backtop


Print 77 comment(s) - last by ditkaamdVSinte.. on Jun 2 at 6:05 PM

DailyTech takes a look at Intel and AMD's claims when it comes to server processor performance, pricing.

With the recent release of Intel's Nehalem-based Xeon processors last month and AMD's Shanghai-based Opteron processor in November 2008, a war of words has precipitated between Intel and AMD.  Intel claimed that it maintained the greatest performance-per-dollar and that its slowest Nehalem was faster than the fastest Opteron, while AMD refutes this and claimed that it provides a better value product.  AMD also claims to have the most energy efficient chip by advertising the world's most efficient and first "40-watt" quad-core CPU while Intel claims that its CPUs are the most energy efficient. 

The result of this public bickering is that it is difficult for chip customers to figure out who to believe.  To settle this dispute, we will analyze each of these claims and provide hard data to validate or debunk each of these claims.

Claim #1:
The slowest Intel Nehalem is faster than the fastest AMD Shanghai Opteron processor.

Reality: No, yes, and not anymore.
If we include the one dual-core Nehalem variant, the answer is no because a lower clock speed dual-core Intel Nehalem isn't going to beat the highest clocked quad-core AMD Shanghai.  If Intel was only talking about all of the other quad-core Nehalem models then yes, even the slowest 2 GHz Nehalem XEON model E5504 will soundly beat the fastest 2.7 GHz AMD Shanghai Opteron model 2384.  But this has changed recently as AMD has just recently released a 3.1 GHz model 2393SE Opteron which would likely be faster than the slowest Nehalem quad-core processor but still lag far behind most Nehalem model numbers.

Table 1a and 1b below shows the latest and best performance scores published at SPEC.org as of 4/27/2009.  SPECint is a general purpose benchmark covering a lot of mainstream applications while SPECfp represents a cross section of scientific and engineering applications.  SPECweb reflects web server performance.  For comparison sakes, we'll also include the fastest Nehalem Xeon 2.93 GHz X5570 which on average is nearly twice as fast as the Opteron 2384.  Since official public benchmarks aren't available at the time of this writing, table 1 below will not include model 2393SE performance numbers.  However, the performance gain of the Opteron 2393SE will likely be no more than 15% faster than an Opteron 2384 which may be enough to put it past the Intel E5504.

Table 1a: SPEC CPU 2006 base performance results as of 4/27/2009

Processor

Chip price

GHz

Cores

SPECint

SPECfp

Xeon E5504

$235

2.0

2x4

127

111

Opteron 2384

$957

2.7

2x4

113

105

Xeon X5570

$1470

2.93

2x4

241

194

The SPECfp data is especially noteworthy because this was traditionally an AMD stronghold before the arrival of Intel's Nehalem processor with QuickPath memory architecture.  SPECfp along with everything else is now clearly dominated by Intel Nehalem.

Table 1b: SPEC WEB 2005 performance results as of 4/27/2009

Processor

Chip price

GHz

Cores

SPECweb

Opteron 2384

$957

2.7

2x4

39,530

Opteron 8384

$1865

2.7

4x4

48,007

Xeon X5570

$1470

2.93

2x4

71,045

SPECweb was another strong suit for AMD before the arrival of Intel Nehalem.  Now a dual-socket Nehalem based Xeon server can even outpace a four-socket Shanghai based Opteron server.

Claim #2:
AMD launched the world's first quad-core processor within a 40-watt thermal envelope and has the most energy efficient processors.

Reality: False
AMD did not release a quad-core processor within the 40-watt thermal envelope and they do not have the most energy efficient chip.  While AMD never officially claimed that their newest Opteron 2377HE chip runs inside a 40-watt thermal envelope, this is the impression they have successfully marketed to the public.  AMD now refuses to disclose the actual Thermal Design Power (TDP) of their latest CPUs to the media and has tried to state that AMD's Average CPU Power (ACP) metric is comparable to Intel's TDP metric.  As a result, every news story, every product label, and every server vendor only show ACP numbers and not the actual thermal ceiling of the processors while Intel continues to use the TDP metric which more accurately reflects peak power consumption.

AMD also claims that Intel's processors have a higher thermal ceiling which makes them less suitable for dense data center deployments, but even this claim is false for most Nehalem models.  The most comprehensive and standardized server power efficiency metric to date is SPECpower_ssj2008 and it shows a wide range of power consumption metrics across a variety of load levels.  Table 2 below will show a comparison of the most efficient AMD Shanghai based servers against the most efficient Intel Nehalem based servers.  SPECpower_ssj2008 is so interesting because it not only tells us the actual measured idle and peak power consumption under a server side java load, but it also tells us the server's server side java performance.  The "score" is determined by an average measurement of performance per watt across all the workloads.  SPECpower scores for the 2.0 GHz E5504 were not available, so the closest model number with the most similar clock speed and power consumption characteristics were used in table 2.

Table 2: SPECpower_ssj2008 results as of 4/27/2009

Processor

Chip price

ACP

TDP

GHz

Cores

Score

SSJ_OPS

Measured system power

Peak

Idle

Xeon L5520

$560

 

60 W

2.26

2x4

1813

439,831

170 W

66.3 W

Xeon X5570

$1470

 

95 W

2.93

2x4

1860

620,270

244 W

81.4 W

Opteron 2376HE

$500

55 W

79 W

2.3

2x4

1044

346,326

210 W

119 W

Opteron 2384

$957

75 W

95 W

2.7

2x4

860

338,577

264 W

121 W

As we can see from the official SPECpower_ssj2008 results above, AMD's "High Efficiency" (HE) products actually consume substantially more power than Intel's low power products despite the fact that AMD advertises the lowest power consumption.  The idle power consumption gap is even larger which is significant since most servers spend most of their time low utilization states.  It's clear that the million plus transistors Intel devoted to power management on the Nehalem chip is paying off.

While these measured peak power numbers don't reflect the maximum power that each system can draw under any workload, it does represent the peak power consumption under this particular type of workload which is generally not as intensive as a High Performance Computing (HPC) workload used in scientific and engineering applications.  What these numbers do confirm is that AMD's claim that their ACP metric is most similar to Intel's TDP metric is clearly false and that AMD's TDP ratings for these processors are far more comparable to Intel's TDP.

It's also noteworthy that this benchmark also shows lower clocked Intel processors performing more server side java operations per second (SSJ_OPS) than the higher clocked AMD processors which further reinforces the data from Table 1.

Claim #3: An equivalent Intel Nehalem server costs almost twice as much as an equivalent AMD Opteron Server.

Reality: This claim is false because there are no equivalent (in performance or power consumption) AMD Opterons compared to Intel Nehalem servers.  While AMD based servers are the cheapest, a low end Nehalem server can deliver more performance at a lower price.  Table 3 below shows server prices from HP with the base configuration plus second processor.

Table 3: Server price comparison

Server name

CPU

Cores

GHz

Memory

Server Price

HP DL380 G6

Intel E5520

2x4

2.26

6 GB

$3,448

HP DL380 G6

Intel X5550

2x4

2.66

6 GB

$6,039

HP DL385 G5p

AMD 2382

2x4

2.6

8 GB

$2,619

HP DL385 G5p

AMD 2384

2x4

2.7

4 GB

$4,039

It appears that AMD is already heavily adjusting their processor prices to the major server makers rather than the unrealistic list prices that were originally set before the arrival of Intel Nehalem.  In fact if we only went by the official processor list prices, we would expect the AMD servers to cost much more than they do in table 3.  It is normal to see a lag in price adjustment immediately after a new chip lunch when customers haven't all caught on to the new products yet and when inventory is still working its way into the channel.  However, we should expect more AMD price cuts to come soon because there's no reason that a server based on an Opteron 2384 should fetch more money than a server based on Intel Nehalem server given the benchmark results.

Implications on the high end multi processor server market

The implications of Intel Nehalem on the server market for the near term are profound.  It not only threatens AMD in the two-socket market, but it threatens AMD's four-socket market and even begins to encroach on Intel's four-socket Dunnington servers.  When you can buy a two-socket Intel Nehalem server at a fraction of the price and exceed the performance of four-socket AMD servers and come close to Intel Dunnington, it takes the thunder out of four-socket servers.  Table 4a and 4b shows two-socket Intel Nehalem servers beating Opteron four-socket servers in SAP and Virtualization performance.

Table 4a: SAP performance *

Processor

Chip price

GHz

Cores

SAP

Opteron 8384

$1865

2.7

4x4

440,000

Xeon X5570

$1470

2.93

2x4

510,670

 

Table 4b: VMware virtualization performance *

Processor

Chip price

GHz

Cores

VMmark

Opteron 8384

$1865

2.7

4x4

20.35 @ 14 Tiles

Xeon W5580

$1707

3.2

2x4

23.96 @ 16 Tiles

These results are stunning because at no time in the last 5 years has the performance gap between Intel and AMD been so large.  This situation won't change until AMD launches 6-core Istanbul Opteron processors which are expected by June 2009 and Intel launches its 8-core Nehalem-EX processor for multi-socket severs expected by the end of Q4 2009.  AMD 6-core Istanbul will narrow their performance deficit not enough to close it against Intel current Nehalem-EP 4-core processors.  However, Intel's Nehalem-EX 8-core processor is expected to widen Intel's performance lead by a significant margin by the end of this year.  But because Nehalem-EX won't be seen in most mainstream two-socket servers, Istanbul will allow AMD to raise their average selling price to a healthier level.

Decision time

It appears that the server market between Intel and AMD is now mirroring the desktop chip wars.  IT customers have a choice of buying low-cost value servers based on AMD Shanghai processors or they can buy premium priced Intel-based servers for better performance and energy efficiency.  However, the value hunter should wait a few more weeks when Shanghai servers settle down to post-Nehalem prices.

* All results and prices as of 4/27/2009





Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Not Suprised...
By ExarKun333 on 5/4/2009 11:08:48 AM , Rating: 2
Isn't this what the Nehalem architecture was made for? These new Xeons are looking VERY good. From the charts in the article, you can choose a lower-end Xeon for less price and similar (or better) performance or a higher-priced option for amazing performance. Those are some great options for anyone looking for a server right now.




RE: Not Suprised...
By themaster08 on 5/4/2009 11:20:54 AM , Rating: 2
Agreed.

With that said though, things are really looking bleak for AMD in a market they not so long ago prospered from.

All of their claims are nothing but FUD. I think most of those interested in purchasing servers are wise to that sort of criterion.


RE: Not Suprised...
By MrBungle123 on 5/4/2009 11:50:05 AM , Rating: 2
quote:
All of their claims are nothing but FUD. I think most of those interested in purchasing servers are wise to that sort of criterion.


No doubt.

Since the introduction of core 2 AMD has been getting walked all over by Intel and AMD's responses kind of remind me of Kim Jong Ill... always superior and capable of virtually anything. (In his own mind anyway)


RE: Not Suprised...
By themaster08 on 5/4/2009 12:03:10 PM , Rating: 2
Definitely. AMD have been on the backfoot in the general consumer market in performance ever since the Core2 architecture was released, but they still remained extremely competitive in the server market due to their use of an IMC in their processors, which Intel sorely lacked until now.

The use of QPI in the Nehalems is just what Intel needed to more-or-less obliterate AMD from being remotely competitive when it comes down to performance.

Pricing on the other hand will be a completely different matter. I'm sure AMD will get on to pricing their chips very competitively, as they have been doing in the general consumer market.


RE: Not Suprised...
By MrBungle123 on 5/4/2009 12:28:23 PM , Rating: 5
quote:
Pricing on the other hand will be a completely different matter. I'm sure AMD will get on to pricing their chips very competitively, as they have been doing in the general consumer market.


Pricing can only get you so far. At some point performance becomes the deciding factor. I mean yeah I can buy Pentium 3's for $4 each and I could easily get more raw processing power down that route (theoretically anyway) for the same money that I paid for my core 2 quad but in practice the single thread performance delta is just too great to even make the Pentium 3 solution a viable option.

I wonder how long it will be before AMD runs into a similar issue with their product line? The current Phenom II stuff (not sure what the codename for the Opteron variant is) even with the price cuts is just barely competitive with the core 2 line, Nehalem is just getting started... If you think about it AMD is 2 generations behind. At some point they are simply going to be offering parts that are too slow to be useful and their appeal even when they are substantially cheaper will start to taper off.


RE: Not Suprised...
By mcnabney on 5/4/2009 2:08:16 PM , Rating: 2
Chips too slow for what?

The Atom is the current hot item right now, and that is anything but powerful. Only some server functions need exceptionally fast CPUs.
If AMD is smart they will target their products toward specific usage. When I built a Home Server I grabbed a 4850e which featured low power, an excellent price, and a very low system price. I am sure that Intel will continue to do business with the spoiled teenagers and trust-fund babies, but since gaming has fled to the consoles the need and market for large numbers of high performance PCs is dwindling. And I say that as a PC gamer.
Ultimately, AMDs failure to compete across all product lines will mean a slower pace for innovation and Intel holding back newer products so as not to canibalize their existing sales.


RE: Not Suprised...
By themaster08 on 5/4/2009 2:37:44 PM , Rating: 3
quote:
Chips too slow for what?

To compete with Intels, of course.

quote:
The Atom is the current hot item right now, and that is anything but powerful.

We're talking about server CPU's here, so this is irrelevant.

quote:
Only some server functions need exceptionally fast CPUs.

But they're all in need of energy efficient CPU's, and the Intels prove more so compared to their AMD counterparts, as well as the increased performance in comparison.

quote:
I grabbed a 4850e which featured low power

If you have read the article properly, you will have read the part about these so-called "energy efficient" models, and that Intel equivalents are more energy efficient even with higher TDP's.

quote:
Intel holding back newer products so as not to canibalize their existing sales.

If Intel didn't hold back, AMD would have disappeared a long time ago. Intel are already moving much too fast for AMD to keep up. This shows with their new line of Phenom II processors, which, don't get me wrong, are excellent processors, especially for the price, but lag in performance compared to a 2 year old QX9650.


RE: Not Suprised...
By just4U on 5/6/2009 2:12:20 AM , Rating: 2
Lag in performance? those PII's seem to be doing a bang up job againts intels 7,8, and 9x lines which is where we see the bulk of Intel's desktop sales. The i7 on the other hand hasnt quite hit the masses yet and likely won't until the i5 is introduced. Even then your still only looking at what?? 20% gains at best, so it's not like the performance increases we got out of the core2.

As to the whole server side of things, you can bet AMD will feel the pinch. Server sales are fairly lucrative and if they get shut out there that's going to hurt.


RE: Not Suprised...
By themaster08 on 5/6/2009 6:41:15 AM , Rating: 2
quote:
Lag in performance? those PII's seem to be doing a bang up job againts intels 7,8, and 9x lines which is where we see the bulk of Intel's desktop sales.

I agree, but my whole point is that the Core 2 Quads are well into their second year of availability, and AMD have only just released something to compete with the higher end Core 2 Quads.

If you've read most reviews, you will see that even the latest Phenom II 955 3.2GHz is more comparable in performance to a Q9550 2.83GHz, and that when compared to a QX9650/Q9650 3.0GHz, QX9770 or perhaps even a QX6950 which is 2 years old, and which are all Core 2 Quad, it trails in almost every benchmark.

The entire point was that if Intel moved any quicker than what they are already doing, then who knows where AMD would stand today.


RE: Not Suprised...
By MrPoletski on 5/6/2009 2:46:23 PM , Rating: 2
hehe I don't think we'll be seeing atom based servers any time soon, but then I don't think you meant that;)

to me, it looks like it's ratcheting up to another price war between the two which, well, just plain rules really.


RE: Not Suprised...
By winterspan on 5/4/2009 4:16:40 PM , Rating: 5
I think a far more important issue that is similar to what you bring up is that pricing can only get you so far when you are talking about chips that use more power per performance unit than Nehalem particularly if the idea was to buy more cheap opterons to equal the performance of a smaller number of more expensive Nehalems. Power is very imporant these days as electricity bills and cooling costs are high and companies want to be as green as possible.

Unfortunately, I didn't realize AMD was in such a bad position in regards to performance/watt. The only way I could see AMD attenuating the crushing onslaught of 2P/4P Nehalem was if they could win the power and efficiency argument, and it is clear that despite AMD's marketing, Nehalem is just an incredible performer in every metric.

Lastly, having seen all the Nehalem numbers for a while now, I can only imagine how insanely fast the 4P Nehalem 'Beckton' will be with quad-channel RAM and quad-channel quickpath.


RE: Not Suprised...
By jrb531 on 5/9/2009 11:48:02 AM , Rating: 2
It depends on the apps...

In most games it matters VERY little if you are running Intel or AMD and the FPS load is all on the Video Card so why pay for something you do not need?

I've also been able to keep the very same motherboard for two years!

Starting with a single core AMD then moving to 2x core then Phenom I and now a Phenom II - all on the same motherboard!

I did not have to change my memory as the "myth" of DDR3 being better than DDR2 is just that... myth!

So while I agree that Intel "currently" makes the best chips... you do pay for them and when I buy or build a system I look at the best speed for your money.

While some people get some form of pleasure trying to show who has the largest (censored) by means of benchmarks... the "smart" people laugh all the way to the bank and take that money they saved by not buying a $300+ Intel CPU and using it toward a faster Video Card.

If money was no object then of course I would get an Intel but I have more important things to spend my hard earned dollars on instead of buying a CPU that never gets more then 50% use in the games I play.

Looks great in benchmarks however :)


RE: Not Suprised...
By themaster08 on 5/10/2009 3:53:30 AM , Rating: 2
quote:
I've also been able to keep the very same motherboard for two years!

Starting with a single core AMD then moving to 2x core then Phenom I and now a Phenom II - all on the same motherboard!

I agree that AMD's upgrade solutions offer excellent compatability and value for those who already own an AM2+ motherboard. With that said though, two years ago I purchased a motherboard and processor (Core 2 Quad), and I'm still using that same motherboard and processor.

I haven't had to constantly upgrade because the performance has always been above par and above anything AMD have had to offer until very recently.

With your constant processor upgrading, ultimately you have paid more than I have, to get more-or-less the same performance as I've had for the past 2 years.

In that scenario, AMD no longer looks like the cheaper option, does it?

quote:
While some people get some form of pleasure trying to show who has the largest (censored) by means of benchmarks

It might have escaped your fanboy brain, but benchmarkers only accomodate for a tiny proportion of the market. If only benchmarkers bought Intel processors then they would not be absolutely dominating the processor market, would they?

quote:
While some people get some form of pleasure trying to show who has the largest (censored) by means of benchmarks... the "smart" people laugh all the way to the bank and take that money they saved by not buying a $300+ Intel CPU and using it toward a faster Video Card.

Ironically enough, it's the Intel users who are laughing at the AMD users. It's true that some may have paid $300+, but their need for constant upgrading is not necessary as the performance is enough to get the user through at least the next two years.
The "smart" people buy a processor that will last them as long as possible without the need to upgrade. AMD obviously don't offer a viable solution to this, which shows due to the constant changing of their processor lineup.

Whilst it's true that some may have paid $300 for their processor, there are Core 2 Quad solutions which are much cheaper. Suffice to say, that they have got more than their moneys worth. I paid £160 for a Core 2 Quad Q6600 two years ago and never looked back, whilst in that same amount of time you've forked out for 3 processors. Ultimately, I'm the one laughing at you.

quote:
and using it toward a faster Video Card

Yeah, because as usual people only buy a computer to play games.

quote:
but I have more important things to spend my hard earned dollars on

Like what? Saving up for your next processor upgrade in 2 months time? Please....


RE: Not Suprised...
By Calin on 5/4/2009 12:40:59 PM , Rating: 2
I tend to disagree
Only the recent server processors from Intel (with integrated memory controller ) were able to dethrone AMD from its position in server processors. As for desktop processors, Intel was better with Core2Duo, and is better now (the Phenom 2 are competitive with the latest Core2Duo processors, though)
Now, AMD is really in a bad spot as its most profitable (per unit) market is being lost.
However, keep in mind that Intel is administering the same beating to its older processors/architectures


RE: Not Suprised...
By psychobriggsy on 5/5/2009 7:49:22 AM , Rating: 2
Some of the writing in this article is straight from Intel's marketing handbook. The rubbish about TDP for example - AMD used to use TDP (Voltage x Amperes) whilst Intel use an achievable (thermal bug) value which is around 75% of the aforementioned value. AMD switched to their new system to match Intel's definition. I think that Intel's definition is reasonable to use, but don't go and start claiming that AMD's redefinition to something similar is somehow dishonest!

Also the scaling on the SPECWeb results for the quad CPU Opteron is highly suspect given that AMD's platform scales very well (and indeed there is no quad-socket Nehalem offering currently). Of course Nehalem performs excellently, but something is really fishy with the scaling for AMD's results, and I suspect that there are platform differences. Again this could simply be because AMD need to move to HT3 on their server platform.


RE: Not Suprised...
By MonkeyPaw on 5/4/2009 3:33:35 PM , Rating: 2
quote:
With that said though, things are really looking bleak for AMD in a market they not so long ago prospered from.


Ironically, it required Intel to adopt AMD's tricks to get there. AMD's IMC+HT is what made the difference in the server world, and now that Intel has mated the same concepts to their superior Core2 (and cache) design, AMD no longer holds all the cards in 4-way+ configurations. What AMD does have still going for it is drop-in compatibility, which is still a decent reward for its socket F clients (relatively cheap upgrades for significantly more power). That will only help them for a little while, as Socket F will be replaced by something with more room to grow(I think quad-channel memory and support for up to 16 core CPUs).

What we can all thank AMD for is the twarting of Itanium--Intel's original 64bit design for the future. Intel has sank billions into IA32/64, but it still remains a niche product. The scalability of i7-based Xeons should all but eliminate the attraction of IA architecture. Perhaps Itanic will finally sink?


RE: Not Suprised...
By ExarKun333 on 5/4/09, Rating: 0
RE: Not Suprised...
By MonkeyPaw on 5/4/2009 7:08:02 PM , Rating: 5
How about reading my post before you light the torch? Nowhere did I say AMD invented ANYTHING. I said that AMD's approach (IMC+HT) was a winning combination in the x86 server space, one that Intel had to duplicate. Besides, it's not just the IMC or HT alone that made the Opteron a hit, it was having both of these features together that solved the memory bandwidth and CPU communication issues that plagued x86 servers before then.

Also, do more research. Intel didn't "pioneer" the IMC either, as Alpha (Intel later purchased Alpha's IP) and HP both had CPUs with integrated memory controllers in the early 90s. Timna, Intel's first IMC chip, was built in the late 90s. Timna was not canceled because it "was not needed," rather it had a flawed memory controller, so the concept was scrapped and boards recalled. Lastly, AMD doesn't pay royalties to Intel for Hypertransport (HT), as HT was created by an open consortium of developers, which included AMD. Intel's version of Hypertransport is called QPI.


Check those VMmark Scores...
By solori on 5/4/2009 8:03:21 PM , Rating: 5
RTFM on VMmark scores:

quote:
If two different virtualization platforms achieve similar VMmark scores with a different number of tiles, the score with the lower tile count is generally preferred. The higher tile count could be a sign that the underlying hardware resources were not properly balanced. Studying the individual workload metrics is suggested in these cases.


(see http://www.vmware.com/products/vmmark/faq.html)

Likewise, the current VMmark score for 8300 series is:

quote:
22.11@15 tiles


based on an Opteron 8393SE ($2,649/ea). If you want a "quick" comparison using VMmark, divide the cost of the system ($) by the VMmark through-put (i.e. 22.11) and multiply by the number of tiles (i.e. 15). This gives you a good rule-of-thumb cost index. The cost index on Nehalem-EP vs. Quad-Opteron is $2,800 higher (more costly) per equivalent tile-load.

Now why would I bother comparing a $2700 processor to a $1500 processor? For the same reason someone would compared a $15K/DDR3-1066 sub-system to a $2,400/DDR2-800. That's right, currently it takes $15K in DDR3-1066 memory to perform the same amount of work that $2,400 in DDR2-800 will do. Small cash-in-pocket advantage to Opteron.

Likewise, top-out the memory footprint of your Nehalem-EP to 144GB and you'll pay $21K to the memory fairy (while dropping to DDR3-800 speed). Opteron 8300 will stretch to 160GB at DDR2-667 and at least 256GB DDR2-533 within a $10K or $16K budget.

Clearly, this is due to the outrageous cost of DDR3-1066 memory, but much of the performance difference between Opteron and Nehalem is due to the memory bandwidth advantage Nehalem has over Opteron. Perhaps this is why there are no VMmark scores listed for Nehalem with 144GB of memory.

Speaking purely from a virtualization standpoint, two systems for the same money is a serious advantage - especially on the low-end. Economically, Nehalem-EP's reliance on DDR3 cripples its overall appeal for large memory applications. The move to quad-channel DDR3 in Nehalem-EX will resolve some of these issues...

As for power measurements, on the high-end - where data center consolidation is a play - the current "ultimate" power savings is when a server is off. Idle-to-off (DRS/DRM in vSphere 4.0, for example) makes idle performance irrelevant and not "real world." Loaded VMware servers DO run on average power utilization statistics - especially when in a DRS environment (i.e. very large scale virtualization.)

For comparative 32GB-Opteron and 24GB-Nehalem (2P) systems configurations, Opteron weighs-in at 35% less CAPEX. If you look at "average" consolidation factors - i.e. 3:1 - at $245/VM vs. $370/VM. Nehalem has a good "speed" advantage with DDR3/1333 vs. DDR2/800 but this is consolidation - not performance.

Likewise, for 144GB/DDR3-800 Nehalem vs. 160GB/DDR2-667 Opteron in SQL consolidation systems Opteron enjoys a 25% price advantage and can run two more instances of SQL (8GB/instance). If your DBA says "1vCPU/core" you jump to 56% advantage Opteron unless you want to fight the HT battle with him...

It is true: the more modern architecture of Nehalem has definite advantages over AMD's older Opteron architecture. Socket G34 is still 9 months in delivery and Nehalem-EP won't have a rival until then for the performance trophy. However, on the CAPEX and price/performance side in virtualization environments - there is a street fight and it is currently in AMD's favor.

If AMD's Istanbul delivers 50% more cores with only a 15% price premium, Nehalem-EP can not compete unless something drastically shifts in the DDR3 food-chain. Likewise, Nehalem-EX must ship with twice the memory capacity of -EP or it will have trouble against Magny-Cours/G34 and it's 4-channel DDR3 in 2P and 4P configurations.

One additional note: Intel has always been the best at chasing benchmarks. How else do you explain the "hammering" that it has given AMD's processors over the years in benchmarks while delivering modest losses in "real world" performance? Yes, the Core/Nehalem architecture is an excellent platform, but is worth that much more money?

This is 2009 with record deficits and unsure sales pipelines. Virtualization creates multiple system purchases which only server to multiply cost factors - not divide them. In a "theoretical" world dominated by SMT tricks and IPC performance, the Nehalem-EP is a monster. In the "real world" of $/VM (both OPEX and CAPEX) the race is not over by a long shot.

See my related blogs at: http://solori.wordpress.com




RE: Check those VMmark Scores...
By GeorgeOu on 5/4/2009 8:29:56 PM , Rating: 1
You're right that DDR3 is more expensive. However, are you being completely honest in your memory price quotations? Sure you can quote memory prices from the big server vendors, but no one in their right mind buys memory upgrades from their server vendor and if they worked for me, I'd fire them. You always buy third party memory (same place the big server vendors gets their memory) and you pay 4 to 8 times less per GB of memory.

At the end of the day, DDR3 is still more expensive but it's not nearly as big a factor in overall system price as you make it out to be.


RE: Check those VMmark Scores...
By Amiga500 on 5/5/2009 2:26:49 AM , Rating: 2
quote:
However, are you being completely honest


Ironic... very ironic.


RE: Check those VMmark Scores...
By GeorgeOu on 5/4/2009 9:10:30 PM , Rating: 2
I checked your link and your claim that DDR3 costs $10,000 is blatantly dishonest and assumes people are dumb enough to buy from the most expensive vendor that marks up prices 800%.

I ran a quick google search and found http://www.provantage.com/kingston-technology-kvr1... 4GB DDR3 DIMMs with ECC for $102.26. That means with 18 4GB DIMMs for a total of 144 GB of RAM, you'll spend $1840.68 on all new memory.


RE: Check those VMmark Scores...
By eldakka on 5/4/2009 10:17:47 PM , Rating: 5
In my experience, most large organisations (government departments, outsourced datacentres from the likes of CSC, EDS/HP etc) will ONLY purchase memory from the server vendor.

You'd get sacked in most large organisations if you bought cheap 3rd party memory for their servers rather than from their Sun/HP/IBM/Dell server suppliers.

Hell, it'd invalidate many support/maintainence contracts by not buying direct from the vendor.


RE: Check those VMmark Scores...
By GeorgeOu on 5/5/2009 12:17:05 AM , Rating: 2
You can't invalidate a support contract because someone uses third party DIMMs. There are actually state laws on the books that make these terms illegal unless the vendor can prove that the third party equipment damaged the server.

Yes there are closeminded bureaucratic departments that will buy the scare stories. It might be true for Government agencies that don't actually have to worry about budget cuts since the Government can always tax us more for more revenues. There might be really large corporate departments that have money to throw around but I'd be shocked if they weren't facing budget cuts these days.


RE: Check those VMmark Scores...
By tshen83 on 5/4/2009 10:18:13 PM , Rating: 3
Huh, George:

That is actually VERY DUMB. For an Asian, 18*4=72GB. That is a very obvious math mistake.

To get 144GB, you need 8GB DUAL RANKS DDR3(Nehalem supports 6 DIMMs of quad ranks REG, or 9 DIMMs or dual rank REG, or 6 DIMMs of dual rank UDIMM per socket) which means you either have to buy 4Gbit DDR3 ICs or use ram fudging technology like MetaRam. 144GB REG DDR3 configuration right now is probably unrealisticly expensive.

The best memory combination for Nehalem is indeed 24GB with (12*2GB UDIMM) configuration for $360. UDIMM DDR3 is at price parity vs DDR2-800 REG already.

Both Intel and AMD should take note of this. Adoption of UDIMM will trump the adoption of REGISTERED DDR3. Hopefully Nehalem-EX will take a hint at this :)


RE: Check those VMmark Scores...
By tshen83 on 5/4/2009 10:27:51 PM , Rating: 2
BTW, the Kingston kit you labeled is a quank rank DIMM, which means, Nehalem supports 6 DIMMs total per socket giving you 12*4=48GB for about $1200 dollars which isn't bad for the price.

Dual rank 4GB REG DDR3 DIMMs are about $130 dollars per DIMM. Look at ewiz.com pricing. They seems to have direct sales connections to Samsung and Hynix and Supertalent. It is sad that the memory manufacturers still don't mention much about ranks and power consumption much.


RE: Check those VMmark Scores...
By GeorgeOu on 5/5/2009 12:27:58 AM , Rating: 2
You know, there's nothing wrong with just pointing out errors or mistakes. Do you really need to bring race into this thread? Besides, I'm sure you're perfect and you've never made these kinds of mistakes.

Nehalem-EP based systems use registered or unregistered ECC DDR3 DIMMs. I believe Nehalem-EX will be using FBDIMM.


RE: Check those VMmark Scores...
By tshen83 on 5/5/2009 8:03:46 AM , Rating: 2
Dude, chill man. It was a joke. I am Asian too man. It is like black dudes calling each other niggas in a rap song. It is definitely not about race. I even praised you for writing this article below. You will take a lot of heat from amd idiots for pointing out the obvious truth that a 230 dollar e5504 is a better CPU than the highest end opteron. And Istanbul isn't the solution.

A major price restructuring is necessary for AMD in the near term followed by a complete platform overhaul.


RE: Check those VMmark Scores...
By bruce24 on 5/5/2009 9:23:19 AM , Rating: 2
Yes George he was dishonest with his memory pricing post. Maybe he though after reading his very long post people would be too tired to verify.

When you check on the actual prices of memory for a Shanghai based systems vs a Nehalem based system, you find they aren't all that far off, for example:

To get 128GB of HP memory in an AMD based DL385-G5p you need to buy 8 of HP part # 408855-B21 (16GB Kit 2X8GB Reg PC2-5300), at a total cost of $13.6K.
http://www.provantage.com/hewlett-packard-hp-40885...

To get 128GB of HP memory in an Intel based DL380-G6 you need to buy 16 of HP part # 516423-B21 (8GB 2RX4 PC3-8500R-7 Kit), at a total cost of $15.3K.
http://www.provantage.com/hewlett-packard-hp-51642...


RE: Check those VMmark Scores...
By tshen83 on 5/5/2009 9:55:00 AM , Rating: 1
8GB DDR3 DIMMs can be found for as little as 450ish now.

http://www.provantage.com/supermicro-memdr380lhl01...

Not sure if that stick is quad rank built with 2Gbit ICs or dual rank built with 4Gbit ICs. Even if it is a quad rank, you can get 12*8GB for a comfortable 96GB ram. $450*12 = $5400

However you will not use 8GB DIMMs in a 4S AMD Server. You will use 32x4GB dimms for as little as $2500($80*32). So there is some difference. I would recommend Dunningtons for anyone who really need 128GB ram.

DDR2 vs DDR3 battle:
2GB DIMMS are already at price parity. 8GB DIMMs luxury DIMMs are pretty much price parity too between AMD and Intel. There is a 20-30% price premium occurring at 4GB DIMM size(75 dollars vs 100+ dollars). I am sure the difference will be resolved very very soon considering Intel's volume economics.

I hate AMD fanboys quoting memory capacity advantage of the AMD 4S socket solution compared to Nehalem-EP. Whoopty freaking duh, 32 DIMMs vs 18 DIMMs. This advantage will be gone sooner than you think.


RE: Check those VMmark Scores...
By bruce24 on 5/5/2009 12:11:24 PM , Rating: 2
re: 8GB DDR3 DIMMs can be found for as little as 450ish now.

I provided prices for HP branded memory because someone in the thread pointed out that many buy memory from their server vendors for support and other reasons. Both DDR2 and DDR3 memory can be purchases from thrid parties for much less than the HP branded memory.



RE: Check those VMmark Scores...
By GeorgeOu on 5/5/2009 4:40:28 PM , Rating: 2
Since I made the error in the parent comment, I rescind my "blatantly dishonest" comment and appologize to Solari for it.


some things to consider
By pkhughes on 5/4/2009 2:07:21 PM , Rating: 5
Full disclosure, i am an AMD employee.

I'm not disputing any of the benchmark results that George has included in his article, those are directly from the SPEC web site.

What's interesting is that those benchmark results are based on Intel systems configured with 24 GB of memory for the SPECcpu results and 96GB for the SPECweb result. Yet, when George does his price comparison later in the article, he uses servers configured with only 6 GB of memory in the case fo the Intel servers. Seems to be a bit of a misrepresentation there, even if these are considered the "standard" configurations the OEM publish on their sites. I would ask that if George is going to highlight benchmark numbers followed by price comparisons, he stay consistent.

Also, George claims that AMD "now refuses to post the TDP" for our Opteron EE processors. At no time have we ever "refused" to disclose that info to anyone. In fact, our MAX TDP number can be found here at www.amd.com/acp. Please note, that Intel's TDP number is not a MAX TDP. Had George taken the time to look on AMD's web site for the info, he would have found this information. If there is an article out there where an AMD spokesperson has "refused" to provide this info at any time, please send it to me.




RE: some things to consider
By iheartzoloft on 5/4/2009 2:13:26 PM , Rating: 5
Bump


RE: some things to consider
By GeorgeOu on 5/4/09, Rating: 0
RE: some things to consider
By bruce24 on 5/4/2009 10:21:53 PM , Rating: 2
PKHughes, George included peak and idle power numbers in the article; unless you are disputing them, I'm not sure what you point is.

George, maybe if you take back "refused" and replace it with "make it difficult to find" he will feel better.

Personally I'm not sure why people care so much about TDP, it's a design limit. The E5580(2.53ghz) and the E5502(1.86Ghz) Nehalem chips both have a 80W TDP, yet they do not consume the same amount of power.


RE: some things to consider
By brianstretch on 5/4/2009 11:31:21 PM , Rating: 2
There are quite a few discrepancies.

On the VMWare test, the AMD test used VMware ESX 3.5 Update 3 (build 120079) while the Intel test used VMware ESX 4.0 build 148783. Might there have been some improvements in hardware VM support between the two?

The web test has the ludicrous difference of 24GB RAM on the AMD box and 96GB on the Intel box as pkhughes noted.

The SPEC CFP2006 tests are using different compilers, Intel's own on the Intel box and PGI and Pathscale on the AMD box. I'm sure that Intel wouldn't invest the man hours necessary to hand tune their compilers to score well on synthetic benchmarks...

That's just what I can see from a cursory look at the linked documents.

Look, I know that AMD really, really needs to get their new server chipsets out and that relying on Broadcom and NVIDIA was a mistake. I know that Intel can throw fantastic resources at chip design and fabrication. But with Intel's history of marketing thuggery we really need to see more precise apples-to-apples comparisons than this.


RE: some things to consider
By kenji4life on 5/5/2009 2:56:46 AM , Rating: 2
There are some serious and solidly grounded concerns about the veracity of this article. I encourage an editor of this site to review the "facts" presented, if they have any regard for the reputation of this site.

That being said, although this is one of my favorite news sites, this is definitely not the first time I have noticed serious and unchecked discrepancies with fact on various articles here.

There was a point at which readers of this site could confidently point to other tech news sites as being sensationalist, bordering on the level of a tabloid, and now I fear that this site has unfortunately in some cases included itself among those.

That being said, I still have great respect for most of the writers here and also for many of the members who participate in the discussions here. I will continue to read the articles on this site and link friends and family to interesting stories that I am likely to continue finding here.

Good day.

-Kevin Hoy


So does this mean...
By kamel5547 on 5/4/2009 11:22:27 AM , Rating: 2
That Intel processors are finally viable in 8-P boxes? That was alwayhs the sticking point, performance wise even if Xeon's beat Opteron's in 2-P configurations they were demolished in 8-P configurations due to general arhcitecture issues.

I figure thats probably fixed with the new architecture, but I would think that would have made the list of topics.




RE: So does this mean...
By themaster08 on 5/4/2009 11:41:08 AM , Rating: 2
I believe the use of QPI over FSB solves such issues found in previous Xeons.

The traffic bottlenecking limitations of FSB were holding it back when compared to AMD's server chips, which have used their own IMC, known as Direct Connect Architecture for years. QuickPath eradicates such limitations giving the CPU a lot more bandwidth.


RE: So does this mean...
By Calin on 5/4/2009 12:49:18 PM , Rating: 2
A common FSB was used for memory traffic AND for processor communication since the days of the Pentium 2 and !!!. As such, it was sorely overworked in anything over 4P. The fact that server processors had slower FSB than desktop processors did no good.
Meanwhile, AMD went with different "FSB" since the Athlon MP days (with memory controller on chipset). Also, the multiple HyperTransport link allowed processors in a 4P system to talk simultaneously and directly with each other, while Intel had to use the same FSB for everything.
As someone from Intel once said, "you can only integrate the memory controller once". Intel was late to this, but it more than made up in the overall speedup.


RE: So does this mean...
By GeorgeOu on 5/5/2009 2:14:59 AM , Rating: 2
The FSB and single memory controller (North Bridge) wasn't a big problem for single socket Core Microarchitecture systems nor was it a problem for lower clock speed dual-socket Core Microarchitecture systems. It became a liability for higher clocked dual-socket Core-based Xeon server systems, but Intel had a strong enough product to stay ahead of AMD from Q3 2006 till November 2008 on most benchmarks except HPC applications that required a lot of memory bandwidth.

Intel was going to produce a Core Microarchitecture chip with QPI and Integrated Memory Controller but that project based out of India was cancelled (around 2005 I believe). Intel mitigated some of these issues by going to DDR2-800 memory with the "Stoakley" platform and they added more cache and a snoop filter to their Dunnington chip used in multi-socket systems. That was enough to keep Intel in the lead until the November 2008 when AMD came out with 45nm "Shanghai" which is used for Phenom II and the 45nm Opteron servers which took a slight lead on things like SPECweb and dominated HPC. However, single-socket Nehalem (the i7) was already beating AMD two-socket "Shanghai" for server applications and we all knew that Nehalem-EP was going to take a massive lead.

But by March of 2009, Intel launched Nehalem-EP with QPI and Integrated Memory Controller and the rest is history.


RE: So does this mean...
By Anonymous Freak on 5/4/2009 4:11:07 PM , Rating: 2
"Nehalem" is not yet available in 4P or 8P boxes, but the 2P boxes demolish AMD's 4P boxes, and come extraordinarily close to Intel's own previous-generation 4P boxes in most benchmarks.

This article mentions a potential release timeframe for Intel's 4P+ Nehalems.


RE: So does this mean...
By GeorgeOu on 5/4/2009 7:42:38 PM , Rating: 2
I explained this at the end of the article.

Current Nehalems are "Nehalem-EP" quad-cores designed for 2-socket servers. However, Nehalem-EP systems are already beating AMD's 4-socket solutions and encroaching on Intel's older Core Microarchitecture based 4-socket Dunnington.

By the end of this year, Intel will have a replacement for Dunnington with the "Nehalem-EX" 8-core chip designed for 4 or more sockets. Nehalem's QPI and IMC will allow it to scale nearly perfectly.


crappiest review
By snakeoil on 5/4/2009 6:24:35 PM , Rating: 4
this is the crappiest server review i have seen, seriously. either intel sent you this pamphlet or you are another victim of the swine flu,which is possible because this review is a mess.




RE: crappiest review
By pachitoone on 5/4/2009 7:01:18 PM , Rating: 1
What do you expect from Georgie stupidOU? Not worth the reading. He will destroy the reputation of this site.


RE: crappiest review
By stmok on 5/5/2009 8:39:22 AM , Rating: 1
George Ou?...Wait a minute! Didn't he used to work at ZDNet as a pro-Microsoft and Intel blogger? I remember those anti-opensource and Apple conspiracy articles he did!

How the heck did he end up at DailyTech?

Watch out folks! George is a skilled spinner!


RE: crappiest review
By Sivar on 5/4/2009 7:57:16 PM , Rating: 2
Yeah, I hate it when there are facts, numbers backed up with references, and direct comparison tables.

What crap!

He doesn't even use any incredibly meaningful words like "amazing", "revolutionary", "TCO", "Adreneline", or the phrase "No obligations -- Act now!"

One word which is thrown around a lot is "fanboy", which I'll avoid using here, other than to say that their posts seem to be heavy on attacks and light on content.


RE: crappiest review
By GeorgeOu on 5/5/2009 7:21:34 AM , Rating: 2
LMAO, you must be referring to Neil Nelson who "guarantees" you an honest analysis or "your money back".


???
By audifan on 5/4/2009 12:56:08 PM , Rating: 1
Why is AMD still in business?




RE: ???
By iheartzoloft on 5/4/2009 2:07:53 PM , Rating: 5
Because without them Intel would not even need to raise the bar and we would all still be using pentium pros.

If nothing else respect the fact that AMD exists to keep intel ever so slightly honest. Without AMD intel would still be uping processor speeds 10 megahurtz at a time.


RE: ???
By Motoman on 5/4/2009 4:51:42 PM , Rating: 3
AMD caught Intel with their pants down a few years ago...it was not that long ago at all that you just weren't a "serious gamer" without an AMD chip in your PC.

If you appreciate the CPUs of today, don't thank Intel - thank AMD. Even if you buy an Intel today.


Question on power consumption...
By Amiga500 on 5/4/2009 2:06:41 PM , Rating: 2
At idle:
Xeon L5520 66.3 W
Opteron 2376HE 119 W

That is a difference of 53 Watts - at idle.

Now - considering AMD quote the average thermal draw of the CPU under *representative* load conditions (AMD's definition - not mine) to be 55 Watts - it does beg the question:

What is the difference in system set-ups (hardware) in this test?

Furthermore, consider the delta between idle and full power tests, they are:

Xeon L5520 104 W
Xeon X5570 163 W
Opteron 2376HE 119 W
Opteron 2384 143 W

The idle to load differences in themselves say nothing, but coupled with the dubious idle power draws for the AMD machines...




By Amiga500 on 5/4/2009 2:15:02 PM , Rating: 2
I would also direct readers to:

http://it.anandtech.com/IT/showdoc.aspx?i=3536&p=1...

Performance Per Watt is still massively in favour of Intel... than is not surprising. But the power draw figures do appear to be in favour of AMD.


By zpdixon on 5/5/2009 12:37:40 AM , Rating: 2
You raise a valid point: there are big differences in the hardware configurations for the Xeon L5520 and Opteron 2376HE:

The PSU used by the AMD system is not 80 PLUS certified (check http://80plus.org/manu/psu/psu_join.aspx and also check the label here http://cgi.ebay.com/ColdWatt-CWA2-0650-10-IT01-Pow... ) therefore probably 70-80% efficient. Whereas the Intel system uses a HP PSU that is 80 PLUS certified (85% efficient if comparable to most HP PSUs - check 80plus.org)

Moreover the PSU of the AMD system is a 650W one, compared to only 450W for the Intel system. Because the efficiency of a PSU dramatically decreases with load, if the system draws 150W internally, the Intel PSU would operate at 33% load vs. 23% load for the AMD PSU, thereby enabling a greater efficiency for the Intel system.

The Intel system also has only 8GB of RAM vs. 16GB of RAM in the AMD system. That probably accounts for 4-8W of additional power.

The Intel system has a 2.5" SSD (1-2W) compared to a regular SATA drives in the AMD system (5-10W).

All these differences --but mostly the differences in the PSUs-- could easily explain a difference of 30-45W. But of course the Dailytech propaganda solely places the blame on the CPUs... This is just ridiculous. Very few people in the IT industry actually know anything about power draw in computers. Comparing these 2 SPECpower_ssj2008 test reports actually tells you virtually [b]nothing[/b] about the CPUs alone.

Note to the reader: I own a clamp-meter and have built power efficient systems for years by carefully measuring the power draw of every single of their components and every single rail. I have validated the efficiency of PSUs. I have pushed PSUs to their limit at the exact amperage rating of some of their rails. I have helped friends troubleshoots why their PCs wouldn't boot because they would overtax a 12V rail. Etc.


real power consumption
By duploxxx on 5/5/2009 3:08:50 PM , Rating: 1
Time to get some real powerconsumption in this review in stead of the cut and past from several sites without a decent comparable platform:

so lets start with REALITY! Some official power measurements from HP OEM site, HP has a special testlap where a huge amount of there systems are tested with various load SW and where you can combine each piece of hardware to see the influence. Both systems have the same base server layout and same latest server generation.

HP DL380G6
2 x L5520
4 x 4GB PC3-8500R LP
2x 146GB 15K 2.5 SAS
2 x 460W HE Power supply

idle: 132.29W
40% cpu load: 171.69W
60% cpu load: 191.72W
100% cpu load: 231.51W

HP DL385G5p
2 x 2376HE
4 x 4GB PC6400 LP
2x 146GB 15K 2.5 SAS
2 x 460W HE Power supply

idle: 106.68W
40% cpu load: 147.94W
60% cpu load: 168.7W
100% cpu load: 209.93W

So dear George you cqn already shut up with the bullshit about ACP and the launch of the opteron EE series, because they ARE more power efficient.

All the rest of your data is just a cut/paste from SPEC and virtual banchmarks where HT is a real advantage, Real world applications tend to behave different, if you would test your self you would know.

That will not change the fact off course that nehalem is really a very good cpu architecture, but it is also way more expensive and does NOT consume less!!!!!!




RE: real power consumption
By tshen83 on 5/6/2009 9:25:29 PM , Rating: 2
There are so many MISTAKES in your post, I don't know where to begin.

1. HP DL380G6 L5520 power consumption figures are WRONG. 231W full load power consumption reminds me of the E5520 power consumption. Where is your data source? I am sure you mixed up the L5520 which is 60W TDP vs the E5520 at 80W. That's why your idle is 50W+ higher than the idle power consumption quoted by standard Specpower test. So I recommend you check your sources again. Here is the math:

Dual L5520 = 120W TDP
5520 chipset = 27W TDP
4x4GB DIMMs = 20W TDP
2x 146GB 15K 2.5SAS = 10W TDP
90% efficient power supply you get a total of about no more than 190W at all theoretical max. The extra 40W comes from the 20W differencial between the L and E Xeon.
BTW, nobody should be loading 4 DIMMs of memory on Nehalems, it is freaking triple channel.

2. Nehalem 4GB DDR3 is likely quad ranks and the AMD 4GB is dual rank dimms. And you are only loading 4 ranks total per Socket on the AMD system, causing power consumption to be slightly lower.

3. Let's say that the L5520 system and the AMD 2376HE system both consume 200W. Your argument still fails because L5520 scores about 200 points in SpecInt and 184 SpecBase, while the 2376HE scores only 120 SpecInt and 101 SPECint_rate_base2006. So if both system consumes the same power, you are talking about a 67% Nehalem performance per watt advantage based on SpecInt or a 80% performance per watt advantage based on SPECint_rate_base2006. If you load more ranks onto the AMD system, I have personally seen 100% performance per watt advantage.

Sources:

L5520 system:
http://www.spec.org/cpu2006/results/res2009q2/cpu2...

2376HE system:
http://www.spec.org/cpu2006/results/res2009q1/cpu2...


By leexgx on 5/4/2009 3:23:27 PM , Rating: 2
ACP is crap i agree, lucky for AMD the only good cases on my list are AMD and the 45w EE X2 K8 cpus are back on the list (the phenom X2 are to hot and so are the 5200+ X2 K8)

hopefuly thay are not lieing about the heat out put

Wish thay would put on box Max heatsink temp when both or all 4 cores maxed out


IBM anyone?
By dastruch on 5/4/2009 8:05:49 PM , Rating: 2
Power6 or "Nehalem" Xeon ?




voice of reason
By tshen83 on 5/4/2009 8:40:11 PM , Rating: 1
Hey George nice article.

This is what I have been yelling on anandtech recently. Too many idiots are biased these days that they are willing to ignore hard data against amd.

Regarding the ACP numbers, it has only been getting worse since Barcelona.

Barcelona 75w acp is 95w tdp.
Shanghai 75w acp is now 115w tdp.
How amd managed to sqeeze 2 more core into the same acp for Istanbul is beyond me. But if I were to have a guess, it is either a significant reduction of clock speed or a deeper fudge of acp, say 130w tdp?

Nice article. The latest opteron 3.1ghz isn't a better performer than the e5504 because SE opterons have as high as 137w tdp. Compared to the 80w e5504 it is a bigger joke than even the opteron 2384.




so wait...
By yacoub on 5/4/09, Rating: -1
RE: so wait...
By The Keith on 5/4/09, Rating: 0
RE: so wait...
By oTAL on 5/4/2009 12:11:49 PM , Rating: 3
WARNING: Do NOT feed the troll!


RE: so wait...
By yacoub on 5/4/2009 1:40:11 PM , Rating: 1
Oh it's not a troll and I don't even understand how you could twist it into one. It's an honest question and as i noted, i don't really care about the data so much as whether or not this is an advertisement care-of Intel or an article DailyTech actually wrote.

And given that every other post that questions DailyTech's news quality and article originality gets rated a '5', I was also hoping for a free '5' rating, too. :)


RE: so wait...
By oTAL on 5/5/2009 11:05:21 AM , Rating: 2
They get a 5 when they are right. You are not. This is a decent article (to say the least) and it backs up its claims with data.

Furthermore, you did not provide any justifi.......

...

... .......

damn.....

I'M FEEDING YOU!... i'm a moron....


RE: so wait...
By Calin on 5/4/2009 12:51:30 PM , Rating: 3
Those benchmarks might be hand-picked for the best appearance - one must however note that the new Intel processors beat the hell out of the old Intel server processors too


RE: so wait...
By rcc on 5/4/2009 1:58:37 PM , Rating: 2
One would hope so. Otherwise, what's the point.


RE: so wait...
By Screwballl on 5/4/2009 3:50:15 PM , Rating: 1
The real question is.... what about a full suite of benchmarks that show the whole picture... not just a few that are shown with Intel at the top.

SPEC 2006 uses the Intel optimized compilers so of course as shown before, favors Intel CPUs when third party testing shows a much different picture.
From http://www.spec.org/cpu2006/flags/Intel-ic10-ia32-... it shows:

quote:
SPEC CPU2006 Flag Description for the Intel(R) C++ Compiler 10.1 for IA32 and Intel 64 applications and Intel(R) Fortran Compiler 10.1 for IA32 and Intel 64 applications


Even VMWare has worked with Intel to increase optimization with their processors giving Intel a lead there as well.

http://www.vmware.com/company/news/releases/intel_...

So next time try using other benchmarking programs that are not so heavily invested and produced by, with or inside Intel.


RE: so wait...
By GeorgeOu on 5/4/2009 7:22:34 PM , Rating: 2
Yacoub, I can assure you that this Intel did not have a hand in this article and they knew nothing about it. Intel did not send me "suggestions" for benchmarks to use. I went to the major benchmarking sites and researched for the best scores from AMD and Intel. Some of the benchmarks were suggested by my friend "JumpingJack" (who is popular in the CPU forums) and he has no affiliation with Intel.


RE: so wait...
By yacoub on 5/5/2009 8:29:45 AM , Rating: 2
Thanks, George. You wrote a very good article.


RE: so wait...
By lopri on 5/5/2009 3:50:05 PM , Rating: 2
No worries there, really. George Ou will PAY to write for Intel. Intel need not move a finger.


RE: so wait...
By DallasTexas on 5/6/2009 8:09:43 PM , Rating: 2
Great article George. The numbers don't lie.

What you are encountering here is a mob of angry AMD proponents that don't like the current situation. What do you expect them to say?

The point is that we need more writers like you that stick to the facts and the numbers and not pander to telling the readers what they want to hear. They should go see mommy for that. Great job.


RE: so wait...
By Screwballl on 5/6/2009 11:04:22 PM , Rating: 2
No he is running into people that have done their own testing of hardware and found that this story has a serious Intel leaning bias. I know Intel Core2s are rocking the consumer market but when it comes to the server CPUs, AMD does have a strong hand and there is no way Intel would manhandle them as bad as this biased story leads them to believe.

One fanboi calling others fanbois does not mean this story proves any point.


RE: so wait...
By GeorgeOu on 5/7/2009 12:11:36 AM , Rating: 2
The article doesn't even mention Core 2, but maybe that's your problem.


RE: so wait...
By ditkaamdVSintel on 6/2/2009 6:04:08 PM , Rating: 2
That got Obama elected. Do your research and make fair comparisons that's all we ask.


RE: so wait...
By ditkaamdVSintel on 6/2/2009 6:05:06 PM , Rating: 2
Sorry, It was meant to say "It's articles like these that got Obama elected"


"I mean, if you wanna break down someone's door, why don't you start with AT&T, for God sakes? They make your amazing phone unusable as a phone!" -- Jon Stewart on Apple and the iPhone










botimage
Copyright 2015 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki