4 cores in 1,Conroe E6600 8Mb Cache,Kentsfield result@MSI 975X

[coolaler]

Default clocks@cinebench2003
mulitiple cpu 1347pts@19.5s

[dogsx2]

19sec at default

Thank you!!!

[theteamaqua]

woot!!!!!

[Ron 61]

The only difference between Coolaler and a crystal ball is that Coolaler has a pulse.

[coolaler]

sisoft test

[Ailleur]

Can we compare the cinebench results with :



If so, thats some scary performances. Quite literally twice as fast. Makes sense i guess

I want that genius boy nn_step to tell me how this is fsb starved.

[Charles Wirth]

AMD

[ethernal]

If so, thats some scary performances. Quite literally twice as fast. Makes sense i guess

I want that genius boy to tell me how this is fsb starved.

Not to be a downer, but that is only a 54% performance increase going from 2 cores to 4 cores, as opposed to an 86% increase going from 1 core to 2 cores (in overall score).

Hard to say if it's from the FSB or because they are seperated dies, but there is certainly a noticable scaling drop.

[Supertim0r]

AMD

hahahahahahaha

[Ailleur]

Not to be a downer, but that is only a 54% performance increase going from 2 cores to 4 cores, as opposed to an 86% increase going from 1 core to 2 cores (in overall score).

Hard to say if it's from the FSB or because they are seperated dies, but there is certainly a noticable scaling drop.

1 core 63 seconds, 4 cores 19 secondes, 63/4=16

Looks pretty damn linear to me.
The image i copied is from another test coolaler ran with a "regular" e6700

[CompGeek]

3D Mark 2006!

[lutjens]

[grovel]

Please show us how well it overclocks...PLEASE!!!

[/grovel]

I'm eagerly looking forward to two of these bad boys together in a nice 8 core Clovertown rig...

[coolaler]

TMPGEnc
4CPU already in the preferences



muliti CPU transfer dual files



compare with

Conroe@2.95G



Merom@2.96G



quad cores win

[xenium]

*has seizure out of amazement*
I demand superpi times! lol

[kdrp]

I was really hoping that it would have the 1333mhz fsb, I mean theres no doubt that a 1066 bus is a bottleneck. The 1333 bus would still be a bottleneck but at least it wouldn't have been quite as bad.

[CanadianTSi]

Puts everything to shame at stock, just wait till it's OC'ed...

[iterations]

Not to be a downer, but that is only a 54% performance increase going from 2 cores to 4 cores, as opposed to an 86% increase going from 1 core to 2 cores (in overall score).

Hard to say if it's from the FSB or because they are seperated dies, but there is certainly a noticable scaling drop.

You realize that you are calculating that 54% performance increase between chips with two different clock speeds?

The scaling is almost perfectly linear:

Single CPU: 63 seconds
Quad Core: 19 seconds

63/(19*4) = 0.82

So there is about an 18% decrease from theoretical max.


The 8 core AMD HT enabled system shows a 4.73x (best case) speed up with 8 cores.

4.73 / 8 = 0.59

Which is a 41% decrease from theoretical max.

Which one do you think is scaling better?

Interesting eh?

[ethernal]

1 core 63 seconds, 4 cores 19 secondes, 63/4=16

Looks pretty damn linear to me.
The image i copied is from another test coolaler ran with a "regular" e6700

Well, I will admit, I didn't compensate for the difference in clockspeed. The E6700 is 11% faster (in terms of clockspeed) than an E6600. So, it actually scales 75%. This is still a bit less than 86%, but not as drastic as 54%. Still, something is clearly slowing it down a tad, but 12% is relatively excusable. Some of that could be the software's threading inefficiency. But still, there is a decrease in scaling (which is to be expected).

[iterations]

I was really hoping that it would have the 1333mhz fsb, I mean theres no doubt that a 1066 bus is a bottleneck. The 1333 bus would still be a bottleneck but at least it wouldn't have been quite as bad.

Remember, this is an early engineering sample. The chip is 1333mhz fsb capable. Maybe it will ship that way. 1333mhz fsb capable mobos will become more common.

Remember Woodcrests are already shipping with 1333MHz FSB, so it is definitely not a chip limitation.

[grimREEFER]

conroe@ 4ghz has worse cinebench than kentsfield at stock...
someone at amd just passed out lol

[coolaler]

stock vcore limit

[CanadianTSi]

stock vcore limit

[dinos22]

stock vcore limit

oh my

what are the temps like compared to Conroe

[n00b 0f l337]

someone at amd just passed out lol

Someone at AMD passed out?
Fugger and I just hit the floor stone cold!

[coolaler]

CPU queen



CPU z