i7 2720QM or i7 2820QM

James D.

Sorry not true, look at the below clocking schedule of a 2720 quad against the 3 top dual's. The quad's are clearly equal or higher than the 5xx series. On the 2620 only single core is higher by 1 bin, equal on 2 cores. The base clocks the quad is several bins lower, but that is with all 4 cores running.

CPU Base 1C /2C/ 4c
2820QM 2.3 GHz 3.4/3.3/3.1 GHz
2720QM 2.2 GHz 3.3/3.2/3.0 GHz
2620M 2.7 GHz 3.4/3.2 GHz
2540M 2.6 GHz 3.3/3.1 GHz
2520M 2.5 GHz 3.2/3.0 GHz

 

True. This part makes me wrong, I've should read spec-sheets. I will edit post above underlining my mistake.
However on another part I believe I was right. If other settings of dual core vs quad stayed the same (dual have less TDP etc) 2720 will never use less power no matter how it will undervolt other 2 cores (as said above) against 2620. On idle dual core always will be (OK, I am sure on 99%) the winner in battery life. Use battery bar to check discharge speed. If you will load processors I think dual core still will use slightly less battery power no matter that "tdp level may be not reached". However I did not check that on SB processors.

 

it actually depends, read something about power gating technologies, and the pentium v pro chips that were embedded in that generation for more compatible instructions

 

Dual cores will always consume less power than a quad core made from the same manufacturing process. Kinda like a 4-cylinder vs. V-8 engine. At idle it may be marginal, but the fact that the cores exist will consume SOME energy.

I don't think there's any benefit between the 2720QM or 2820QM though in any games or applications. Difference is 100MHz lower and top end clock speeds and 2MB more cache.

 

QFT
10 char

 

I agree, I saw a chart recently (Oh I wish I could remember which site) that showed idle power of a Dual and a Quad 2nd Gen, and they were virtually the same.

 

notebookcheck also shows virtually similar benches between the two. the 2630 scores lower than both though.
although im a video editing enthusiast, i dont even see how the realworld marginal gains of the 2720/2820 (over the 2630) would matter. only the 2920 offers realworld gains, even at stock, and if this kind of cpu performance mattered to your profession, then price would be a secondary concern.

 

To clarify, TDP is a maximum value of heat dissipation, not power draw. If a processor has a TDP rated at 45 watts, this does not means that it is constantly eating up 45 watts, it means that the cooling system will need to be able to dissipate up to 45 watts of heat.

Even at full load, some processors won't hit their TDP because they physically can't, the ceiling is artificially too high for them to reach.

Imagine this hypothetical scenario:

You have two runners, A and B, who are exactly the same. Same height, same stride length, same genetics, etc. Except, runner A can run at 10 MPH maximum while runner B can run at 9 MPH maximum.

When they both start a race (Note: the length of the race is unlimited for this hypothetical situation, they will continue running forever), runner A will be always ahead of runner B because he simply can run faster. We will relate this to CPU frequency.

Now, because they have the same genetic makeup, except for maximum speed, you would expect runner A to dehydrate faster than runner B because he is stressing himself harder by running faster. Dehydration will represent heat output and stress represents power draw.

At some point, runner A will need to replenish his body with water so he can continue running at his maximum speed or else he will need to slow down or completely stop. Runner B will need to do this less often, because he is stressing his body less than runner A. Water represents cooling.

As you can see, because runner A is faster, he would complete a race faster than runner B, but require more water as a direct result of stressing himself more. Unless runner A or B overexerts himself and runs at 15 MPH, where he is dehydrating faster than water can rehydrate himself, he can continue running forever. NOTICE: 15 MPH is a give-or-take threshold; it is not absolute. Both can safety run at 12 MPH or even 14 MPH forever because water can keep them hydrated at those speeds and heat outputs (This is easier to explain using math, but to keep it simple this is all you need to know). This is why 'Extreme' processors that allow for overclocking are given a higher TDP so they have more headroom for running at higher frequencies; the cooling system will need to be able to handle the extra heat.

Because CPU frequencies and voltages are dynamic, once they finish a task (running a race), they downclock when not needed to do anything (walking when not racing to stay cool). But, when stressed, a marginally faster clocked processor will draw marginally more power and output marginally more heat -- the cooling system will be sufficient enough either way because Intel overestimates the TDP as a safeguard. But, in terms of battery life, unless you will be consistently stressing both the processors at 100%, the slower clocked one will have marginally more battery life.

It seems that every new generation of CPUs the same question is asked. Because we can't defy the laws of physics, this applies to any CPU of the same class/architecture. The 2630QM will MORE than suffice for daily tasks and gaming. My P8400 still gets the job done.

Wikipedia is extremely helpful for these kinds of questions as lots of people jump to incredibly misleading conclusions.

TLDR;

 

The low end parts is completely inadequate.

The i3 from arrandales provided lesser battery life than the i5, all things the same, i.e. same notebook.

They dont take into account that all the dual cores are made using the same process, and that by the quality of the chip, i.e. how many functions that the said chip contain, that the division is done. The same is valid for quad cores. This is only valid for intel.

 

Physics instantly disproves your claims, I don't understand what you are trying to say.

Here is what notebookcheck had to say (Last picture labeled "Power Consumption"). The i3 beats out both the i5 and i7 in power consumption.

Battery life is inversely related to power consumption; i.e.: The higher the power consumption, the lower the battery, all things being equal.

So what I don't understand is how a CPU which draws less power can have a lower battery life, which is what you are trying to argue.

Could you elaborate on this or reword it? It's very hard for me to make out what you are trying to say.

 

The i3 is a fixed clock CPU. The i5 and i7 have variable clocks. So in idle instances, while the i3 is churning away at 2.4GHz for example, the i5 or i7 could be running at only 2.0 GHz.

 

The i3's may lack turbo boost, but I'm sure Intel implemented some kind of EIST technology into those chips for them to be able to idle ~20 watts. Check out the link above from notebookcheck.

 

I beg to differ, the i3's along with the i5's and i7's all downclock when idle, I had a Toshiba A665-S5170 with an i3 380M and it idled at 900Mhz, my Sager idles at 1.2MHz. so that does not explain the difference that notebookcheck is showing. It could be the idle clocks are different on each series or the unreliable nature of some of the things that notebookcheck does.

 

Agreed on the i3's downclocking. Stepspeed/EIST have been around for a long, long time, it would surprise me for them not to include such a critical feature in a notebook processor. The i3's lack turbo boost.

Notebookcheck can definitely have some sketchy results, but the load power usage seems correct, but as you said, maybe i5's can clock lower making them more efficient? Interesting.

 

What makes a part higher binned?

How are turbo bins distributed?

What makes turbo boost work?

the 2nd and the 3rd, can make use of what I asked you to search, the pentium v pro chip.

 

Mr MM, your answers are very cryptic and inadequately explained, you ignore various questions and problems other have with your ideas, and you tell others to search for proof of your claims.

 

Look at that graph a little more carefully. Mr. MM is not explaining it, but he is not wrong. The power consumption of a chip is given roughly by:

P = C * f * V^2

where C is a constant parameter dependent on the architecture and the silicon being used, f is the frequency the chip is running at and V is the voltage. The chips are binned based on the silicon: Intel has certain power requirements and it tries to push up the frequency as high as it can without exceeding them. Thus, the higher binned chips tend to be those with lower C.

Now, when an i5 is running at full power, this only helps a little -- it is nowhere near enough to match the linear increase due to frequency (even if the voltages are the same) relative to an i3. However, when the chips are idle, they downclock to the same frequencies and voltages and the only difference is the constant. Thus, in the chart you linked, the i5-540M consumes less power at idle (Min/Max: 17.6/20.5) than the i3-330M (18.1/20.7).

Of course, with the iX processors, this is a rather small effect and it is quite likely swamped by statistical fluctuations (particularly since a lot of Intel's binning is artificial), but with sufficiently distinct silicon and excellent power-gating, it is not impossible for a quad-core to consume less power at idle than a dual-core. I don't think this is true of any current processors of the same architecture, but it's possible.

 

Makes sense. But, it still doesn't explain how a 2820QM will have better battery life than a 2720QM, unless the computer is idling more than it is being used. At a higher frequency, we can agree that a chip will consume more power than a chip with a lower frequency, as long as they architecturally similar. Unless you are comparing idle times of battery life, then according to what you said, the higher frequency chip will consume marginally less power, but under load the lower frequency chip will consume significantly less power than a chip running at a higher frequency.

So, my question still remains, how can the 2820QM achieve a higher battery life, even though it runs faster?

 

He made several statements, most of whch after investigating I believe are wrong, including saying that frequency has no effect on power consumption and that TDP is power consumption.

Do you have any source suggesting that there is any measureable variance in capacitance of a CMOS device of fixed design and manufacturing method? I have been looking and I cannot find anything suggesting that. I just don't understand how one processor could have any measurably different capacitance than another built with the same die on the same manufacturing process. It seems more like you and Mr MM are trying to brainstorm for reasons why a "higher binned" processor might have better battery life, without even knowing that this premise is even true.

 

Battery life is almost always determined by idle power consumption. Very, very few people run tasks that push their CPUs to the limit while on battery. The typical things people do on the go (email, web browsing, etc.) put almost no load on a modern CPU and even watching a movie doesn't push Sandy Bridge much.

This is an experimental fact well known to any overclocker -- some chips can run at a given frequency with a given voltage and a given power consumption while others cannot, even though they are nominally the same chip. In fact, there is considerable variance even within chips of the same model. I don't know of a source that spells it out explicitly, but here is an article that discusses a "bad" chip (in fact, this one actually needs a higher voltage at idle so it's not just the C that is screwing it):

 

i'd go with the 1st one i cant remember its name but its the proscessor i have

 

i7 2630QM 2.0Ghz is the 1 u want if u care about battery life. still a very good proscessor

 

You are drawing the conclusion that the capacitance of processors must be different if the clock rate and voltage abilities of two processors are different, but you can't. All it takes is one transistor to malfunction out of hundreds of millions, and the processor won't clock as high or require more voltage at a given clockspeed. Do I think that it may be possible for similar processors to have a slightly different capacitance? Maybe, but I don't think so, and if it is the case, it is not really a measureable difference. But something tells me that the capacitance of similar processors made on the same assembly line would have to be within way less than 1% of each other.

But you're also just making a guess based on another guess.