Anandtech benches Nehalem

[Shintai]

Q4 for servers, not desktop, or only very limited

Q4 2008 for servers, mobiles and X58 coupled versions.
Q1 2009 for performance/mainstream desktops with Ibexpeak
Q2 2009 for value desktops with Ibexpeak and GPU

[RealTelstar]

It is the Multithreaded era.

Tell that to the software houses that are lagging 2y+ behind cpu development.

[GoThr3k]

Q4 2008 for servers, mobiles and X58 coupled versions.
Q1 2009 for performance/mainstream desktops with Ibexpeak
Q2 2009 for value desktops with Ibexpeak and GPU

availability is the question here

[Metroid]

Tell that to the software houses that are lagging 2y+ behind cpu development.

They know about it. I would not say they are lagging behind as it is difficult and costly. I'm saying this is just the right time to make the implementation and Nehalem is a good marketing product for the purpose given the prospectus.

Metroid.

[Spawne32]

man those numbers are AWESOME! i couldnt have even hoped for that much this early. This does not bode well for AMD AT ALL.

[alucasa]

availability is the question here

I feel that it can't get worse than Barcelona availability fiasco.

[Metroid]

(core2 = nearly bankrupt AMD).

AMD did it to itself. AMD bought ATI without any short return investment.

Companies have to be prepared for the worse as the technology changes everyday and the competition is just in the outdoor waiting for your drop.

AMD had many management issues, paying higher salaries to people who were doing nothing. The cash cows have stopped and AMD have become what we know as of today. A demolished but at least a structured company.

AMD knew about the threat that it would face as Conroe benchmarks were shown earlier on March 2006. So the question is why did AMD buy ATI knowing about threat? Joy Of Life? Excessive Money?

Probably lack of business knowledge of the leadership or executive team , over confidence, perhaps naiveness with hypocrisy.

AMD had its competitive advantage now is history, AMD can not knock over a company that is almost 30 times more valued, miracles may happen after all.

Metroid.

[terrace215]

availability is the question here

There is not going to be a problem with availability for the enthusiast desktop segment-- only small volumes are required there.

Did you also note the various comments over the past few days that Nehalem seems to be ready sooner than expected? I think Q4 was a conservative schedule that they will not have any trouble meeting-- they could probably start in Q3 if they wanted to.

[Macadamia]

Building a 8 core Harpertown rig this month. I am gonna be on tight budget throughout this month.
Gamers will never know our pain.

Too bad that Nehalem is 4 physical and 4 logical though.

Off topic, when is 8 physical-core coming ? What codename is it under ?

Beckton EX, Q409 IIRC.

Really, I think you're off scaling better with 2 dual barceys, and probably enough money for another Q6600 rig.

[Zucker2k]

AMD did it to itself. AMD bought ATI without any short return investment.

Companies have to be prepared for the worse as the technology changes everyday and the competition is just in the outdoor waiting for your drop.

AMD had many management issues, paying higher salaries to people who were doing nothing. The cash cows have stopped and AMD have become what we know as of today. A demolished but at least a structured company.

AMD knew about the threat that it would face as Conroe benchmarks were shown earlier on March 2006. So the question is why did AMD buy ATI knowing about threat? Joy Of Life? Excessive Money?

Probably lack of business knowledge of the leadership or executive team , over confidence, perhaps naiveness with hypocrisy.

AMD had its competitive advantage now is history, AMD can not knock over a company that is almost 30 times more valued, miracles may happen after all.

Metroid.

Answer: AMD: "Damn, we're f*!ked! Let's go grab ATi while we still can."

Seriously though, hasn't ATi proven to be the AMD saver?

[Periander6]

Answer: AMD: "Damn, we're f*!ked! Let's go grab ATi while we still can."

Seriously though, hasn't ATi proven to be the AMD saver?

No. To buy them AMD gutted its balance sheet, they basically spent all of its profits from the K8 era and took on a ton of debt as well and they have gotten nothing but losses in return. The former ATI has lost money every single quarter since they were bought.

[Roger_D25]

I just finished reading through the whole Anandtech article and then came right to this section of the forum thinking I'd post the news for everyone to check out, lol! Like usual I'm far behind the ball as there is already 5 pages here, thankfully I read through a few threads before posting the article! Like most I'm excited about this new archietecture, very impressive especially given the early hardware tested in the article!

[savantu]

There is not going to be a problem with availability for the enthusiast desktop segment-- only small volumes are required there.

Did you also note the various comments over the past few days that Nehalem seems to be ready sooner than expected? I think Q4 was a conservative schedule that they will not have any trouble meeting-- they could probably start in Q3 if they wanted to.

IIRC it's something like this :

Q4 2008 - XE and Xeon DP
Q1 2009 - mainstream
Q2 2009 - low-end with IGP

[GoThr3k]

There is not going to be a problem with availability for the enthusiast desktop segment-- only small volumes are required there.

Did you also note the various comments over the past few days that Nehalem seems to be ready sooner than expected? I think Q4 was a conservative schedule that they will not have any trouble meeting-- they could probably start in Q3 if they wanted to.

i was talking about Q1 and Q2, i dont think nehalem is that ready, but it is shaping up quite nicely though
Maybe some chips are in awesome condition and they are trying to send a strong signal they are ready for it, and AMD is not.
altough the initial results look really good, i just hold off til i see some more meaningfull benchmarks

[BrowncoatGR]

The way the numbers from these tests are presented is totally uninformative but in multithreaded apps Nehalem seems to do great. And SMT scaling is just WOW.

first consider this - the tri chanel nehalem setup had one extra stick of RAM. that's what 5-10 W right there
they have 6 months to improve manufacturing...
do you really think that nehalem will launch using more power at a given clock speed?

Yes absolutely. Nehalem is a much larger and complex chip than Penryn and also has a lot more logic circuitry. The opposite would be surprising actually.

@Shintai and Co --> L3 latency = 39cycles

[gallag]

Actually guys, If you nip over to AMDzone scientia and the_ghost have some pretty compelling theories as to why we should not get to excited about this.

A guy says "I think even the most optimistic person would say AMD is in deep trouble."

and scientia cleverly points out "I would say that anyone who doesn't care about proper testing would say that AMD is in deep trouble."

Yeeeeaaaaahhhhh AMD is still safe anyway. Thanks to scienta thinking outside the box he deduced that AMD need not worry because Nehalem was tested using to many threads which really never happens (((some of the shrewder of you may very well point out that many AMD supporters including sci have eagerly pointed out that phenom was actually a better cpu than penryn or at least viable because under highly multi threaded apps it performers and scales quite well and to be frank single threaded performance is not really important anymore))) and when in the real world k10 with ddr3 will probably compete nicely.

At the end of the day though this is all moot because the_ghost suggests that anandtech is pretty much owned by Intel so all this actually means nothing.

Thank the gods for rational folk http://www.amdzone.com/phpbb3/viewto...?f=52&t=135197

[Donnie27]

@ GoThr3k

We also ran the single-threaded Cinebench test to see how performance improved on an individual core basis vs. Penryn:

Cinebench shows us a 25% increase in performance even when looking at a single Nehalem core vs. a single Penryn core. The single-to-quad core scaling is actually a little better on Penryn, which we suspect may be caused in part by the memory issues on the motherboard we tested with.

Did you skip this part or did Anand revise it? Yet, even if performance was the same or slightly less per core, Nehalem ends up being faster due to other improvements. Most apps are Dual Core and WinXP and up uses Dual Core as well. At a minimum, one core runs the OS while another would run a single thread.

To the other guy, this does sound or read like the same comments during 2006 when C2D was tested by this same person, WOW! After over two years, you'd think they'd learn something by Now.

@the others

Please don't forget Timna had IMC, EV6 is from Alpha, Fusion is a copy of Timna, HT cn be traced back to AMD's RAMBUS license long before Lighting DATA or etc.... Oh brother!

[makaka]

Where is the gaming performance ???. This look to me as core 2 performance + Hyper threading

[flippin_waffles]

Actually guys, If you nip over to AMDzone scientia and the_ghost have some pretty compelling theories as to why we should not get to excited about this.

A guy says "I think even the most optimistic person would say AMD is in deep trouble."

and scientia cleverly points out "I would say that anyone who doesn't care about proper testing would say that AMD is in deep trouble."

Yeeeeaaaaahhhhh AMD is still safe anyway. Thanks to scienta thinking outside the box he deduced that AMD need not worry because Nehalem was tested using to many threads which really never happens (((some of the shrewder of you may very well point out that many AMD supporters including sci have eagerly pointed out that phenom was actually a better cpu than penryn or at least viable because under highly multi threaded apps it performers and scales quite well and to be frank single threaded performance is not really important anymore))) and when in the real world k10 with ddr3 will probably compete nicely.

At the end of the day though this is all moot because the_ghost suggests that anandtech is pretty much owned by Intel so all this actually means nothing.

Thank the gods for rational folk http://www.amdzone.com/phpbb3/viewto...?f=52&t=135197

Hmm, good points over there. There's certainly some educated people on that forum. Thanks for the link! (you do know it's not in marketing's best interest to point to something that legitimately questions these early results, yes?)

[Shintai]

@Shintai and Co --> L3 latency = 39cycles

You mean the CORESPEED L3?

21.5 ns to get a request back from Phenom's L3 vs. 14.6 ns with Nehalem's 4times bigger cache?

50% faster at 2.66Ghz. And most likely alot faster at higher speeds.

[RPGWiZaRD]

Holy already 5 pages lol, thought I had missed something.

The memory performance I'm most satisfied with, I've wanted seeing an improvement in this for a while since I like mem performance tweaking. Hopefully timings will become useful to mess with again such as in DDR1 S939 AMD era.

I was really gonna wait for the "tick" version of Nehalem as I bought this Wolfie quite recently anyways but looking at these tests I already start becoming a bit worried if I can resist or not.

[Mad1723]

Tell that to the software houses that are lagging 2y+ behind cpu development.

Reminds me of the Wirth's Law:

Software gets slower faster than hardware gets faster

[Donnie27]

Hmm, good points over there. There's certainly some educated people on that forum. Thanks for the link! (you do know it's not in marketing's best interest to point to something that legitimately questions these early results, yes?)

I think he was being sarcastic LOL!

[BrowncoatGR]

and scientia cleverly points out "I would say that anyone who doesn't care about proper testing would say that AMD is in deep trouble."

Regardless of how good or bad Nehalem is, he's right that the testing done by Anandtech is incredibly bad.

[clayton]

First time I'll consider extreme editions for high clocked Nehalem. >=)