Idf

[Toferman]

LOL @ a wafer laying under the seat.

Thanks for the updates, guys

[Movieman]

They are "faking" the score providing a Cinebench10 version with Cinema4d 11.0 Render engine and call it a usual Cinebench10 score. Pretty lame

I can't comment on that but I was told by a VERY reputable source that those posted scores were very possible so I retracted.
If they "played" with the engine then they know but I don't.
All I know is that 45,000+ In Cinebench10 is huge.
A dual socket 8 core Harpertown at 4000mhz is app 30,000 in that benchmark for example.

[Sampsa]

That Cinebench score is pretty hard to beleive..
Dual Gainstown at 4500mhz maybe?
Think on that one for just a minute.
It does seem off the scale.

Edit: statement retracted and I will go and eat my cold crow in the other room..

I can confirm the score of 45850, Cinebench info says:

Intel(R) CPU 000 @ 3,2 GHz
8 x
Windows 32 BIT 6.0.600
ATI Radeon HD 4870 X2

But there is a sticker in front of system which says: "Nehalem overclocking". They had a system on display, normal case, normal aircooling.. nothing special..

[Movieman]

I can confirm the score of 45850, Cinebench info says:

Intel(R) CPU 000 @ 3,2 GHz
8 x
Windows 32 BIT 6.0.600
ATI Radeon HD 4870 X2

But there is a sticker in front of system which says: "Nehalem overclocking". They had a system on display, normal case, normal aircooling.. nothing special..

Thanks.
Was this a single socket with HT turned or or a dual socket with HT turned off?
If it's a single socket and running an unmodded Cinebench10 then those are HUGE numbers and I want one now to stick on the cancer research we do.

[informal]

Thanks.
Was this a single socket with HT turned or or a dual socket with HT turned off?
If it's a single socket and running an unmodded Cinebench10 then those are HUGE numbers and I want one now to stick on the cancer research we do.

Movieman,JC already shown us the Bloomfield 2.93Ghz numbers for 32bit Cinebench(see the quote below).There is no way in hell that score was done with regular Cinebench 10.
From what i see (Sampsa's post) it indeed was a single socket SMT enabled Bloomfield 3.2Ghz chip,although overclocked and running 32bit OS and tested with modded Cinebech10.
User Boschwanza already explained that 45K result several times...
http://www.xtremesystems.org/forums/...0&postcount=20

Its just a bit logical . Cinema4d 11 Render engine will give you a boost up to 2.5x.

Nehalem 2.93 scores about 14k points with Cinebench 10 32Bit .. 3.2 will be around 17.000 * 2.5 makes 42.500 .. pretty close isnt it ?

With regular Cinebench10(unmodded with Cinema11 core), you can't jump from 17K(or 20K with 4ghz clock for Nehalem) to 40+K no matter how you look at it... From the exact 45850 figure,you can even calculate the OC speed of that 3.2Ghz Nehalem chip.It was OCed approx. ~1.2-1.25x,which gives us 3.84-4Ghz range (4850/ (17K x 2.2) ).I used the ~2.2x speedup for modded Cinebench since it varies a lot on different systems.

[Movieman]

Movieman,JC already shown us the Bloomfield 2.93Ghz numbers for 32bit Cinebench(see the quote below).There is no way in hell that score was done with regular Cinebench 10.
From what i see (Sampsa's post) it indeed was a single socket SMT enabled Bloomfield 3.2Ghz chip,although overclocked and running 32bit OS.
User Boschwanza already explained that 45K result several times...
http://www.xtremesystems.org/forums/...0&postcount=20

With regular Cinebench10(unmodded with Cinema11 core), you can't jump from 17K to 40+K no matter how you look at it... From the exact 45850 figure,you can even calculate the OC speed of that 3.2Ghz Nehalem chip.It was OCed approx. ~1.2-1.25x,which gives us 3.84-4Ghz range.

I understand but I want to hear the answer to my question from Sampsa as he was there and saw the machine.
I don't doubt what JC said but his was a prerelease board on single channell memory and a beta bios.
I am assuming that the machine shown by Intel yesterday could be a little "better" for lack of a better word.

[informal]

Hmm,i thought JC showed(near the end of his preview) the completely functional 3 channel Bloomfield system.The numbers from that system were even used by Hornet to make one of his charts for comparison with K10 and C2Q if i'm not mistaken.

What i said in my previous post is only based on logic.No 3 channel/better tuned system can produce 2x the score,it is simply against logic(and what we know from uarch. evolution-Nehalem is the same 4 issue 128bit SSE/fpu capable core).

[twilyth]

I'm hesitant to say anything here since I will probably come across as clueless, but hasn't it been an issue in the past regarding benchmarking software that some programs were optimized for the Intel micro-architecture and instruction set? Maybe the new incarnation of cinebench is an instance of that. However even in that case it's hard to see how you would get 2x the score so I see Informal's point.

But this might be a good thing for Agena and Shanghai. Anything optimized for Nehalem will probably also favor AMD's chips due to similarities in the architectures. Does that make sense or am I completely clueless?

[Movieman]

Hmm,i thought JC showed(near the end of his preview) the completely functional 3 channel Bloomfield system.The numbers from that system were even used by Hornet to make one of his charts for comparison with K10 and C2Q if i'm not mistaken.

What i said in my previous post is only based on logic.No 3 channel/better tuned system can produce 2x the score,it is simply against logic(and what we know from uarch. evolution-Nehalem is the same 4 issue 128bit SSE/fpu capable core).

Very good point and the score did seem very high to me for a single socket system which is why I want to talk to Sampsa on this..Or Fugger if he saw the system.

I'm hesitant to say anything here since I will probably come across as clueless, but hasn't it been an issue in the past regarding benchmarking software that some programs were optimized for the Intel micro-architecture and instruction set? Maybe the new incarnation of cinebench is an instance of that. However even in that case it's hard to see how you would get 2x the score so I see Informal's point.

But this might be a good thing for Agena and Shanghai. Anything optimized for Nehalem will probably also favor AMD's chips due to similarities in the architectures. Does that make sense or am I completely clueless?

You may well be right..

[AliG]

I've been thinking the same thing, there are enough similarities among k10.5 and i7 that anything optimised for one will be good for the other. Thing is k10.5 is stuck with a 3 issue core unlike nehalem/conroe's 4 issue core, my bet is that is what is killing performance, not the l3 cache. But enough of shanghai, we'll know soon enough if amd sinks or floats due to shanghai

[Shintai]

I've been thinking the same thing, there are enough similarities among k10.5 and i7 that anything optimised for one will be good for the other. Thing is k10.5 is stuck with a 3 issue core unlike nehalem/conroe's 4 issue core, my bet is that is what is killing performance, not the l3 cache. But enough of shanghai, we'll know soon enough if amd sinks or floats due to shanghai

Uhmm...no? They are massively different.

You want to read this:
http://www.realworldtech.com/page.cf...0208182719&p=1
http://www.anandtech.com/cpuchipsets...spx?i=3382&p=1

[Cooper]

Uhmm...no? They are massively different.

You want to read this:
http://www.realworldtech.com/page.cf...0208182719&p=1
http://www.anandtech.com/cpuchipsets...spx?i=3382&p=1

So you provided two different articles covering only one part of the question and claim they are absolutely different. Well from my understanding you either claiming that nehalem is absolutely different from nehalem (hope you not THAT insane) or your comment has no base.

[twilyth]

Uhmm...no? They are massively different.

You want to read this:
http://www.realworldtech.com/page.cf...0208182719&p=1
http://www.anandtech.com/cpuchipsets...spx?i=3382&p=1

I haven't read the articles yet, but I will.

What I had in mind was the change from a massive L2 to a small L2 + shared L3. That's a standard design feature for AMD. IIRC, there are certain benchmarks that can reside happily in a large L2 and give an advantage to chips using one.

[Shintai]

So you provided two different articles covering only one part of the question and claim they are absolutely different. Well from my understanding you either claiming that nehalem is absolutely different from nehalem (hope you not THAT insane) or your comment has no base.

Try stay on focus for once. And compare Nehalem to K10. They might look the same with 3 issue and L3 cache etc. But its day and night in difference.

And I provided 2 articles since I assume he knows about K10(.5)

Mod trolling? And personal attacks since you consider me insane on a certain degree?

[Movieman]

Put's on scratchy mod hat..
How about everyone take a breath in here and calm down..
Thanks..

[twilyth]

Put's on scratchy mod hat..
How about everyone take a breath in here and calm down..
Thanks..

I'm calm. Who says I'm not calm? HUH? WHO? WHO THE says I'm not calm?

I'll show you calm. I'm more calm than any of you will ever be!!!

DO YOU HEAR ME DAMMIT!?!?!?!?!

[Movieman]

I'm calm. Who says I'm not calm? HUH? WHO? WHO THE says I'm not calm?

I'll show you calm. I'm more calm than any of you will ever be!!!

DO YOU HEAR ME DAMMIT!?!?!?!?!

I don't get angry either..
I just reach into my bag and pull out the ...

[twilyth]

*imitates yelping of scared wiener dog and runs away*

[Lestat]

charles were those temps you posted idle temps? or did you have something running in the background?
i assumed idle because the cooling wasnt engineered for the new quad's. or... ????

[Cooper]

Try stay on focus for once. And compare Nehalem to K10. They might look the same with 3 issue and L3 cache etc. But its day and night in difference.

And I provided 2 articles since I assume he knows about K10(.5)

Mod trolling? And personal attacks since you consider me insane on a certain degree?

Try to put your thoughts clear next time. I have no reason to attack you cos there non left. You already have a certain reputation here.

[Charles Wirth]

charles were those temps you posted idle temps? or did you have something running in the background?
i assumed idle because the cooling wasnt engineered for the new quad's. or... ????

Temps were just after benching, when I do a screen shots.

[metachronos]

Damn FUGGER what brand of notebook were you using, sounds pretty badass.

[Sampsa]

Thanks.
Was this a single socket with HT turned or or a dual socket with HT turned off?
If it's a single socket and running an unmodded Cinebench10 then those are HUGE numbers and I want one now to stick on the cancer research we do.

Yes, normal desktop PC with single socket and HT on. AFAIK it was Core i7 Extreme Edition 3,2 GHz overclocked to ~3,6 GHz. On display it had Cinebench R10 open with the +45k result.

PS. Here is a pic of me and my buddy in IDF

[STaRGaZeR]

Yes, normal desktop PC with single socket and HT on. AFAIK it was Core i7 Extreme Edition 3,2 GHz overclocked to ~3,6 GHz. On display it had Cinebench R10 open with the +45k result.

PS. Here is a pic of me and my buddy in IDF

Who's your buddy? Left or right?

[JumpingJack]

Who's your buddy? Left or right?

You mean the one sitting down drinking the bottle of water .... Sampsa' attention seems more focused on the other person... I am beginning to wonder about his priorities

BTW -- WTF is a hot babe like that doing at a geek fest -- is she into computers? What a nerd dream girl.