Intel nehalem benchmark

[makaka]

Hi there ;
finally the first nehalem benchmark result , i can't believe my eye's .
8 second in super pi 1 mb for 2.66 Ghz only
http://xtreview.com/addcomment-id-4208-view-Intel-Nehalem-benchmark.html

 

[Avenger_K]

Hmmm....... very interesting. Nice find, even if it ends up turning out to be faked.

 

[aggiestudd07]

not thinking so (in reference to the e6750 @ 2.66ghz), my q6600 @ 3.2 do a 1M calculation in 16.7s. I have difficulty believing that post.

i would be very jealous if it were true though

 

[makaka]

maybe they are referring to core 2 duo E6750 at 444 system bus and 1:1 memory

 

[makaka]

http://www.pcstats.com/articleview.cfm?articleid=2156&page=9
in the link above the E6750 has 18 s in super pi @ stock speed using crappy i650 chipset

 

[apache_lives]

Hi there ;
finally the first nehalem benchmark result , i can't believe my eye's .
8 second in super pi 1 mb for 2.66 Ghz only
http://xtreview.com/addcomment-id-4208-view-Intel-Nehalem-benchmark.html

is that all? my Cyrix gets 8 minutes easy

SuperPi has always been intels ace benchmark, wouldnt supprise me, P4's were scoring ~40 seconds while an E6600 gets 20 seconds stock, but 8 seconds, dang.

 

[jimmysmitty]

Its hard to say really. I mean it looks fake to me but could be real if Intel is trying to at least keep Nehalem on schedule.

I for one doubt it due to the fact that with Penryn around this time until about September Intel would release their own in house benches. Then around late September / early October the enthusiast sites such as THG and XBit got their own ES's to play with.

So don't expect anything unless its an official release from Intel. but this seems to look like the performance jump Intel expects. I hope it is better than this.

 

[apache_lives]

Its hard to say really. I mean it looks fake to me but could be real if Intel is trying to at least keep Nehalem on schedule.

I for one doubt it due to the fact that with Penryn around this time until about September Intel would release their own in house benches. Then around late September / early October the enthusiast sites such as THG and XBit got their own ES's to play with.

So don't expect anything unless its an official release from Intel. but this seems to look like the performance jump Intel expects. I hope it is better than this.

Well look how long the 65nm Pentium 4 and Pentium D lineup lasted, the last of the series before a overhaul. Intel typically uses a manufacturing process to test a series and then uses the results to make there next gen chips, there using Core 2's to test it and work out any minor details (eg G0 65nm Core2's are cooler and clock higher), before the real deal.

On the other hand, the screen shots of CPU-Z are interesting, it detects the FSB as 3 x 444mhz rather then a QDR setup (4 x 333), is this true or perhaps it cant read the new interlink design, and the cache levels (L1) are the same, and i thought i heard that Intels next gen would have less transistors then current Peryn's because of the IMC, the cache wouldnt be needed as much (unlike now with the limited FSB design), but then again if thats 4 cores (monolithic) with 8mb shared then thats 33% less then a Peryn (12mb) with an IMC, thats still loads of cache.

2.66ghz is also sounding healthy, we could be in for mid 3ghz monsters pushing ~5 seconds, its amazing if true, a P4 with LN could barely break the 20 second barrier @ 7.4ghz, this has 4 times the cores (super pi is single threaded) and air cooled (well i presume it still is ), this is going to be a monster.

Speculation only, of course, if it were true, only time will tell whether it is, but it does make sense.

 

[imrul]

i bet its fake... but then again nahelem is a different architecture change just like prescott to core

 

[MU_Engineer]

Hi there ;
finally the first nehalem benchmark result , i can't believe my eye's .
8 second in super pi 1 mb for 2.66 Ghz only
http://xtreview.com/addcomment-id-4208-view-Intel-Nehalem-benchmark.html

Oh, god, not this again. Didn't we all go through this with the Conroe and the Phenom already? The first screenshots are always fakes, ESPECIALLY if some random person on a forum posts them. The first real benchmarks and screenshots typically come from very well-known and widely-respected people in certain forums (such as Coolaler) that know people doing initial ES testing that sneak a chip out for them. They are followed closely by the major enthusiast websites as the real ES chips get dispersed roughly at the same time and more people can get a hold of one for testing. Just wait until you see a bunch of benchmarks if you really want to know how it performs.

 

[jimmysmitty]

Oh, god, not this again. Didn't we all go through this with the Conroe and the Phenom already? The first screenshots are always fakes, ESPECIALLY if some random person on a forum posts them. The first real benchmarks and screenshots typically come from very well-known and widely-respected people in certain forums (such as Coolaler) that know people doing initial ES testing that sneak a chip out for them. They are followed closely by the major enthusiast websites as the real ES chips get dispersed roughly at the same time and more people can get a hold of one for testing. Just wait until you see a bunch of benchmarks if you really want to know how it performs.

Or as I said Intels in house benchmarks and screens will come first then foolowed by what you said and then the ES's get out to the test sites. Hopefully if all goes well will blow you mind in number shattering performance. Or maybe Nehalem will complete it faster than 8 seconds and make your jaw drop to the floor drooling from a severe heart attack.

 

[starcraftfanatic]

Probably fake, anyone remember the posts about the 3Dmark scores of the "super" Phenom?

 

[jimmysmitty]

Probably fake, anyone remember the posts about the 3Dmark scores of the "super" Phenom?

I think that was the "cherry picked" 3GHz Phenom (pronounced fee-nom) and a quad CFX setup which either way there is not that many 3GHz capable Phenoms and no support for CFX yet even with 790FX chipests.

Or maybe it was their "what we want it to do" vs what it could do.

 

[yomamafor1]

Probably fake, anyone remember the posts about the 3Dmark scores of the "super" Phenom?

You mean the one where Charlie made his "dance in the aisle?" No screenshot, or any form of verification has been posted. Charlie's former colleague, Fuad, even openly stated 30,000 in 3DMark06 is impossible to reach with a Phenom and two HD3870s.

 

[jimmysmitty]

You mean the one where Charlie made his "dance in the aisle?" No screenshot, or any form of verification has been posted. Charlie's former colleague, Fuad, even openly stated 30,000 in 3DMark06 is impossible to reach with a Phenom and two HD3870s.

But it was close with a Skulltrail system and SLI 8800Ultra setups. I think it hit over 29,500.

 

[Badbonji]

It is probably fake. Nehalem supports SSE4.2 instruction set which I cannot see in the instructions on the picture, and my Yorkfield even supports 4.1... I assume the first part is from a normal core 2 extreme whereas the benchmark is probably from a LN2 cooled Quad core at 4Ghz+

 

[Shadow703793]

Why can't ppl. just wait for the real thing.......

 

[jimmysmitty]

Why can't ppl. just wait for the real thing.......

Because we want it all....and we want it now. Like the Dr Pepper commercial.

I understand what you mean. But I really would have liked to believe that was a real benchmark. I wich I could get one too. It would be nice to say I am the first person with a Nehalem.

 

[speedbird]

What's the point of this I say? Like Super PI is a true indication of overall actual performance anyway Then there's the question of this being a fake, which it probably is. Intel are hardly going to let some nobody have one so they can do pathetic pointless benchmarks like super PI.
If their going to fake benchmarks then at least make it believable and use a decent Benchmark only fools will take this seriously.

 

[radnor]

First off, fake, because:

"Too soon Executus, you have awaken me too soon!!"

Second, if they were to publish ANY benchmark, it wont be superPI. Thats a test that GPUs will trash real soon.
So they wont brag on it, knowing that will be trashed soon. And we are talking about a single-threaded app in supposed king of multi-threading ?

Thats just silly.

 

[chise1]

Hey guys, it's a fake, don't worry... Here's the original forum post from 2006 when a B0 X6800 was tested, but the results were photoshopped. It is not Nehalem:

http://www.xtremesystems.org/forums/showpost.php?p=1547196&postcount=21

The forum post where they explain it can be found here: http://www.xtremesystems.org/forums/showthread.php?t=176526 (scroll 3/4 of way down)

 

[iluvgillgill]

But now it looks like the Nehalem is much faster then all the oldies we got now!
http://www.vr-zone.com/articles/Intel_Bloomfield_On_X58_Board_Benchmarked/5825.html

MU_Engineer get ready to cry!!!

 

[MU_Engineer]

But now it looks like the Nehalem is much faster then all the oldies we got now!
http://www.vr-zone.com/articles/Intel_Bloomfield_On_X58_Board_Benchmarked/5825.html

MU_Engineer get ready to cry!!!

Why would I be crying? The only reason I'd be crying is if they price these chips and boards the way they did back in the Bad Old Days or shackled it to yet another slow, hot, only-Intel-uses-it memory type a la RDRAM. That doesn't seem likely as we know that Bloomfield uses standard DDR3 (which will go down in price quite a bit as demand grows) and Intel has seemed to be wanting to play ball with prices these last few years. They could have very easily sold us $350 E6300s on launch day and kept the quads over $500 even now, but they chose not to. So I think this looks like nothing but good news.

If this is in fact a genuine screenshot, I am quite excited. The thing that really got my attention was that Vcore. It is in ULV CPU territory (my notebook's 1.06 GHz C2D U7500 has a 0.850/0.875 Vcore) so this ought to be one very cool-running CPU and possible very overclockable. The only reason I doubt the screenshot is that it says the chip has 4x32 KBytes L2 when it should have 4x256 KB. That could be CPU-Z not correctly recognizing the processor or it could be a faked screenshot. If it is real, then there should be a bunch of benchmarks coming out soon and it will be nice to see them.

 

[TechnologyCoordinator]

I don't trust that result.

 

[chris312]

But now it looks like the Nehalem is much faster then all the oldies we got now!
http://www.vr-zone.com/articles/Intel_Bloomfield_On_X58_Board_Benchmarked/5825.html

Wow, that looks promising! I really want to know pricing now, but the mainstream procs won't come out for another 9 months from now.. oh well, that puts it in time for graduation for me :lol: