Why are the Core i3/i5 considered low-power?

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
wendell
Posts: 26
Joined: Sun Jan 16, 2011 12:30 pm
Location: USA

Why are the Core i3/i5 considered low-power?

Post by wendell » Sun Jan 30, 2011 10:47 am

I frequently see people recommending an i3 or i5 for a low-power/heat system. But a Core i3-540 is 73W TDP. That sounds pretty toasty to me.

Also, as far as I can tell, none of the CULV processors are available retail.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Sun Jan 30, 2011 11:14 am

wendell wrote:I frequently see people recommending an i3 or i5 for a low-power/heat system. But a Core i3-540 is 73W TDP. That sounds pretty toasty to me.
73W TDP is a theoretical maximum running a burn. In real life the sip electricity compared to other CPUs.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

fumino
Posts: 298
Joined: Thu Aug 05, 2010 4:38 pm
Location: ontario

Re: Why are the Core i3/i5 considered low-power?

Post by fumino » Sun Jan 30, 2011 11:18 am

that tdp includes the on package gpu as well.
my foamed mini p180 | corsair cx-400 w/ fan swap | scythe mugen-2 | thermalright hr-05 | unlocked sempron 140 | asrock 880g | 8gb mushkin |vertex 30gb boot | 7TBs storage |

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why are the Core i3/i5 considered low-power?

Post by HFat » Sun Jan 30, 2011 11:56 am

If you look at a table, it's obvious the TDP values you're talking about are wrong. The rated TDP is probably the TDP of the fastest dual-core, not of the one you're buying.
That said, these CPUs are *not* low-power. People who say so lack perspective. They're not P4s alright and Clarkdales have low idle power consumption but that's not the same thing. Unless you can underclock the CPU in the BIOS or something, it will require a serious power supply and cooling system.

You can get the regular mobile CPUs at retail. They're not worth the price IMO. I think there are some fairly low-power Xeons available at retail as well. But you should really be buying boards which come with a CPU if you want low-power.

victorhortalives
*Lifetime Patron*
Posts: 207
Joined: Wed Aug 08, 2007 1:50 am
Location: Xhystos

Re: Why are the Core i3/i5 considered low-power?

Post by victorhortalives » Sun Jan 30, 2011 1:03 pm

Don't entirely agree.

If the objective is to build a very quiet system, then a very worthwhile goal is to be able to run the system with only one slow running fan.
That means being able to use a passive PSU (probably a Pico) as well as ideally (but not mandatory) an embedded CPU.
In the absence of an embedded CPU, then a single fan that cools the CPU and does case ventilation is OK.

So a "low-enough" power CPU is OK, such as an i3 with a 120mm fan.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Sun Jan 30, 2011 1:31 pm

victorhortalives wrote:So a "low-enough" power CPU is OK, such as an i3 with a 120mm fan.
What about a
65W TDP i5-2400S $204.99 at newegg vs.
73W TDP i3-560 Clarkdale $149.99 newegg
73W TDP i3-540 Clarkdale $119.99 newegg

Then keep in mind that the 95W i5-2400 and i5-2300 (184.99) pretty much matches the 65W TDP i5-2400S in actual energy usage until you push its performance beyond what the i5-2400 performance envelope.

see:
http://www.xbitlabs.com/articles/cpu/di ... html#sect0

Does that not tip your decision over to an i5-2400 and i5-2300?
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why are the Core i3/i5 considered low-power?

Post by HFat » Sun Jan 30, 2011 1:40 pm

@victorhortalives:
With the right case and the right heatsink you can cool pretty much any CPU with a 120mm fan. Not that it makes much sense not to use more if it that allows you to spin them slower (and not to have too much overheating and downtime if your single fan craps out).
You're free to use your own definitions of course but if a Clarksdale is low-power then so are the dual-cores branded as i5 and pretty much every modern dual-core for that matter. So you'd need a new word for chips that burn less power than mainstream CPUs.

@ces:
The Sandy Bridge quad-cores are expensive. Intel is apparently planning a 35W $70 dual-core Sandy Bridge. That's perhaps more of a Clarksdale replacement, and arguably somewhat low-power to boot as it's a special model with lower computing power and power consumption than a regular Sandy Bridge.
Last edited by HFat on Sun Jan 30, 2011 1:45 pm, edited 1 time in total.

victorhortalives
*Lifetime Patron*
Posts: 207
Joined: Wed Aug 08, 2007 1:50 am
Location: Xhystos

Re: Why are the Core i3/i5 considered low-power?

Post by victorhortalives » Sun Jan 30, 2011 1:43 pm

Well, since I already have an i3, I'm not making a new decision.

However, I do have another PC with a Zotac Atom/ION that I AM looking to make a new decision about replacing, but not yet.
If I do, then I'll be looking at an AMD solution or waiting for a new i3 variant that allows me to reuse a 60W pico PSU.


" So you'd need a new word for chips that burn less power than mainstream CPUs " - Low Enough Power ? LEP CPUs ?

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Sun Jan 30, 2011 2:11 pm

victorhortalives wrote:a new i3 variant
When the 35W $70 dual-core i3 Sandy Bridge CPUs that hfat refers to arrive, I can't for the life of me imagine why anyone would use an atom... unless they already own it.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why are the Core i3/i5 considered low-power?

Post by HFat » Sun Jan 30, 2011 2:31 pm

The $70 SB won't be an i3. Intel's branding is confusing and irrelevant.

Cheap dual-core Atoms are 13W. That's a good bit lower than 35W. Prices are quite different as well. Atoms and similar patforms never competed with mainstream gear and mainstream gear still doesn't compete on price or power consumption. They have different uses. Atom's main competitors are AMD's low-power gear as well as VIA and ARM chips.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Sun Jan 30, 2011 2:54 pm

HFat wrote:The $70 SB won't be an i3.
I was referring to the 35 watt i3 2100t... which is basically the Sandy Bridge version of the dual core Clarkdale that virtualizes an additional 2 cores.

I have an atom netbook. It can handle windows... barely. A 35 watt i3 2100t will never have to make apologies to anyone about handling windows or anything else. It will always have a crisp response to it limited only by other components. That and an SSD will handle anything anything short of heavy duty gaming.

For a few watts and a few bucks.. why put up with an atom?
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why are the Core i3/i5 considered low-power?

Post by HFat » Sun Jan 30, 2011 3:39 pm

You've answered your own question.
ces wrote:I have an atom netbook.
Chances are it's a single-core with a TDP under 6W. It's got the name "Atom" but all Atom are not equal.
Dual-core Atoms *with enough RAM* "handle Windows" just fine. It's not a particularly difficult job.
But enough about Atoms.

The 2100T price is like $145-$150 here. Perhaps $125 for you? There will be cheaper Sandy Bridge CPUs.

wendell
Posts: 26
Joined: Sun Jan 16, 2011 12:30 pm
Location: USA

Re: Why are the Core i3/i5 considered low-power?

Post by wendell » Sun Jan 30, 2011 4:10 pm

ces wrote:73W TDP is a theoretical maximum running a burn. In real life the sip electricity compared to other CPUs.
Where do I find numbers for energy use in realistic tasks? How do I compare, for example, the Core i3-540 and Celeron SU2300?

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why are the Core i3/i5 considered low-power?

Post by HFat » Sun Jan 30, 2011 4:25 pm

These CPUs don't compare. One is low-power and the other isn't.

Google and you shall find:
http://www.anandtech.com/show/3869/zota ... dias-ion/3
It's not the best comparison because they don't tell you the boards they're using. And the inclusion of an old Atom is silly. But it should give you a fairly good idea.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Sun Jan 30, 2011 5:52 pm

HFat wrote:These CPUs don't compare. One is low-power and the other isn't.

Google and you shall find:
http://www.anandtech.com/show/3869/zota ... dias-ion/3
It's not the best comparison because they don't tell you the boards they're using. And the inclusion of an old Atom is silly. But it should give you a fairly good idea.
This isn't the best comparison either, but it is enough in my opinion to question why anyone would any longer be interested (in most applications) in either an atom or a Clarkdale:
http://www.xbitlabs.com/articles/cpu/di ... html#sect0

Now keep in mind these are 4 core Sandy Bridge's being compared with a 2 core Clarkdale. How do you think a two core / 4 virtual core Sandy Bridge will perform?????

I think at the end of the day, there is no comparison. For a few extra bucks (not that many) you get a real computer that can really run windows with no excuses and no delay... and it will live within any reasonable thermal requirements... even little ITX boxes.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Sun Jan 30, 2011 6:05 pm

wendell wrote:
ces wrote:73W TDP is a theoretical maximum running a burn. In real life the sip electricity compared to other CPUs.
Where do I find numbers for energy use in realistic tasks? How do I compare, for example, the Core i3-540 and Celeron SU2300?
I wish I had a better link at my finger tips, but this should provide a lot of help. Hfat's link is a more direct answer to your question, this one I think is a more direct answer to your problem.... I think (read it and you tell me) :
http://www.xbitlabs.com/articles/cpu/di ... html#sect0
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

wendell
Posts: 26
Joined: Sun Jan 16, 2011 12:30 pm
Location: USA

Re: Why are the Core i3/i5 considered low-power?

Post by wendell » Sun Jan 30, 2011 11:25 pm

ces wrote:Hfat's link is a more direct answer to your question, this one I think is a more direct answer to your problem.... I think (read it and you tell me) :
http://www.xbitlabs.com/articles/cpu/di ... html#sect0
Since this is SPCR, when we talk about energy usage, we are concerned with the cooling required. So whether a difference in wattage is significant depends on whether it pushes cooling to unacceptable noise levels. I don't have any feeling for what level of heat would be too much in a Mini ITX box...30W?..70W?..100W? I'm sure there is no answer for everyone.

It seems that idle power would be important, and your example illustrates that some of the Core i3/i5 compete with the Atom and Celeron in that category.

But if low heat is a significant concern, I would think that a Core series might be a problem, since it could get quite warm under heavy load. Whereas, a low-power processor would take longer for a task but not get so hot.

Celeron SU2300 vs. Atom 330 vs. AMD E-350
Celeron SU2300 vs. Atom 330 vs. Core i3-540

Aren't the CULV i3/i5/i7 processors designed to avoid that very problem? I see retail prices listed for some of them, but I haven't found where you can buy them.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Why are the Core i3/i5 considered low-power?

Post by quest_for_silence » Mon Jan 31, 2011 4:52 am

wendell wrote:a low-power processor would take longer for a task but not get so hot.

AFAIK either we are talking from a silence perspective, or from a green-computing one, the wasted energy (heat) is a function of time.

With this respect, I think you have to take into account the computing efficiency, and giving a certain specific task an Atom well may be regarded as a "less" low-power CPU than an high-end Sandy Bridge.
On the contrary, I mean that if we take into account the total energy over a certain period (including idle times), more probably that not the things would go far better for the Atoms and comparatively way worse for the SBs.
So, the certain general conditions are the ones to determine (mostly) such an outcome.

That saids, as published efficiency benchmarks usually (if not always) are made with reference to very different tasks either for Atoms-like CPU and for general purpose ones, I guess that all our guessworks cannot be backed by any meaningful data, whichever CPU we may support for.
Regards,
Luca

Support SPCR, use these links when you buy: NCIX, Amazon and Newegg

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Mon Jan 31, 2011 5:43 am

wendell wrote:Aren't the CULV i3/i5/i7 processors designed to avoid that very problem? I see retail prices listed for some of them, but I haven't found where you can buy them.
You can buy CULV cpus from newegg... the problem is finding a motherboard with which to use them.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Mon Jan 31, 2011 6:24 am

wendell wrote:But if low heat is a significant concern, I would think that a Core series might be a problem, since it could get quite warm under heavy load. Whereas, a low-power processor would take longer for a task but not get so hot.
Theoretically that is so. But reality is a bit different.

What constitutes a "heavy load" for you? Take a look at this test of a top of the line Core i5-2500K
http://www.silentpcreview.com/article1159-page5.html

At idle, running on an intel DP67DG motherboard, overclocked to 4Ghz the CPU and motherboard together are using 28 watts. Running an H.264 playback they use only 5 more watts. No one, not even gamers use the power that a Prime 95 burn in uses. This has always been true, but it is even more so with these new chips. They just have a wider dynamic range than prior generations. They will use up only so much power as is required by the work they do. And even when heavily loaded, they run fairly cool (that is they are not wasting a lot of the energy they are using)

Here
http://www.xbitlabs.com/articles/cpu/di ... html#sect0
is data on an i5-2400 cpu coupled with a Gigabyte GA-H67MA-UD2H motherboard. At idle the i5-2400 CPU is using under 4 watts, and the motherboard another 14 watts. That is 16 watts total compared to 28 watts total for a dual core clarkdale.

I would expect that a load that maxes out an atom will barely budge the needle of a a four core i5-2400.

The authors say "Moreover, the LGA1155 system is overall surprisingly energy-efficient. It is not without reason that Intel claims Sandy Bridge to be a great choice for HTPC." and they are comparing it with the already super efficient Clarkdale.

Prime 95 was always an artificial test. With these new chips that is even more so. For them, running a H.264 playback is basically the same as running at idle. Under any reasonable load, and most unreasonable loads, they still run cool.

==================

These new CPUs represent game changing technology in which everyone must adjust their old intuitive models of power consumption, heat generation.... and the relationship of these two factors to noise. The CPU is no longer the most important factor.... it has become the motherboard. If you want to obsess over power consumption and heat generation, you need to move your focus from the CPU to the motherboard. That is where the greatest differences will be found.

I don't know this for a fact, but I just don't think there is that much difference in efficiency between an atom motherboard and an equivalently sized 1155 motherboard. They have to do basically the same work.

Maybe someone else can comment on that. Someone here must know the energy usage of a robust Zotac Atom ITX motherboard and the energy usage of the Zotac 1156 ITX motherboard.... and how much they vary with the load. I bet between equivalent boards they are not more than 5 or 6 watts different at idle... and full load doesn't add more than 10 watts to either.

Does anyone know?
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Mon Jan 31, 2011 6:36 am

oops
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

wendell
Posts: 26
Joined: Sun Jan 16, 2011 12:30 pm
Location: USA

Re: Why are the Core i3/i5 considered low-power?

Post by wendell » Mon Jan 31, 2011 11:38 am

ces wrote:You can buy CULV cpus from newegg...
Ah ha, I think you are thinking of the mobile version, not the CULV. For example, the i5-560M (2.66 GHz, 35W) is easy to find, but not the i5-560UM (1.33 GHz, 18W).

I've looked for these CULV processors given with retail prices here, but couldn't find any sellers.

Celeron SU2300 $134
Pentium SU4100 $289
Core 2 Duo SU9600 $289
Celeron U3400 $134
Celeron U3600 $134
Core i5-560UM $250
Core i7-680UM $317

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Mon Jan 31, 2011 12:04 pm

wendell wrote:
ces wrote:You can buy CULV cpus from newegg...
Ah ha, I think you are thinking of the mobile version, not the CULV. For example, the i5-560M (2.66 GHz, 35W) is easy to find, but not the i5-560UM (1.33 GHz, 18W).

I've looked for these CULV processors given with retail prices here, but couldn't find any sellers.

Celeron SU2300 $134
Pentium SU4100 $289
Core 2 Duo SU9600 $289
Celeron U3400 $134
Celeron U3600 $134
Core i5-560UM $250
Core i7-680UM $317
Celeron SU2300
$179.99 at newegg as part of a motherboard/CPU bundle
http://www.newegg.com/Product/Product.a ... 6813500059

Core i7 680UM
newegg installed in a laptop ($849.99 for the laptop with the chip)
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why are the Core i3/i5 considered low-power?

Post by HFat » Mon Jan 31, 2011 12:17 pm

The OP is apparently looking for CPUs without boards.
The current ULV CPUs are listed in some shops in Europe but nobody seems to be stocking them... and why should they? Compared to Atoms and suchlike, their prices are crazy! Even the boards are very expensive at retail. And the performance isn't great. OEMs don't pay nearly as much for the stuff.
Better underclock (and undervolt) mainstream chips. Or gut an old laptop to make a ghetto desktop. Or find a used mobile C2D board in desktop form factor. Or something... anything but buying retail.
wendell wrote:I don't have any feeling for what level of heat would be too much in a Mini ITX box...30W?..70W?..100W? I'm sure there is no answer for everyone.

It seems that idle power would be important, and your example illustrates that some of the Core i3/i5 compete with the Atom and Celeron in that category.

But if low heat is a significant concern, I would think that a Core series might be a problem, since it could get quite warm under heavy load. Whereas, a low-power processor would take longer for a task but not get so hot.
These comparisons with the old desktop Atoms are silly. You can beat the measurements you linked to with the low-end of the low-end now. You can tell the clueless and the disingenuous by how they talk about "atom this" and "atom that". In fact power consumption varies a lot depending on the models, just like it does with more powerful gear. Apple got their C2D SFF PC to idle very low by using mobile parts for instance.There are lots of boards which can beat Clarkdales (and therefore all mainstream desktop CPUs) at idle power consumption with the right PSU.
We don't have good measurements for the E-350 yet but it looks like it could do as well as some of the more efficient Atoms for less money. Plus it's got better features and better single-thread performance.
But none of this matters much unless you're on solar power or something. A handful of watts at idle will not make a real difference in your electricity consumption and you've got to take power consumption at load into account for cooling. As you say, how much power consumption is too much depends on your goals.
If you want absolutely no noise, you're essentially looking at Bobcat or Atom unless you want to spend a lot of money for mediocre performance.
If very little noise is OK (and it should be unless you're a perfectionist), you should be able to cool fairly beefy chips in a mini-ITX box. But mini-ITX cases come in different shapes and sizes. If you wanted to run a regular CPU you'd need a larger case and you'd need to spend a good bit more money, to put in a lot more work and so on.


ces wrote:These new CPUs represent game changing technology in which everyone must adjust their old intuitive models of power consumption, heat generation.... and the relationship of these two factors to noise. The CPU is no longer the most important factor.... it has become the motherboard.
You've reached a new level in hyperbole. That's not new. How has the game changed?
ces wrote:I don't know this for a fact, but I just don't think there is that much difference in efficiency between an atom motherboard and an equivalently sized 1155 motherboard. They have to do basically the same work.
No.
All motherboards do not run at the same speed. That's most obvious by looking at the RAM they support but some boards underclock dynamically which is less obvious. That's less of an issue now that more stuff is being integrated in the CPU die but it's still there.
Motherboards also need to power the CPUs with low-voltage current. It's less of an issue now that the trend is moving away from very power-hungry CPUs but boards still need to do that job. Motherboards which need to support less power-hungry CPU tend to be more efficient. But at the same time cheaper boards tend to be less efficient. So if you're looking at the low-end, you'll see some fairly inefficient low-power boards. But mainstream gear can't touch the efficiency of the more sophisticated low-power boards.
There's no such thing as "an atom motherboard". Different models have very different efficiencies. The first desktop Atoms boards were very inefficient... shockingly so actually. At roughtly the same time, Intel used Atom to release most efficient modern x86 systems on the market by a long shot. Prices were of course different.

With regards to powerful Sandy Bridge CPUs, you need to understand a couple of things:
-relying on low CPU utilization to keep your system cool means you're one bug or greedy piece of software away from a cooked system... a simple webpage could cook your system
-buying expensive CPU when you're never going to use their power is foolish... wait until less powerful Sandy Bridge chips are relased if you love the platform so much (it looks like the boards are buggy so you're in for a wait anyway)

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Why are the Core i3/i5 considered low-power?

Post by ces » Mon Jan 31, 2011 1:23 pm

HFat wrote:No.
You've reached a new level in hyperbole...... (emphasis added)
All motherboards do not run at the same speed. That's most obvious by looking at the RAM they support but some boards underclock dynamically which is less obvious. That's less of an issue now that more stuff is being integrated in the CPU die but it's still there.
Motherboards also need to power the CPUs with low-voltage current. It's less of an issue now that the trend is moving away from very power-hungry CPUs but boards still need to do that job. Motherboards which need to support less power-hungry CPU tend to be more efficient. But at the same time cheaper boards tend to be less efficient. So if you're looking at the low-end, you'll see some fairly inefficient low-power boards. But mainstream gear can't touch the efficiency of the more sophisticated low-power boards.
There's no such thing as "an atom motherboard". Different models have very different efficiencies. The first desktop Atoms boards were very inefficient... shockingly so actually. At roughtly the same time, Intel used Atom to release most efficient modern x86 systems on the market by a long shot. Prices were of course different.
OK Hfat, let's keep the "hyperbole" out of this discussion.

One assumption I am making is that people are going to use properly licensed Microsoft software and that the system is likely to be used as more than a server. It will be used as an extra general purpose computer. And when they do use it, they don't want the irritation of waiting waiting waiting. A few hundred dollars extra on a system that I am going to load $700-900 (a retail full license for Office and Windows and some other misc licenses) of software isn't a big deal. After paying for and loading in this software... waiting around, tapping my fingers, while an atom flails around is a big deal... at least to me.

It's not unlikely that some of us may take the same facts and form alternate opinions. People have different values and priorities. Crisp user interface response is pretty high up on the list for me. For others it may be squeezing that law bit of wattage out of a system. And for others it will be all about going all passive without burning up any components. Some people cheat on the licenses and are looking at different economics.

I suspect my speculation on the numbers is fairly on the mark. My proposition is that comparing a reasonable high end atom ITX motherboard with a reasonably high end Sandy Bridge (or even Clarkdale) ITX motherboard isn't going to show a lot of difference in energy usage (I even proposed some wattage figures I suspected the differences might be) under many circumstances... and for some user profiles, under all circumstances. A Zotac to Zotac ITX comparison between boards that are similarly provisioned would be strong evidence one way or the other.

But I made clear I didn't know what that difference was for a fact. You seem very confident you know. And you certainly seem conversant on the subject.

And what constitutes a large difference or a small difference will be different for different people. But the numbers won't be different. Everyone is entitled to their own opinion, but not their own facts.

Do you have any facts you can share with us? Names of specific boards, wattage numbers ect... supporting what you say? Without that it's hard to discern whether something is "hyperbole" or not. I have a tendency to disagree with many of the judgments and opinions on a good many of your postings... but I will not disagree with your facts. I truly don't know how the numbers compare. Without those numbers I don't know how you, I or anyone else can form informed opinions.

If you have them, or know where to find them, would you be kind enough to post internet verifiable numbers supporting the opinions and judgement you have expressed and share them with us. Just if it isn't too much effort. It will help others of us to form informed opinions.

Thanks
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
"Aristotle calls man the rational animal. All my life I have been seeking evidence to confirm this" Bertrand Russell
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former" Albert Einstein

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Why are the Core i3/i5 considered low-power?

Post by quest_for_silence » Mon Jan 31, 2011 1:51 pm

ces wrote:Thanks

ces, now (you can) sit down and rest. :wink:

Am I the only one who see this entire conversation as becoming a bit pointless? Moreover that you are - more probably that not - talking about - maybe very - different things? :?:
Regards,
Luca

Support SPCR, use these links when you buy: NCIX, Amazon and Newegg

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why are the Core i3/i5 considered low-power?

Post by HFat » Mon Jan 31, 2011 2:53 pm

It may or may not be pointless. If I believed it was pointless to talk with ces I wouldn't answer her more than others. I won't bother to respond on the licences business however because I think that's pointless.

Now, as to something which might be useful:
I didn't point out specifics because I'm not talking about obscure gear for the most part.
There's gear that consumes less at load than the best Clarkdale setup at idle so there's no way it'd end up consuming as much when you integrate power consumption (see lower for some examples).
There are few manufacturers which try to get a lower than usual idle power consumption for desktops. I doesn't make a lot of economic sense for most people. But this is SPCR.
So far as I know, Zotac does not make particularly efficent boards, Atom or otherwise. Zotac's market is apparently more people who want features at low cost. But Intel or Jetway for instance do make efficient Atom boards. The most powerful 2010 models carry the N550 CPU. But let's not be so focused on Atoms which are by no means the only game in town!
There's a lot of talk right now about Zacate boards right now. You've read Anandtech's latest article about them, yes? Surely you have noticed the alledged surprisingly low idle power consumption (the PSU used was not a good match and possibly cherry-picked so we won't know for sure the until a more serious outfit like SPCR gets their hands on one of these boards). And Zacate isn't even the low-power line. That's Ontario at about half the TDP. They're the only modern x86 chips which consume as little as efficient Atoms at load.
But let's look at gear which is actually in the hands of end-users. While more powerful gear can not compete with puny CPUs at load consumption, being competitive at idle consumption is not out of the question. Apple for isntance has a well-known SFF PC line which consumes a lot less power at idle than the competition. Back when Atom was introduced they did this by using regular mobile C2Ds (so far as I know they didn't use the low-power versions) and the Atom chipsets with the lowest power consumption handily beat them. But they have since then started to use ULV C2Ds the OP is lusting after and have almost caught up as far as idle power consumption is concerned. Obviously we're not talking about cheap gear here. Virtually all of the manufacturers who used the Atom chipsets I'm talking about put them in laptops because that's where they make the most sense of course. But there's a small company which sells ultra-SFF desktops based on them called fit-PC2.
I'm sure you can Google the exact power consumption figures as well as me with the help of the pointers above.

edit: perhaps you can also Google figures for the existing AMD mobile gear. It tends to be more affordable than Intel's. I don't know the figures but I was surprised at how low a UPS measured a M880G board with a Neo CPU when I looked at that. It might well be a bogus reading and I don't have a reliable meter to confirm.

Post Reply