game is not build to run faster on multithreads/multicore hardware, THAN the clock speed still the main indice to compare them and not the cpu architecture.
So in MONO, D960@3.6 runs faster than Quad@2.6

BUT

in a multithread application (like cinebench) the clock speed isn't a right indice.
So MULTITHREAD, D960@3.6 IS VERY SLOWER than Quad@2.6

have a good day !

 

[shadowthor]

I don't agree with the above post, the Quad with the better architecture will be better than the Pentium D as well as the amount of L2 cache that's in the quad, that most games use. As we remember clockspeed doesn't mean everything, its how fast they operate at their clockspeed.

 

[dagger]

I don't agree with the above post, the Quad with the better architecture will be better than the Pentium D as well as the amount of L2 cache that's in the quad, that most games use. As we remember clockspeed doesn't mean everything, its how fast they operate at their clockspeed.

Lol, yeah. Newer quads can literally execute 4 times the instructions per cycle per core compared to Pentium D. 1ghz = 4ghz. Unless the Pentium D gets chocked on low cache, in which case the difference will be even bigger.

Not sure if anyone remembers the legendary single core old type celerons that goes to 5ghz on air. Ghz is everything!

 

[Evilonigiri]

Not sure if anyone remembers the legendary single core old type celerons that goes to 5ghz on air. Ghz was everything!

Fixed.

 

[dagger]

Fixed.

Meh, it's not funny anymore.

 

[Evilonigiri]

I didn't intend it to be funny...

It's the truth.

 

[dagger]

I didn't intend it to be funny...

It's the truth.

You ruined the joke.

 

[Evilonigiri]

...what joke?

 

[shadowthor]

lmao

 

[B-Unit]

i don't think so guy sorry !
a simple Quad@2.6 is NOT better than a D960@3.6 in a MONOTHREAT task.

If the application/game is not build to run faster on multithreads/multicore hardware, THAN the clock speed still the main indice to compare them and not the cpu architecture.
So in MONO, D960@3.6 runs faster than Quad@2.6

BUT

in a multithread application (like cinebench) the clock speed isn't a right indice.
So MULTITHREAD, D960@3.6 IS VERY SLOWER than Quad@2.6

have a good day !

Epic fail.

Your statement would be tru if we were comparing a Core2 Dual to the Quad, however since this is the good ol Pentium D, even at 2.6 the newer quad will dominate the OCed D960. Same can be said in a comparason between this 960 and a C2D clocked at 2.6

 

[iluvgillgill]

Not sure if anyone remembers the legendary single core old type celerons that goes to 5ghz on air. Ghz is everything!

yeah i do remember. in fact i got one in my room!!!lol

but back then on a 945 i only manage to get it stable at 4.8Ghz.FAIL!!!!5GHZ!!!!MY DREAM NUMBER!!!!

 

[USA Mike]

hi again !

hey guys, dont forget i'm speaking about ONLY MONO-THREAD application/task.
So, with only ONE CORE, AT 2.6GHz , a penryn cpu IS NOT SPEEDER than a 3.6GHZ PRESLER cpu.

Newer architecture deson't offer a gain of 1GHZ !!! Open your eyes and hears, specially with the same FSB !!

In Opposite, if the FSB changes, then a highest cpu fsb wins !For example, a T9600 (montevnia centrino2 FSB 1066) will be faster than a classic X9000 CPU (less FSB).

 

[dagger]

hi again !

hey guys, dont forget i'm speaking about ONLY MONO-THREAD application/task.
So, with only ONE CORE, AT 2.6GHz , a penryn cpu IS NOT SPEEDER than a 3.6GHZ PRESLER cpu.

Newer architecture deson't offer a gain of 1GHZ !!! Open your eyes and hears, specially with the same FSB !!

In Opposite, if the FSB changes, then a highest cpu fsb wins !For example, a T9600 (montevnia centrino2 FSB 1066) will be faster than a classic X9000 CPU (less FSB).

You're not listening. Cpus with newer architecture can execute more instructions (basically, do more work) per cycle. This is comparing on a basis of 1 vs 1 core. Google any benchmarks using old single core programs.

 

[iluvgillgill]

actually a Core 2 working at 800FSB below 3Ghz is more than enough. so a 2.6 Penryn will outperform the PD960. the chart below is showing a benchmark.its double threaded which the PD960, X6800, E6850 and E8400 is dual core.and as you can see the Quad core clocked at 3Ghz is not faster. so only 2 core will be used. i also picked different FSB at the same CPU clock.


http://www.tomshardware.co.uk/charts/desktop-cpu-charts/itunes-7-1-1-5,378.html?p=1267%2C1265%2C1274%2C1269%2C1300%2C1298%2C1297

comparing PD960 to E6700. speed wise the PD960@3.4Ghz is 35% faster then E6700. the time it takes to encode is 38% slower then the E6700 @2.66Ghz. So add those numbers together that will make the PD960 70%+ slower then Core 2 processor Clock for Clock.

and to beat that the PD960 need to be 70% faster then the Core 2 to make them have same performance. so in theory the PD960 needs to run at 4.58Ghz to do the same task at the same length of time.

 

[USA Mike]

thanks for this analyse !
you said a PD960@4.6 is equal to a C2D@3.6 ...i would have graphical chart which could say it !

does anybody can resolve this equation/query :

Penryn@Y = X Presler@Y ? (X is the factor, Y the clock speed) ?

 

[I800C0LLECT]

I can't believe this is even in question.

If you want the answer go find out the IPC ratio between the two architectures and apply that to whatever scaling you wish.

The architecture that Conroe brought forth is much more efficient than Prescott, Northwood, or any other P4 variant.

USA Mike...I'm not trying to bash or be a jerk but enthusiasts realized a long time ago that over all performance is based on speed AND ARCHITECTURE. You can't ignore one or the other when making comparisons.

In this case, the Kentsfield will win at default speeds.

But if this kid has been sitting on his P4 this long I'd wait a bit more and grab nehalem. That way he's not dead in the water for upgrades.

 

[Scotteq]

Mike, it seems to me that your mind is already made up. So apologies if the following amounts to post mortem abuse of an Equus Caballus...

What we're telling you is that a Core 2 is MUCH more efficient per clock cycle than a Pentium D. Meaning - It does more work per Mhz. Just look at any of the benchmarks that gillgill linked, or navigate to the CPU charts from the main page and do a little back of the envelope math. To pick a random benchmark: 3D Mark 06: P4EE at 3.2Ghz = a score of 1581. Divide the score by the clock to get an "efficiency" of 0.49 points per clock (1581/3200). Now you can use that 0.49 and multiply it by your 4.16 to get 2038. Using the 5GHZ you were referring to (but not attaining apparently), you get 2450.

The Q6600 you mentioned in your original post?? 2.4Ghz and a score of 3507, so 3507/2400 = 1.46 per clock. Compare a stock Q6600 to the calculated score of your "Best Possible Achievable" 5Ghz Pentium: (2450/3507 = 0.698), and we see that even at the highest clock speeds a Pentium can achieve, a STOCK Q6600 is still 30% better than the Pentium 4 on this particular benchmark. *And* it achieves that 30% performance margin on less than half the clock speed. Plus it won't drive up your electicity bill or heat your outdoor swimming pool while doing so. Of course, if cooking hot dogs in your case during LAN Parties is your thing, then I'm sad to say a Q6600 isn't going to work. But a 5GHZ P4?? Yah Baybee!

This is only one example, though, so please please please feel free to repeat the math on the benchmark of your choice. Just divide the score by the speed to find out how fast per clock. You will find similar results across the board: The newer processors may have lower stock clocks, but do far far more work per cycle than what you have in your machine.

Another point which hasn't been brought up is that, single threaded or not, your game does not operate by itself, even if it's all that *you* are doing. Just open up the system tray to see all of the things which are active at any given time. Now: When there's more than one available, the operating system can dedicate a core to your game, and run background tasks on others. Your P4 has to take cycles away from the game to perform those same tasks. A Core 2 Duo or Quad does not.

In short - Any Core 2 Duo or Quad will put any Pentium over it's knee and SPANK it. HARD.