Nehalem-EP......BLOOMFIELD

[JumpingJack]

why bus 133mhz ?

why not?

133 MHz is not the bus anyway, it is imply the system clock. The 'bus' as we know it for Intel will no longer exist. Any bus clock derived from this system clock is dictated by PLLs and dividers.

AMD's bus is 200 Mhz, with 133 and multipliers Intel has a little more fine granularity in the control over their processor binning.

Jack

[villa1n]

No offense , but that's dumb.Based on a single test which doesn't touch the improvements of Nehalem , you extrapolated to a 2% gain across all apps?

no offense but you didnt read my response to jacks answer.

I was stating what was written, and cinebench seems like a fairly good indicator of cpu performance. I was also factoring in in singlethreaded apps, the yorkfield seemed to keep pretty close to nehalem in performance, so seeing a 13% improvement on an architecture, that was designed to be fast in single threaded environments, is quite impressive, especially since it is even more apt in multithreaded environments.

[bowman]

why bus 133mhz ?

It's the reference clock, not the bus. It's just an external clock for the processor to atune itself to.

I hope it'll be like Phenom's clocks with separate clock domains for each core and each component. Fun tweaking times ahead.

[Jacky]

Does the 1m Spi test fit entirely into any of the caches, 4,6,8mb?
What architectural feature of Nehalem influences single threaded performance so positively in that scneario, while it does nothing for cinebench?
Any guesses?

[GoThr3k]

Does the 1m Spi test fit entirely into any of the caches, 4,6,8mb?
What architectural feature of Nehalem influences single threaded performance so positively in that scneario, while it does nothing for cinebench?
Any guesses?

what about an IMC and some shared 8MB L3 goodness

[Jacky]

AMD K10 has a full 3 pages of improvements over K8 and can't touch Core even with a 10 foot pole in single threaded apps.Imagine what work went into Nehalem to offer even that measly 2% in Cinebench.

btw there might be a chance that in 64bit cinebench it is more than 2%. I just realised that nehalem has improved 64bit macro-ops fusion according to kanter's article.

[Dualist]

mmmm, run wprime please

[virtualrain]

why not?

133 MHz is not the bus anyway, it is imply the system clock. The 'bus' as we know it for Intel will no longer exist. Any bus clock derived from this system clock is dictated by PLLs and dividers.

AMD's bus is 200 Mhz, with 133 and multipliers Intel has a little more fine granularity in the control over their processor binning.

Jack

Exactly! In addition, it is easy to maintain a very stable clock at lower frequencies.

Here's my projections on how the clock domains will work on Nehalem...

http://www.nehalemnews.com/2008/05/e...r-domains.html

Cheers,
-Chris.

[Movieman]

I've seen SS at 2667 and at 3200+ so this beast will OC.
How it was done I don't know but I know the people that took the SS and trust them.


Hey Andre, run cinebench10 64 bit on that sucker if you can..

[GoThr3k]

I've seen SS at 2667 and at 3200+ so this beast will OC.
How it was done I don't know but I know the people that took the SS and trust them.


Hey Andre, run cinebench10 64 bit on that sucker if you can..

everyone also saw phenom at 3ghz+ before release, no guarantee for overclock

[gOJDO]

Does the 1m Spi test fit entirely into any of the caches, 4,6,8mb?

No, it can't fit in 8MB and the algorithm of SPi is not designed to try to hold the data in cache.

What architectural feature of Nehalem influences single threaded performance so positively in that scneario, while it does nothing for cinebench?
Any guesses?

Improved OP fusion, 33% better parallelization, better & faster cache architecture, IMCs. To see if there is difference in cinebench or not we'll have to wait for the retail CPUs and mainboards. The available test are made with a single channel memory config and with crappy memory timings.

everyone also saw phenom at 3ghz+ before release, no guarantee for overclock

yeap, everybody saw phenom on 3GHz, but nobody was allowed to come closer to the system and not to even think about touching it. Right now there are many people who have nehalem and they are saying that it is very good and very cool OC-er.

[Donnie27]

everyone also saw phenom at 3ghz+ before release, no guarantee for overclock

Not even close to being the same. Only a few folks thought 3GHz would be the norm for Barcelona, while most folks think 2.66GHz Nehalem is way under clocked. Intel showed 3.2GHz and sent out 2.66GHz ES models. The first Conroe ES processors showed up at 2.66GHz so there's nothing that's not routine here.

Oh BTW, I learned a long time ago to NOT doubt MovieMan.

[GoThr3k]

Not even close to being the same. Only a few folks thought 3GHz would be the norm for Barcelona, while most folks think 2.66GHz Nehalem is way under clocked. Intel showed 3.2GHz and sent out 2.66GHz ES models. The first Conroe ES processors showed up at 2.66GHz so there's nothing that's not routine here.

Oh BTW, I learned a long time ago to NOT doubt MovieMan.

i dont doubt him

i'm just viewing it from another perspective

[Shintai]

i dont doubt him

i'm just viewing it from another perspective

There is a massive difference on Intel and AMD marketing lately on that point.

There is 2.66 and 2.93Ghz ES samples out. Plus Intel easily shows off 3.2Ghz and alot higher.

AMD had what speed barcelonas out? And only 1 single 3Ghz that was basicly locked away. Not to talk about the endless FUD slides of how fast it was compared to Core 2s. 50% faster!!!

3Ghz+ is not even a challenge for Nehalems.

[GoThr3k]

There is a massive difference on Intel and AMD marketing lately on that point.

There is 2.66 and 2.93Ghz ES samples out. Plus Intel easily shows off 3.2Ghz and alot higher.

AMD had what speed barcelonas out? And only 1 single 3Ghz that was basicly locked away. Not to talk about the endless FUD slides of how fast it was compared to Core 2s. 50% faster!!!

3Ghz+ is not even a challenge for Nehalems.

where?

[Movieman]

i dont doubt him

i'm just viewing it from another perspective

oh my friend, if you haven't decided on a career yet definately study to be a diplomat.
Damn, that was good..

[Shintai]

where?

http://www.engadget.com/2008/06/03/i...d-at-computex/

http://www.tgdaily.com/content/view/37781/135/

But seriously, with ES samples out at 2.93Ghz. You really have any doubt?

[STaRGaZeR]

oh my friend, if you haven't decided on a career yet definately study to be a diplomat.
Damn, that was good..

His perspectives are indeed interesting

[GoThr3k]

http://www.engadget.com/2008/06/03/i...d-at-computex/

http://www.tgdaily.com/content/view/37781/135/

But seriously, with ES samples out at 2.93Ghz. You really have any doubt?

thats not 3.2GHZ+, but doesnt matter

i dont have any doubt i believe nehalem will OC pretty nicely, my point was just that sometimes not all our dreams or hopes come true, remember phenom....
but some people don't even bother to read that, they just make fun of someone's opinion, even an XS moderator, retardation of society i think

[ryboto]

http://www.engadget.com/2008/06/03/i...d-at-computex/

http://www.tgdaily.com/content/view/37781/135/

But seriously, with ES samples out at 2.93Ghz. You really have any doubt?

In both of those articles, neither system revealed the true core speed...still speculation. So, I don't see why you would be providing it as fact to support your position. Out of curiosity, do you have links to any information regarding the 2.93ghz ES chips?

[Jacky]

2,93ghz - Anand's word as a proof: http://anandtech.com/cpuchipsets/int...spx?i=3326&p=5
Apparently there is no mention of 3.2ghz, but according to movieman it seems realistic.

[Movieman]

thats not 3.2GHZ+, but doesnt matter

i dont have any doubt i believe nehalem will OC pretty nicely, my point was just that sometimes not all our dreams or hopes come true, remember phenom....
but some people don't even bother to read that, they just make fun of someone's opinion, even an XS moderator, retardation of society i think

When the Clovertowns first came out they were limited to 2667 tops.
remember, this was December 2006.
The heat issue was killing anyone not on top water or phase.
No one was getting over 3000 on air because of crap heatsinks available back then. I got lucky and matched some Dynatron H46G's to some vantec tornadoes and was able to get 3157 stable and 3258 for benching on air.
Now if some dumb smuck like me can do that then some of the top people in the world are getting a crapload more than 3000 out of the Nehalems.
I rest my case

[Neuuubeh]

gotta me save moneeeh!!! :P

[Philly_Boy]

When the Clovertowns first came out they were limited to 2667 tops.
remember, this was December 2006.
The heat issue was killing anyone not on top water or phase.
No one was getting over 3000 on air because of crap heatsinks available back then. I got lucky and matched some Dynatron H46G's to some vantec tornadoes and was able to get 3157 stable and 3258 for benching on air.
Now if some dumb smuck like me can do that then some of the top people in the world are getting a crapload more than 3000 out of the Nehalems.
I rest my case

That's dumb schmuck.....spell it right!!

[GoThr3k]

When the Clovertowns first came out they were limited to 2667 tops.
remember, this was December 2006.
The heat issue was killing anyone not on top water or phase.
No one was getting over 3000 on air because of crap heatsinks available back then. I got lucky and matched some Dynatron H46G's to some vantec tornadoes and was able to get 3157 stable and 3258 for benching on air.
Now if some dumb smuck like me can do that then some of the top people in the world are getting a crapload more than 3000 out of the Nehalems.
I rest my case

your point being?

i dont see any relevance with this, and the thing i posted...
maybe you should become politician, they also start talking about something else....