Anandtech benches Nehalem

Printable View

Show 100 post(s) from this thread on one page

06-05-2008, 10:33 AM
Donnie27

Quote:

Originally Posted by eleeter

I certainly do. And it's an effective strategy for Intel.

QFT!
06-05-2008, 10:36 AM
thebluemeanie1

AMD will live on because Intel has no reason to sell these CPUs at any affordable price. The Q9xxxs didn't get released until AMD had the 9850s ready. I think that the mainstream or even the less expensive LGA 1366s won't be around for more than another year. Intel is just going to release another CPU like the QX9650 to taunt AMD for a few months.
06-05-2008, 10:38 AM
Nedjo

Quote:

Originally Posted by bowman

To say that Intel's official roadmap and product plans are FUD would definitely be stretching it.

Well I'd like to see that official roadmap and transition timeframes from Core to Nehalem architecture, wit appropriate segmentation percentage...

Point is - this CPU that Anand tested is "Extreme" version of Nehalem, intended for (again) BMW M3 market segment.

I'm aware that more Nehalems are in pipeline, but according to this unofficial roadmap: http://pc.watch.impress.co.jp/docs/2.../kaigai01l.gif they all are planed for H2 2K9 (most probably for holiday’s season), and they'll be penetrating market step-by-step, and not by cutting off current lineup.

In that sense AMD's own K10 isn't intended for competing with Nehalem but with Conroe gen. derivates. Currently AMD can wrestle with Intel’s offering in Low-to-Mid end markets, and as long is Nehalem out of that market segment (and somehow I don’t see that Intel can make more money on much bigger Havendale then on Celeron/Pentium+chipset combo) AMD has enough maneuvering space, and it is not in cut-throat grapple of Nehalem, as some people including the Anand are trying to convince us.
06-05-2008, 10:57 AM
jas420221

Quote:

Originally Posted by BrowncoatGR

You want the WOW effect to happen on or close to release to help maximise sales. Besides whats the point of an NDA on reviews if you wanted all the info out there.

This reaches more people period. Sales are all about numbers. The more people you can 'reach' the more sales you will get. This goes back to basic math... so you have 3 things to learn now... :p: :D

Quote:

Originally Posted by Nedjo

Anand has proven to be less, and less informed about what's going on in this industry (I still remember his ridiculous CeBIT "diaries")

This sentence: "2009 should hold a new architecture for AMD, which is the only thing that could possibly come close to achieving competition here" further prove that! As far as I'm aware (maybe I've missed some announcement) but AMD for 2009 only plans to further shrink/refine K10 architecture...

Well, Anand is no FUD.

I dont know any roadmaps...so who cares if they got it wrong about AMD's plans. That makes it even WORSE for them!!!! Phenom wont improve clock for clock to surpass Penryn, none the less Neh....

These are sad days. :(

EDIT: What is Montreal?
http://download.amd.com/Corporate/Ma...AnalystDay.pdf (Page 21)
06-05-2008, 11:10 AM
CBONE

Quote:

Originally Posted by Nedjo

Anand has proven to be less, and less informed about what's going on in this industry (I still remember his ridiculous CeBIT "diaries")

This sentence: "2009 should hold a new architecture for AMD, which is the only thing that could possibly come close to achieving competition here" further prove that! As far as I'm aware (maybe I've missed some announcement) but AMD for 2009 only plans to further shrink/refine K10 architecture...

I don't think he means to say that their roadmap does contain a new architecture, but that it should (needs to) contain one to try and make this a competitive race.
06-05-2008, 11:28 AM
savantu

Quote:

Originally Posted by eleeter

...

I started that thread over at AMDzone.com. I have been a very strong AMD supporter for years, but I see the writing on the wall. Scientia is a very, very knowledgeable person and has excellent technical understanding. But he also tends to find the silver lining for AMD more often than not. I agree with a lot of what he says yes, but in this instance I think he is grasping a bit. AMD IS in trouble. Their last bastion, the server CPU with IMC and multi-socket scaling due to HT is coming to an end.

:rofl::ROTF::wierd:

This has to be the joke of the week.

Quote:

I really think AMD's biggest problem right now is their fab tech. If they had their ducks in a row, K10 would probably be running at above 3Ghz with decent power consumption....

How come ? According to your mentor Scientia , AMD has the best FABs and process in the world , their APM supermumbojumbo ,etc...

Moving on to more serious matters , you're partly correct.AMD's process tech on 65nm is actually pretty decent ( although it has 10-15% lower transistor performance than Intel IIRC ).Their problem comes from scaling , or its lack thereof.AMD's 65nm process uses thicket gates than Intel in order to keep leakage under control.By doing so you get excellent power characteristics at low clocks , but it hurts badly when you need high frequency.Basically , once you exit the sweetspot and try to go higher , the power consumption sky rockets.

K10 was designed to run at 2.2-2.8GHz@95w.In reality at 2.3Ghz it needs 95w , 125w for 2.5GHz and 140w for 2.66GHz.Part of the blame is on the process , part is on the design.
By reusing the K8 core , which is pretty complex anyway , and further increasing the complexity ,they've basically said goodbye to any frequency increase.
06-05-2008, 11:35 AM
savantu

Quote:

Originally Posted by GoThr3k

i dont know, maybe i missed it, i think singlethread nehalem wil have a 10-15% advantage, but not more, on the multithreaded side however :banana::banana::banana::banana:

In multithreaded scenarios , certain parts ( like the L3 ) need to be shared and as a result of arbitration there is an increase in latency.

Quote:

Originally Posted by Nedjo

c’mone Browncoat, do you honestly think that Intel isn’t behind this!?
It’s clear as a day that sole purpose of this “review” is to make anyone considering buying AMD’s 4-way Barcelona, or making plan for 4-way Shanghai, to rethink idea once more, and hopefully go in idle ‘till Nehalem based Xeon is out!

...

IMO you're seriously delusional if you think Intel wants to send a message to its server customers through Anandtech. :ROTF:
The important ones already have Nehalem servers for testing.
06-05-2008, 11:44 AM
MrHydes

Quote:

Originally Posted by Hexus

..the immutable fact remains that AMD's fastest quad-core offering is slower than Intel's slowest..

that's a hard true to swallow and too bad for final consumers :down:
06-05-2008, 11:53 AM
therightway

Quote:

Originally Posted by Nedjo

Anand has proven to be less, and less informed about what's going on in this industry (I still remember his ridiculous CeBIT "diaries")

This sentence: "2009 should hold a new architecture for AMD, which is the only thing that could possibly come close to achieving competition here" further prove that! As far as I'm aware (maybe I've missed some announcement) but AMD for 2009 only plans to further shrink/refine K10 architecture...

Anandtech is far and away the best review site on the net. If you question their quality of reviews, then every review site sucks. Don't tell me you read the crap from hardocp.
06-05-2008, 11:55 AM
ceevee

Quote:

Originally Posted by jaredpace

clock for clock appears to be ~ 36% increase over Penyrn in general.

nope. i ran the benches of his that i could easily duplicate and it looks more like 10-15% improvement clock for clock depending on the benchmark.

someone said this is all because i was running vista 64, but i find it hard to believe that me running vista 64 is going to make my penryn performance 10-20% better clock for clock on every single benchmark used than his penryn performance. :rolleyes:

his posted penryn results are extemely low and unrealistic, especially because the rig i am using has not been tuned and optimized at all yet.
06-05-2008, 12:04 PM
Omastar

Article won't load at the moment. :(
06-05-2008, 12:06 PM
terrace215

Quote:

Originally Posted by ceevee

nope. i ran the benches of his that i could easily duplicate and it looks more like 10-15% improvement clock for clock depending on the benchmark.

someone said this is all because i was running vista 64, but i find it hard to believe that me running vista 64 is going to make my penryn performance 10-20% better clock for clock on every single benchmark used than his penryn performance. :rolleyes:

his posted penryn results are extemely low and unrealistic, especially because the rig i am using has not been tuned and optimized at all yet.

Uh, did you notice that the Nehalem results used alpha silicon on a CRIPPLED motherbaord with only 1 of 3 memory channels working?
06-05-2008, 12:07 PM
LowRun

Quote:

Originally Posted by therightway

Anandtech is far and away the best review site on the net. If you question their quality of reviews, then every review site sucks. Don't tell me you read the crap from hardocp.

Then i suppose every review site sucks cos' Anand produced some utterly crap reviews already, the A8R-MVP being one, where they labeled the board OCer's dream and couldn't even figure the board was reverting to 2T when they thought it was 1T :rolleyes:
06-05-2008, 12:07 PM
BrowncoatGR

He was talking specifically about Cinebench and it's true that the 64bit version performs better.. Though i'd bet that your right and Nehalem is about 10% faster in single threaded apps
06-05-2008, 12:09 PM
gallag

Quote:

Originally Posted by ceevee

nope. i ran the benches of his that i could easily duplicate and it looks more like 10-15% improvement clock for clock depending on the benchmark.

someone said this is all because i was running vista 64, but i find it hard to believe that me running vista 64 is going to make my penryn performance 10-20% better clock for clock on every single benchmark used than his penryn performance. :rolleyes:

his posted penryn results are extemely low and unrealistic, especially because the rig i am using has not been tuned and optimized at all yet.

Are you being serious? can you not think of the variables that could give two different computer set-ups different results?
06-05-2008, 12:19 PM
TG01

Quote:

Originally Posted by gallag

Are you being serious? can you not think of the variables that could give two different computer set-ups different results?

Please do tell...

I'm very interested at the variables that would yield a 10%-20% performance increase between two computer systems running the same processor at the same clock speed...
06-05-2008, 12:29 PM
jas420221

Quote:

Originally Posted by TG01

Please do tell...

I'm very interested at the variables that would yield a 10%-20% performance increase between two computer systems running the same processor at the same clock speed...

Mobo, Memory, backround apps, processes...he didnt list his clockspeeds...shall I go on? :shrug:
06-05-2008, 12:29 PM
savantu

Quote:

Originally Posted by TG01

Please do tell...

I'm very interested at the variables that would yield a 10%-20% performance increase between two computer systems running the same processor at the same clock speed...

10% out of the hat for Cinebench under Vista 64 vs. 32

http://www.extremetech.com/article2/...2280813,00.asp
06-05-2008, 12:33 PM
Periander6

Quote:

Originally Posted by savantu

10% out of the hat for Cinebench under Vista 64 vs. 32

http://www.extremetech.com/article2/...2280813,00.asp

Oops, looks like that one was shot down in a hurry.
06-05-2008, 01:03 PM
Donnie27

Quote:

Originally Posted by Periander6

Oops, looks like that one was shot down in a hurry.

Yepp but they have short memories LOL!
06-05-2008, 01:13 PM
gallag

Quote:

Originally Posted by TG01

Please do tell...

I'm very interested at the variables that would yield a 10%-20% performance increase between two computer systems running the same processor at the same clock speed...

Man , you were being serious:ROTF:. Others have saved me pointing out the obvious.
06-05-2008, 01:20 PM
ceevee

Quote:

Originally Posted by jas420221

Mobo, Memory, backround apps, processes...he didnt list his clockspeeds...shall I go on? :shrug:

QX9770 @ stock 3.2Ghz
8GB RAM @ 1066Mhz
Foxconn Blackops

its in my sig. it is NOT at all tuned atm due to problems with the MB.

anyway i did the math so that i could see what clock for clock difference would be.

the only two tests I duplicated were cinebench and pov-ray. all anand's other tests were proprietary and he does not give the info needed to duplicate them, but I saw the exact same result on pov-ray. anands penryn results were way too low on this one as well.

is pov-ray also 10% better on vista 64?? :rolleyes:
06-05-2008, 01:28 PM
Shintai

Quote:

Originally Posted by ceevee

is pov-ray also 10% better on vista 64?? :rolleyes:

If 64bit yes.
06-05-2008, 01:28 PM
Donnie27

Quote:

Originally Posted by GoThr3k

i dont know, maybe i missed it, i think singlethread nehalem wil have a 10-15% advantage, but not more, on the multithreaded side however :banana::banana::banana::banana:

I tend to think the performance gap will be app dependant. These might range from 8% to 50%.
06-05-2008, 01:28 PM
fiveprime

I think I am going to wait till after the big release to purchase, one, because I want to wait and see what a little maturation will do, as well as (hopefully) some price rudection, and seecondly I wont have the cash till then regardless. The real question is when to upgrade my graphics, before or after Nehalem.

The one thing that can be said about this chip, is it is a culmanation of all current cpu technology , regardless of who or where it came from, Intel got it on the die first.

Show 100 post(s) from this thread on one page