Deneb previewed

[epsilon84]

http://nl.babelfish.yahoo.com/translate_url?doit=done&tt=url&intl=1&fr=bf-home&trurl=http%3A%2F%2Fwww.hardspell.com%2Fdoc%2Fhard%2F79405.htm&lp=zt_en&btnTrUrl=Vertalen

Excuse the poor translation... just look at the pretty graphs mmkay? :lol:

It looks like Deneb is about 5 - 10% faster per clock, but even more impressive is the much improved power consumption under load - almost half that of current Phenoms!

Overclocking is only so so, but this is apparently an early stepping so latter steppings may show better results, and maybe even better performance? Who knows. It should be a pretty good match for Yorkfield, their only problem is that Nehalem is right around the corner...

Anyway, at least AMD is heading in the right direction.

 

[kassler]

It should be a pretty good match for Yorkfield, their only problem is that Nehalem is right around the corner...

nehalem will use more power for same speed

 

[NMDante]

The power consumption was nice. Wish they would have done one at 3.2Ghz, since they ran many of their test at that frequency.

Otherwise, it does look like a good product coming out of camp Green.

 

[Slobogob]

Either the increase from 2 to 6 MB L3 has a decent impact or AMD didn't just shrink it but modified the Phenom. The review used DDR2 800, so with better memory the performance could be improved further - at least a little.
If this preview is correct amd might claim parity with intels current line up, if not a slight edge in some applications.

However, i simply refuse to believe anything until i read a hand full of in-depth reviews by respected hardware sites. Right now the 45nm could be what AMD intended the Phenom to be or it could be another marketing fluff like back during the messed up introduction of the B2 Phenom.

 

[epsilon84]

nehalem will use more power for same speed

Right. Show us some comparative benchmarks of Deneb against Nehalem then.

Judging by this preview, Deneb is about ~5% slower per clock than Yorkfield, which would put it some 30 - 40% behind Nehalem in mulithreaded apps.

Thats taking nothing away from Deneb, I think AMD did a great job with the shrink, Nehalem is just in a different league performance wise. Only a rabid AMD fanboy like yourself would fail to realise that.

The power consumption was nice. Wish they would have done one at 3.2Ghz, since they ran many of their test at that frequency.

Otherwise, it does look like a good product coming out of camp Green.

Considering they took 1.5V to hit 3.2GHz, it wouldn't have been pretty. As I said, its an early stepping though (C0, apparently they are up to C2 according to XS), so there is scope for further improvement I'm sure.

 

[kassler]

I think AMD did a great job with the shrink, Nehalem is just in a different league performance wise.

Nehalem has 130 watt in tdp, you don't now yet if it is faster than deneb.
also nehalem will not be out for normal users before late 2009

 

[zenmaster]

Right. Show us some comparative benchmarks of Deneb against Nehalem then.

Judging by this preview, Deneb is about ~5% slower per clock than Yorkfield, which would put it some 30 - 40% behind Nehalem in mulithreaded apps.

Thats taking nothing away from Deneb, I think AMD did a great job with the shrink, Nehalem is just in a different league performance wise.

Let the FanBoy Dream - Don't Dash his hopes that much. But I think teh Difference would only be about 25-30, not 40%. But still its the same point.

 

[spuddyt]

seriously kassler, are you expecting nehalem to be slower than the current CPU's from intel? because that seems to be the only way they could be at the same level as Deneb....

 

[epsilon84]

Nehalem has 130 watt in tdp, you don't now yet if it is faster than deneb.
also nehalem will not be out for normal users before late 2009

Actually, I do know Nehalem will be faster than Deneb. Substantially faster at that. The 5 - 10% performance increase in Deneb takes it about level with Kentsfield, and slightly short of Yorkfield. So unless you are claiming Nehalem will be no faster than Kentsfield...

The TDP figure means nothing, its just an upper limit for a given class of CPUs. A QX9650 has a 130W TDP and consumes ~65W under full load. An E8400 has a 65W TDP and consumes ~30W under load.

Anyway, you can believe whatever you want. All will be revealed in a few short months.

 

[kassler]

seriously kassler, are you expecting nehalem to be slower than the current CPU's from intel? because that seems to be the only way they could be at the same level as Deneb....

it is a different processor and works like phenom.

L2 cache on nehalem is only 256 KB and 11 clocks

L2 cache on core 2 is 6 MB for new processors and 15 clocks

L3 cache on nehalem is 8 MB and 39 clocks

L3 cache on Phenom is 43 clocks I think

How these processors behave depends on what type of application you test. Also the first version of Nehalem has tripple ddr3 and I think one computer powered by nehalem will be more then twice the price of deneb AND it will use more power

 

[speedemon]

Im pretty sure the crysis benchmarck is incorrect. I have not seen a phenom proc. achive 30+ fps in this game, no matter the speed.

 

[epsilon84]

it is a different processor and works like phenom.

L2 cache on nehalem is only 256 KB and 11 clocks

L2 cache on core 2 is 6 MB for new processors and 15 clocks

L3 cache on nehalem is 8 MB and 39 clocks

L3 cache on Phenom is 43 clocks I think

How these processors behave depends on what type of application you test. Also the first version of Nehalem has tripple ddr3 and I think one computer powered by nehalem will be more then twice the price of deneb AND it will use more power

Pictures speak louder than words.

 

[cjl]

You can say it all you want, but until you show sources for your assumptions about Nehalem's pricing, power, and speed, they are completely worthless.

Oh, and Nehalem is supposed to launch at the low end around $330, IIRC. I wouldn't call that outrageous.

Speedemon: that's the Crysis CPU benchmark. It is intentionally non graphically limited, and could very easily be run at 30fps.

 

[kassler]

http://images.anandtech.com/graphs/nehalempreview_060508030043/17015.png

Pictures speak louder than words.

And you beleve anandtech?

 

[gamerk316]

Basically, the 3.2GHz Phenom is slower than the 2.66GHz Quad, which is slower than the 2.66 Nehalem.

Of course, the pictures could be photoshopped as part of an Intel conspiracy...

 

[epsilon84]

And you beleve anandtech?

No, I think they used a random number generator to come up with those results. :lol:

 

[B-Unit]

TDP represents how much heat the proc lets off, it has nothing at all to do with how much power it eats.

 

[njalterio]

It amazes me how the cold hard facts can be displayed right in front of someone and they will still deny the fact.

@ Kassler, Anandtech is a very legitimate benchmarking source. Don't even try to deny their validity. I don't think it's possible, but we might even take you less seriously for that.

From the looks of it, Nehalem will soundly beat Deneb. However, AMD has not lost this round yet. We still have to see what pricing will take place, and how well the new processors can overclock.

 

[kassler]

@ Kassler, Anandtech is a very legitimate benchmarking source. Don't even try to deny their validity. I don't think it's possible, but we might even take you less seriously for that.

No it isn't. Check the commercials and it is very easy to spot that they are VERY pro Intel.

 

[kassler]

http://anandtech.com/cpuchipsets/intel/

where is AMD on anandtech?

 

[NMDante]

No, since Scientia (head of the "Green" fanboy club) has decided that Anandtech is biased, almost every one of the "green" club has believe that to be fact also.

Of course, there are other review sites that come up with the same benchmark results, that weren't called "biased", but that's a different story, I guess.

 

[cjl]

http://anandtech.com/cpuchipsets/intel/

where is AMD on anandtech?

Nowhere in a link to the intel category. Try this one:

http://anandtech.com/cpuchipsets/

 

[epsilon84]

TDP represents how much heat the proc lets off, it has nothing at all to do with how much power it eats.

'Heat' is related to temperature. TDP is supposed to represent power consumption, but in the case of Intel, it is generally quite conservative. You will rarely find an Intel CPU reaching the stated TDP.

 

[epsilon84]

No it isn't. Check the commercials and it is very easy to spot that they are VERY pro Intel.

I've refreshed the Anandtech main page 3 times, and there was a Zipzoomfly ad, a Foxconn ad and an Antec ad. Only the zipzoomfly ad is remotely related to Intel, since it advertised the P45 chipset, but its still an advertisement for ZZF rather than Intel. So, as per usual, your claims are groundless.

Btw, can we please get back on topic instead of discussing Anandtech?! This thread is about Deneb, not Anandtech.

 

[Scotteq]

Hmmmm...

Misinterpreted 'facts' and made up suppositions being thrown around about products which haven't even been released yet??

CHECK!!

Epic doses of Truthiness??

CHECK!!


Points made by one side being ignored by responding with unrelated data??


CHECK!!


Accusations of Fanboism being thrown around like popcorn at a PowerRangers movie??

CHECK!! AND DOUBLE CHECK!



OK... Move along folks. Nothing to see here.