Leaked Nehalem Performance Numbers Posted

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
George Oh has posted leaked six-core Xeon performance information and specs today that are worth checking out. Not sure how long the info will last so get a look while you can.

Reliable sources have reported in the past that Intel’s Nehalem processor will have three channels of DDR3 memory per CPU versus two channels of DDR2 memory per AMD Barcelona or upcoming Shanghai processor. That would mean that AMD’s massive memory bandwidth advantage will turn in to a large memory bandwidth.
 
First thought was that it looks very impressive. However i dont see it mentioned anywhere if its the 8core chip, if so then thats not as impressive. ( i didnt check the aces hardware thread, so it might be mentioned in there)
 
I was honestly impressed with the performance of AMDs upcomming chip, although it seems like that's probably just an estimate. I saw another link to the article that said the numbers were for a six core chip.
 
It will be funny to see how theinq and fuad spin this.
 
Wow AMD is close to holding its own, im more impressed by that to be perfectly honest.
 
Wow AMD is close to holding its own, im more impressed by that to be perfectly honest.
I think you read it wrong. The performance gap is going to widen significantly when Nehalem is released.
 
Does "Nehalem" remind anyone else of Final Fantasy VII?

Don't cry for me, Nibelheim.
 
Uh, hey Steve. I do not see "6 cores" anywhere on that page. Am I just being really blind?

Nehalem-EP was previously thought to be a 4core 8thread design.
 
Interesting. Looks like the next round of processors will give us great performance from either the green or the blue camps.
 
Hmm, let me see if I got this right, that chart is comparing a new 6 core chip against 4 core chips? I'll wait until the product hits or someone actually does a proper performance preview of this thing. I'm sure the chip as a whole is refined and Intel is indeed churning out some fine technology, but of course 6 cores will show a big improvement over 4 in those benchmarks (as questionable as they may be).

I'm interested in seeing how well that dual thread per core technology works in real world scenarios. Then I'd like to see how much of an improvement there really is over the current chips by comparing 4 cores of Nehalem against other quad cores. It's not fair to do 6 against 4. Damn, another platform change (nothing new, I know) but last I checked DDR3 isn't cheap for 4GB at about $550 or so for a kit (Triple channel so 3 or 6GB kits?). Lord knows what the motherboards are going to cost initially.

Well, I'll just have to wait and see. I'm kinda excited though, in theory the sum of all it's parts should equal a sick system.
 
You might as well just compare Nehalem against a dual core Opteron. There are no quad core Opterons available still, I'm not holding my breath on Shanghai even existing anytime soon.
 
Interesting. Looks like the next round of processors will give us great performance from either the green or the blue camps.

What cause we dont have spectacular performance right now? What we need are more consumer related programs/applications that take advantage of multi-core cpu's.
 
Also these charts are unbelievaably questionable. I mean the Inq probably comes out with more factual stuff than this thing. Dont get me wrong though I'm sure it'll be a fine cpu but this link on its own is hardly enough to stand on its own.
 
What cause we dont have spectacular performance right now? What we need are more consumer related programs/applications that take advantage of multi-core cpu's.

I never said that we don't have spectacular performance as of right now. Shoot, my old Opteron 170 running at stock 2GHz and 2GB of RAM did everything just fine. Fast forward a year and a half later (present day) to my 6400+ running at stock 3.2GHz with 4GB of RAM. Nothing noticeable in terms of performance gains not taking into account my move from 7800GT SLI to a single 8800GT, except Windows boots a little faster.

I was merely stating that these next gen procs look to offer a pretty big boost over current gen procs by way of the benchmark results. Real world testing and reviews are what I need, however... but just from the benchmark scores, they impress me,

I also agree that it would certainly be nice to have programs/applications/games (emphasis on the later) that can take better advantage of the technology as it sits today. However, I unfortunately don't see that happening anytime soon, so hardware development must make up for that. These new procs certainly seem to be a huge step in the right direction in terms of that.
 
ahhh yes, more fine journalism from George, anything for Intel. The numbers look great but its a 6 core vs 4 core. That being said I look forward to what Nehalem can do, and left wondering how much money does he get for sucking so much Intel ...... you fill in the blank.
 
Also these charts are unbelievaably questionable. I mean the Inq probably comes out with more factual stuff than this thing.
These were prepared from internal documents leaked on Sun's web site. That's 10x more reliable than the usual inq garbage which is sourced from, um, nothing.

The iffy part of the data in the graphs is what points he (George Ou) chose to start the comparisons. There are still no "valid" quad core Opteron SPEC scores, and there are many different Xeon SPEC scores to choose, although it is likely that Intel used Q3 or later scores. That's a minor quibble though, the relative performance differences will still remain the same regardless of the exact numbers.

These are just projections for Shanghai and Nehalem anyways. Although one of the two companies that makes those products is pretty good at meeting projections. ;)

If you want hard numbers, you'll have to wait at least 3 months before either chip ships in systems.
 
Also these charts are unbelievaably questionable. I mean the Inq probably comes out with more factual stuff than this thing. Dont get me wrong though I'm sure it'll be a fine cpu but this link on its own is hardly enough to stand on its own.


QFT - I am sad to see this sort of journalism on ZDNet. It is beneath them. Or I guess not.
 
What cause we dont have spectacular performance right now? What we need are more consumer related programs/applications that take advantage of multi-core cpu's.

Do you want the CPU companies to stop making CPUs to go focus on software? Because that's what your message seems to imply. That's up to the software makers, they'll catch up eventually, but for most it doesn't make sense to go multithreaded, most programs I (and probably everybody else) run use <10% of my CPU, so there's no reason to multithread them. Games/Apps which could use the performance are already multithreaded or being multithreaded or will be in the near future, for the most part.
And then consider, most games that give bad performance are held back by the GPU for which multithreading and number of cores makes no difference.
 
ahhh yes, more fine journalism from George, anything for Intel. The numbers look great but its a 6 core vs 4 core. That being said I look forward to what Nehalem can do, and left wondering how much money does he get for sucking so much Intel ...... you fill in the blank.

You should have mentioned this on the first page, I won't waste my time reading the second page of this thread if I knew that it is a George Ou article, it is like forcing me to read a Fuad article ;)
 
I'm just looking forward to a g0 type of stepping for the current 45nm quads. won't want to spend more money for a new mobo too. I'm glad progress still continues in the cpu realm (wish it went to the video cards too)
 
sure is fun to theorize huh?

while reading the number and graphs is entertaining. there are many unanswered questions. in fact too many unknowns to be of much use other than propaganda or to fuel the ever persistent fanboy flame wars. i do think Intel has learned that they (intel) can be caught and beat, and you cannot rest on your laurels. prescot to conroe case in point.

i hope for our own sakes AMD can compete again. it helps to serve as price control so i can afford my addiction :rolleyes:

True or indicative or not, the performance is already here, and will get better. six cores huh? <insert your favorite Pinky and the Brain comment here>
 
I have learned that synthetic benchmark means nothing. Other than compiling linux, encoding and servers, what else is it good for?
 
Either that was sarcasm that didn't set my meter off, or you fail to realize that there are people who would by this CPU to do nothing but what you listed.

Also, I fail to see what the problem is with the article. The source is Intel's own slides, George just did some maths to estimate real scores from the data Intel provided. Yes, it may be a six- or eight-core CPU vs a quad-core AMD offering, but that's just how this SPEC benchmark works, and, frankly, that's the state the market will likely be in at that point. As much as I'd like to see it, I don't see AMD pulling an eight-core CPU out of their hat.
 
Nehalem EP is the quad core. I still don't understand why people are talking about Dunnington in this thread. :p The link in the news article has nothing to do with Dunnington.

This is the slide from Sun's server that is getting people all riled up: http://img254.imageshack.us/my.php?image=nehalemtv6jw6.jpg (I have the whole 21 page PDF and it looks pretty real)

Repeat: it's just a projection. AMD does it too. It's not unusual. I don't see what the big surprise is either. Shanghai is a tweak of a K10 core that lags 1+ year old Intel quad core CPUs. Nehalem is a new architecture with a 3 channel ODMC and high speed interconnect, and probably to AMD's displeasure, is a "native" quad core CPU.
 
It will be funny to see how theinq and fuad spin this.

"New operating systems that need 4GB+ Ram to run smoothly now central processing units that need to 16GB just to carry out a single equation!"
 
Wow AMD is close to holding its own, im more impressed by that to be perfectly honest.

what? does anybody seriously think AMD will fade away to dust? they didn't get where they are by jacking off wasting time...

AMD will STOMP back again..

then hopefully soon after that INTEL will STOMP back again ;)
 
what? does anybody seriously think AMD will fade away to dust? they didn't get where they are by jacking off wasting time...

AMD will STOMP back again..

then hopefully soon after that INTEL will STOMP back again ;)

And back. And forth. And back, ad nauseum.

We don't want to see AMD go the way of 3dfx, now do we?
 
Back
Top