Sign in with
Sign up | Sign in
Your question

Dual Core Processors For Low-Power, High-Performance Desktops

Last response: in Memory
Share
April 24, 2006 10:47:32 AM

Intel's Core Duo is worth serious consideration if you care about performance, low noise and high efficiency in a PC system. We compared it to efficiency specialists Pentium M and Turion 64, but found the Athlon 64 X2 to be a potential show-stopper.
April 24, 2006 12:33:31 PM

It's sad to see that despite the fact that the Core Duo stomped all over the AMD lineup in the benches, the Core Duo still earns that conclusion Core Duo Loses The Platform Game.
April 24, 2006 1:41:32 PM

Quote:
Clearly, the Athlon 64 X2 offers superior performance and efficiency, but we are looking for high efficiency solutions, which draws our attention to Intel's latest mobile dual core processor: the Core Duo.


What exactly do you mean here? I'm not trying to troll or anything, I know you intended to say something else here. I genuinely want to know what was meant to be said or what the statement intended to say.
Related resources
April 24, 2006 2:16:17 PM

Quote:
Clearly, the Athlon 64 X2 offers superior performance and efficiency, but we are looking for high efficiency solutions, which draws our attention to Intel's latest mobile dual core processor: the Core Duo.


What exactly do you mean here? I'm not trying to troll or anything, I know you intended to say something else here. I genuinely want to know what was meant to be said or what the statement intended to say.

Actually, the T2600 goes to-to-toe with the X2 in terms of performance, and whomps it in efficiency, at less of a cost. I'm not sure where THG is getting thier info nowadays, but I doubt they're paying much for it.

http://techreport.com/reviews/2006q2/core-duo/index.x?p...
April 24, 2006 3:07:16 PM

Quote:
It's sad to see that despite the fact that the Core Duo stomped all over the AMD lineup in the benches, the Core Duo still earns that conclusion Core Duo Loses The Platform Game.


Whizzard9992,
The statement Core Duo Loses The Platform Game is accurate as far as meaning that there are not motherboards currenly available to take full advantage of the powersaving aspects of the Core Duo design. I do not take the statement to mean that the Core Duo loses in the benchmarks shown in this article. It might have been better put to say that the Core Duo is so new that the motherboard manufacturers have not yet produced products for the CPU.

Looking at the benchmarks the Core Duo is at the top of many and close to the top in the rest of the performance benchmarks. In the Power consumption tests the Core Duo uses from 70W to 93W. The Pentium M 780 is the only one of the 4 CPUs tested with a descrete graphics card that uses less wattage (59W to 83W). The Athlon X2 3800+ uses 74W to 133W; and the Turion uses 79W to 101W.

Assuming that the motherboards will appear in the future which take advantage of the design features of the Core Duo (and appropriate BIOS), the performance of this CPU should improve.

By the way, why is this thread under Memory rather than CPU?
April 24, 2006 5:22:50 PM

Anyone notice the screenshots of CPU-Z on pages 15 and 17? Athlon X2 4600+? RAM at 240?
April 24, 2006 7:18:55 PM

That doesn't necessarily mean that the RAM is set at 240MHz base clocks. That just means that HT is set at 240 with a multiplyer of 10x that sets the CPU to 2.4GHZ. The Memory is on it's own bus. You can set that independent of HT. That one on page 17 does show that though.... hmm.
April 24, 2006 7:34:06 PM

Quote:

Bt the way, why is this thread under Memory rather than CPU?


:wink: to hide the thread from the likes of MMM and conroe, and all the other flamers and fanboys in the CPU forum :roll:
April 24, 2006 9:25:26 PM

I didn't see a single benchmark where the 2.16GHz Duo lost to the 2.26GHz Centrino.

Is THG aware that a great many folk jump straight to the conclusion of these?
What a poor choice for of words for the conclusion title?

Look at the benchmarks in which the Duo beat out the Centrino:
Open GL ... Duo won
Quake ... Duo won
3DMark2005 ... Duo won
Pinnacle Studio Plus ... Duo won
Divx transcoding ... Duo won
Xvid transcoding ... Duo won
Main Concept MPEG encoder ... Duo won
Ogg Vorbis ... Duo
WinRAR ... Duo
3D Studio Max .. Duo wtfpwn'd em all
SiSoft 2005 Synthetic ... Duo
SiSoft 2005 ... Duo
PCMark05 ... Duo by much
PCMark05 CPU ... Duo by much
PCMark05 Memory ... Duo

Now lets look at the benchmarks in which the Centrino beat out the Duo:

NONE!

They tied in the Lame MP3 encoder bench for which THG used a beta version of to test (this software is just in its Dual Core optimization infancy).

Furthermore, the energy consumption charts are bogus ... on page 1 of the review:
"The Core Duo uses the same processor socket as the Pentium M, but requires some electrical modifications, which means that you cannot use existing Socket 479 motherboards. Suitable products are not yet available, but several motherboard makers are working on them. We had a look at one of the first solutions back in March, AOpen's i975Xa-YDG 975X chipset is not known to be energy efficient."

So they grabbed some AOpen POS thats KNOWN to be an energy hawg to show meaningful results? Ummm no ...
April 24, 2006 9:27:42 PM

Quote:
As mentioned in the article, GPUs are becoming also problematic, both noisy and major power consumers, would it be possible as well to have an efficiency test for GC as well?


I believe I've read elsewhere that the 7900GT (from what I've heard, comparable to the 7800GT) produces much less in the way of excess heat. So maybe nVidia is taking a step in the right direction (the shrink down to 65 (?) nm helps, of course)
April 24, 2006 9:50:05 PM

Quote:
Intel's Core Duo is worth serious consideration if you care about performance, low noise and high efficiency in a PC system. We compared it to efficiency specialists Pentium M and Turion 64, but found the Athlon 64 X2 to be a potential show-stopper.

Absolutely right conclusion - all three mobile variants are great CPUs, we just don't have a price-viable platform for any just yet... at the moment low-power, yet reasonably priced computing seems rooted in socket 754 with the pre-Turion CPUs. Not an ideal solution by any means, if you're looking for performance.
April 24, 2006 10:46:37 PM

Quote:

Whizzard9992,
The statement Core Duo Loses The Platform Game is accurate as far as meaning that there are not motherboards currenly available to take full advantage of the powersaving aspects of the Core Duo design. I do not take the statement to mean that the Core Duo loses in the benchmarks shown in this article. It might have been better put to say that the Core Duo is so new that the motherboard manufacturers have not yet produced products for the CPU.


Don't get me wrong: I understand and agree. Not a criticism of the article, just raising a point. It's a shame such a good processor has to be shunned by lack of platform support. I wonder if the i945 overstock has motherboard makers wary, especially in light of Conroe's pending release.

I don't agree that the X2 clearly crushes the Core Duo, though :)  It does outperform it in some areas, but the Duo takes the crown in others. It gives me hope that the conroe will be a true contender, and not just an over-hyped marketing scheme.

This is, of course, based on limited information. It would be nice if THG did thier own Core Duo versus X2 platform duel ;) ;);) My money is on the X2, but I still think it would be an interesting read.
April 25, 2006 1:19:16 AM

Mr Schmid and Mr. Ross,

Sorry, if it is already discussed. This is not a flame thing, but just an observation on the way THG conclude thing.

I do have problem with the statement "Core Duo Loses The Platform Game" as the conclusion. As I count thru the tests, (for comparison purpose, I ignore the onboard graphics components), here what I found:

Given a tie if less than 2% difference in result:

Core Duo: 9 wins of which 3 can be considered tie. (also win in 4 Power consumption test)

X2 3800 : 4 wins

Turion: 1 win (consider tie)

PM: 6 wins

If we exclude the PM, then Core Duo's record is 13 wins.

The statement should have said: "Core Duo wins big in benching but having no solid platform" or taken a page from THG's own style "Core Duo assaults hard .... but having no solid platform". Simply put "Core Duo Lose the Platform Game" is too simplified to the point of misleading for noobies like me, who just may jump to the conclusion and pay no attention to benching. THG may inadvertenly appear bias ... though I don't really think so.

Respectfully,

Chime
April 25, 2006 3:15:25 AM

Quote:
Our conclusion for the Core Duo processor is particularly interesting, because in theory it is capable of enabling the assembly of a dual core desktop system that requires only 45 W. This, however, requires a motherboard that uses the 945GM chipset, which is not yet available.


Yes it is, in the Mac Mini. Since you wanted a system under 45W, there is no need for those PCI slots. You can buy the Core Solo version and swap in a high-end Core Duo, and boot Windows already.
April 25, 2006 3:41:11 AM

Would like to point out a couple of things I find strange in the test since I'm running a similiar setup with a K8NGM-V and a 25W Mobile Sempron.

Quote:
We suppose that this is caused by the integrated graphics unit remaining active after plugging in the PCI Express card.


- There IS a bios option to turn off the integrated graphics on the K8NGM-V.

- Turion MT's runs on 0.8Vcore-1.2Vcore stock while the K8NGM-V only supports 1V-1.3V so it's running on a lil more juice than it should. I think readers should know that.

- Since this is a low power test, why is the test done with 2 HDD's? And a Raptor at that. It sorta skews the results.
April 25, 2006 6:54:35 AM

Quote:
- Since this is a low power test, why is the test done with 2 HDD's? And a Raptor at that. It sorta skews the results.


Wow. That's a really good point. I didn't notice that.

Idle power consumption for the raptor is 8 watts. The samsung spinpont would be the appropriate drive here: I doubt anyone is going to want to RAID a low-power computer. Do the idle stats include the drives in an ACPI "Shutdown" state?

The Mac Mini is also a good example of a 45W Core Duo platform. The Mac Mini's DO run Windows legitimately. This would have been a good addition to the article... They have DVI and modest gfx capabillities.
April 25, 2006 10:44:28 PM

I am just impressed the Core Duo performance is anywhere near the Desktop X2.
We might see some "Mac Mini" type PCs which are notebooks without the monitor and keyboard.
However, I understand the article is about PLATFORMs, not just the CPUs.

Even after the chipsets get updated, I will still not buy a Core Duo until there is full 64-bit support. I'll be happy with a company Core Duo until then 8)
April 26, 2006 3:40:45 AM

I'd say performance is very comparable. The Core Duo is almost 10% faster than the X2 (2.16 GHz vs. 2.00 GHz) and has twice the cache. I bet if you put my 2.2GHz X2 4200+ against the T2600 instead of the 3800+, it would likely tie or edge it out in just about everything. The P6 and K8 architectures are very, very close in performance per clock- the IMC of the Athlon and the larger cache of the Core Duo are about the only real differences, and they only show up in a few tests.
April 26, 2006 3:45:15 AM

Well yes, the Core Duo would lose as it has a maximum clock of 2.16GHz and the fastest X2 runs at 2.4 GHz (2.6 if you count the FX-60 as an X2- I do.) The sheer clock speed would stomp the Core Duo into the ground.

But what this DOES show is that the Conroe chips, based heavily on the Core Duo, have good promise and will fight the K8s well.
April 26, 2006 5:34:28 AM

lol MU_Engineer. I hope you aren't serious....

Core Duo + 64-bit = Conroe

P6 = Pentium Pro, not Core Duo. Saying Conroe is based on the PIII or P6 is like saying the Mercedez CL 55 is based on the Ford Model-T.

Conroe has 4 instructions per clock, K8 has 3 (though not a determinite for performance, it was noteworthy per your clock-for-clock comment).

Lastly, as shown in this article, even the 4800+ X2 takes second chair to the Core Duo, or only loses by a minimal margin, so I doubt your 4200+ X2 would "stomp the Core Duo into the ground."

The Core Duo is not meant to compete with a Desktop Processor such as the X2, but it does so very well nonetheless.

And to be honest, unless I'm missing something, I fail to see how THG comes to the conclusion the X2 to be a potential show-stopper.... the X2 gets taken to the cleaners in EVERY bench test THG performed. How is the X2 a clear winner over the Core Duo in platforms, if that is indeed the point of the article? As I see it, the idle Pentium M with the 800XL takes only 5 more watts than the X2-onboard, and takes 50 fewer watts under heavy load, and beats the X2-onboard by more than 500% in performance. This article about low-power concludes that the largest power consumer is a potential show-stopper? wtf?

If we're talking low-power platforms, the pentium-m is the clear winner. If we're talking a low-power internet machine (I don't know many entusiasts looking to save $20/year on thier energy bill for 10% less performance), the clear winner is the Core Duo. If you want to save money on your energy bill, turn your computer off when you're done with it.

THis article needs a disclaimer, "For entertainment purposes only." The conclusions are shady at best.
April 26, 2006 5:53:48 AM

EXAMPLE:

Your cost of electricity is $0.10/kWH
You use your computer 10 hours a day, 7 days a week

You will save $3.65 per year for every 10W you save on your computer usage...

How is a person with 2 WD raptors going to be concerned with saving $4 per year?

Moreover, if you REALLY want low-power, why not just get a laptop and plug a mouse into it?

*edit*
April 26, 2006 12:05:26 PM

Okay, a few things:

1. The AMD chip tested was an X2 3800+ not a 4800+. And yes, the X2 was very competitive with the T2600, beating it in several tests and when it did get beaten, it was by small margins (a few FPS or seconds.) I attribute this to the 160 MHz that the Core Duo had over the 2GHz X2 3800+.

2. Okay, technically the Pentium M architecture is "P6+," not just P6. But the microarchitecture is pretty similar- ask anybody around here and they would agree. The Core Duo is almost exactly the same as the Pentium M, but on 65nm and with a second core and a higher FSB. Conroe makes some changes to the formula, but it's not akin to the major changes that differentiated the Pentium 4 from the PIII or the Pentium Pro you mentioned from the i586 Pentium. Pentium M/Core -> Conroe is supposed to be similar to the Pentium II -> Pentium 3 or K7 -> K8 in terms of difference in architecture. (But we don't have a Conroe yet to do real benchmarks on, so just chill.)

3. Most chips run at an average of roughly 2 IPC in real-world usage from what I have heard. A 4 IPC chip will benchmark very well in synthetic benchmarks but be not as fast as the benches indicate in real-world usage. Here's something to chew on: the Intel Itanium 1 had 6 IPC and it's not several times faster than, say, a PIII or K8 at the Itanic's same speeds. If I remember correctly, the Itanic 2 issues 11 IPC and still gets its butt handed to it in by Opterons and Xeons in all but some very specific applications. IPC capability is good- but if you can only use 2 IPC, it does not matter whether you can issue 2 or 350 IPC.

4. I never said a 4200+ would "stomp the Core Duo into the ground." I think you are getting my posts switched up- I said a 4800+ or FX-60 would as their clock speeds are 240 and 440 MHz faster than the T2600. I said that the Core Duo performing well with the K8s means that Conroe has promise as it is based on a very similar microarchitecture. I also said "I bet if you put my 4200+ against the T2600, it would likely tie or edge it out in just about everything."

So you should double-check the posts that you reply to and make sure that you're really reading what is written instead of pulling things from various posts about different things.
April 26, 2006 7:15:17 PM

Quote:
Damn - this is my dream ...
I would like to thank Tom's Hardware who kindly listened to its readers and finnaly had this test.
Yet while reading this artice a few questions hit me:

1- As mentioned in the article, GPUs are becoming also problematic, both noisy and major power consumers, would it be possible as well to have an efficiency test for GC as well?
2- what are the technical problems if a mobile graphic card is used instead of a regular one, and are they more efficient also under STRESS?

and I had a wild thought:
would it be COST efficient (just as it is energy and noise efficient) to base a desktop (or HTPC) comuter on a mobile MOBO ? or in other words, just buy the parts for a notebook and building it in a "regular" case instead of a notebook shell ...! (I think this is possible!)

thank you


Yes I 100% agree with the author. Although I would like to permamently see a heat index and power consumption ratings for all hardware, especially graphic cards and recently HD's which have also started to become problematic.

I'm wondering why a graphics card sitting at idle with desktop applications should be too hot for the touch. I don't understand the engineering with that.

I've been seriously looking at Shuttle systems lately, just because I don't like the trend of noisy and hot. I'm testing whole board imerssion with mineral oil cooling and an automobile transmission cooler. But I can't help wondering if this is a permanent PC trend or just a temporary phase we are going through.
April 26, 2006 8:43:08 PM

Quote:
EXAMPLE:

Your cost of electricity is $0.10/kWH
You use your computer 10 hours a day, 7 days a week

You will save $3.65 per year for every 10W you save on your computer usage...

How is a person with 2 WD raptors going to be concerned with saving $4 per year?

Moreover, if you REALLY want low-power, why not just get a laptop and plug a mouse into it?

*edit*


As Whizzard9992 noted, the savings from the power difference are very minimal. However, the difference in cost will be quite large in comarison. Even if saving 100W of power, quoting Whizzard9992 figures, that would save $36.5 per year. That is, you'd have to run the Core Duo system for some 10 years to make the cost difference in the processor alone.

The Athlon64 X2 3800+ costs under $300, the Pentium D 805 sells for around $125 online, the Pentium D 920 for around $250, and all provide a full range of motherboards to suit every need and taste and are full fledged dual core processors. However one would have to pay some $650 for the T2600 processor alone. Thats over twice the price of the X2 3800 and Pentium D 920, and over 5 times the price of the 805, and dont even get me started on the price difference between the motherboard prices for all those processors in comparison to the board featured in the review.

At the end of the day, the only factor that would make any real difference for the end user should be how compact he/she wants their system to be. Building an SFF with a Pentium M based processor is a lot easier than building a system with the same dimentions with an Athlon64 X2 or Pentium D processor. Also, the P-M based system would run much quieter than an X2 or Pentium-D based one due to the smaller heatsinks and fans required to keep the system cool.
April 26, 2006 9:22:11 PM

There are a lot of ancillary costs to hot running equipment. If you live in any place where air conditioning is in use, you know that air conditioning is a very expensive utility. Running even two 60 watt light bubs in a 10x12 room ads a considerable amount of heat build up and causes the air conditioner to cycle on and off observably more often. In fact, you can feel the heat building in a room with this modern CPU and video equipment. It’s that bad!

Additionally, there are other costs which ultimately will be realized in premature equipment failure. Electronics run like this is more prone to failure, especially hard drives. You can test this out for yourself. Ttry to run a disk for three days without a fan, no side rail mounting and just see what happens.

So cooling and efficiency should most definitely not be dismissed. Furthermore, a lot of the electricity being produced to power this equipment in our homes is the product of fuels that are clearly 100% having a detrimental effect on our living environment.

So I guess I would have to strongly argue that it DOES matter a great deal and not just because of the dimensions for the equipment that you buying.

I think it would take quite a bit of math to really determine the true costs of this increasingly hot (wasted energy) equipment; but for me it is irrelevant. I just simply do not want a spot heater in the middle of my house. And when I do, I will go get one.

I just want a video card that runs as cool as the onces I used to own five years ago. Plus 3d with lots of pipes ;-)
April 27, 2006 4:29:10 AM

There appears to be a couple inaccuracies in this article:

Energy consumption increases exponentially with the clock speed.

As I recall, energy consumption increases linearly with clock speed, and exponentially with voltage.

This, however, requires a motherboard that uses the 945GM chipset, which is not yet available.

I just checked newegg.com, and saw that the Asus N4L-VM DH board (which supports Core Duo) is listed - it uses the 945 GM chipset and has onboard video.

When you take this last fact into account, it seems the conclusions drawn by this article about "core duo losing the platform game" seems premature.[/i]
May 2, 2006 4:02:44 AM

woot woot they used my mobo XD
May 2, 2006 6:24:51 AM

Quote:
Additionally, there are other costs which ultimately will be realized in premature equipment failure. Electronics run like this is more prone to failure, especially hard drives. You can test this out for yourself. Ttry to run a disk for three days without a fan, no side rail mounting and just see what happens.


Just to be a thorn in your side, if you live in a cool area, you could actually SAVE money on heating by purchasing a computer with a greater heat dissipation ;) 

The math isn't hard. Factoring in cooling:

BTU per Hour = Watts * 3.4129

For every 100 watts of power your computer consumes, you add a strain of 341 BTU/hour on your A/C unit.

If your air condition has a SEER rating of 10, then you will be using an additional 0.034 kw/H per hour.

Assuming my previously stated usage of 10 hours per day, 7 days a week, this adds an additional cost of $1.24 per year for every 10W of power your computer comsumes (12.41 for every 100 W).

So, in short, even with A/C cooling factored in, you'll be saving $4.69 per year for every 10W you save on your system.

So in conclusion, while it's nice to know that our processors are going Greener, its not worth the price difference to invest in a low-power PC. If you want low-power, get a laptop. At least you can carry it around, and I can guarantee that it will consume the least amount of power and produce the least heat (That is, after all, what they are designed for).
May 2, 2006 6:33:42 AM

I agree with StrangeStranger (who somehow changed his name to IraqiGeek in his post this thread...).

If you want performance, get an X2 or wait for Conroe. If you want a low-power PC, get a laptop or wait for the next-gen chips.
May 16, 2006 2:34:08 AM

45 pages and no print version?!?
Make some noise here, help request a single page print version. NOW!
May 17, 2006 3:31:40 AM

Quote:
45 pages and no print version?!?
Make some noise here, help request a single page print version. NOW!


...

no
May 19, 2006 6:11:01 AM

Quote:
lol MU_Engineer. I hope you aren't serious....

Core Duo + 64-bit = Conroe



are u serious?
May 19, 2006 2:47:06 PM

Quote:
lol MU_Engineer. I hope you aren't serious....

Core Duo + 64-bit = Conroe



are u serious?

Yeah that was a mistake... I meant Merom :oops: 
May 24, 2006 2:42:32 PM

The power consumption is one of the main factors for datacenters.
If you have to fill-up 3000 square meters of datacenter, you consider about the powerconsumption, heat displacement. For every 1 Watt heat, you need 1 Watt for cooling. How lower, how cheaper.
May 25, 2006 4:23:33 PM

Quote:
If you want performance, get an X2 or wait for Conroe. If you want a low-power PC, get a laptop or wait for the next-gen chips.

Well, i consider the Core Duo to be a laptop-CPU, and likewise the motherboards and chipsets suitable for the Core Duo to be closely related with their laptop equivalents...

I must admit that i'm an AMD fan, but i'm impressed by the performance the Core Duo delivers. IMHO it has two 'faults': the pricetag and that it lacks 64-bit (i care because i mainly use linux, where most apps are 64-bit native).

I'm thinking of getting a Core Duo T2300 (1.66 GHz) and an Asus N4L-VM DH i945GM for a HTPC build. I would dearly like to see a performance comparison between the T2600 and T2300, does the difference equal the pricedifference (it usually doesn't)?

It would also be nice to see an atempt to overclock these things (par example o.c'n the T2300 to T2500 (2.0 GHz), how much effort is needed?).

Quote:
woot woot they used my mobo XD

Does the above mentioned motherboard support changing the FSB speed?

/andrnils
!