Mobile Kepler review

[Lonbjerg]

Except each has drastically different power consumption, a different feature set and very large gaps in performance.

Define "different feature set"?

And ...again...what performance requirements do you have?

 

[Mopetar]

Not sure what your point is....

My point is that regardless of what kind of chip you have, if you don't use a cooling solution that's designed to handle the chip's TDP, it's going to get hot. Kepler may very well run hot by design, but we don't have nearly enough data points to reach that conclusion and it's far more likely that Acer doesn't have an adequate cooling setup for the chip, evidenced by the fact that it's an ultrabook and those things aren't exactly known for having beefy fans or large vents. There's no good reason to assume it's a Kepler problem, especially after Anand's review didn't report nearly as much of an issue with heat.

 

[Meaker10]

Define "different feature set"?

And ...again...what performance requirements do you have?

Well kepler and fermi have different features such as low idle power etc. Do I really need to explain that. Also this is not a purchase decision for me, so what do you mean my requirements? You are not making any sense.

 

[Lonbjerg]

Well kepler and fermi have different features such as low idle power etc. Do I really need to explain that. Also this is not a purchase decision for me, so what do you mean my requirements? You are not making any sense.

So you have nothing?

You wrote:

"Except each has drastically different power consumption, a different feature set and very large gaps in performance."

That means you cannot suddenly merge power into "features".
I am still wondering what this featureset you are talkin about is...because you you think DX11 vs DX11.1 is going to mean ANYTHING in laptops...well...

So what are the different "feature sets"?

 

[Mopetar]

So what are the different "feature sets"?

:hmm:

I'm pretty sure one of them comes with a lovely set of stickers, whereas the other doesn't.

:sneaky:

😛

 

[Arzachel]

That means you cannot suddenly merge power into "features".
I am still wondering what this featureset you are talkin about is...because you you think DX11 vs DX11.1 is going to mean ANYTHING in laptops...well...

So what are the different "feature sets"?

DX11.1 and three monitor support. Physx will most likely also be tweaked for Kepler, if you care about that.

 

[taltamir]

DX11.1 and three monitor support. Physx will most likely also be tweaked for Kepler, if you care about that.

eh, those are fairly insignificant feature differences not drastic differences.
That being said, they do have drastic differences in power consumption and performance so that point is correct.

 

[Meaker10]

So you have nothing?

You wrote:


That means you cannot suddenly merge power into "features".
I am still wondering what this featureset you are talkin about is...because you you think DX11 vs DX11.1 is going to mean ANYTHING in laptops...well...

So what are the different "feature sets"?

28nm vs 40nm, if I have to explain myself further then there is no point.

Seriously I worry about this forum, first someone who can't see the advantage of a quiet and cool running heatsink, next someone who can't see the advantage of a smaller process chip in a notebook.....

 

[Magic Carpet]

I'm glad most people are impressed with the performance... but if i read the above statement... I'm more scared than impressed..
(also for less ventilated or lower max temps the performance will be different...(lower))

Temps in F?

 

[taltamir]

28nm vs 40nm, if I have to explain myself further then there is no point.

what? that isn't a feature. That is the improvement that resulted in power efficiency improvements.

you said

"Except each has drastically different power consumption, a different feature set and very large gaps in performance."

You seperated power consumption and performance from features. (since each is an extremely important feature that is understandable... the most important features are power consumption, cost, and performance).
But such a separation means that when asked what the "features" that aren't power consumption are, you can't repeat power consumption.

As for

Seriously I worry about this forum, first someone who can't see the advantage of a quiet and cool running heatsink, next someone who can't see the advantage of a smaller process chip in a notebook.....

Unnecessary personal attack. Nobody here is disputing how important power consumption is. You are just being called out on your claim that there are major differences BEYOND power consumption; and not because there are no differences at all, but because you have been unable to name a single one of them and all you can name is power consumption when asked what those features are. Granted, there ARE differences which other users went and filled in for you (DX11.1 vs DX11 & 3DV improvements) but you were unable to name them. Nor are they really relevant (unlike cost and power consumption and performance which are hugely relevant).

 

[Meaker10]

You will notice I said drastic differences in power consumption, then said differences in features and performance.

Yes the features were mentioned, so I was right in saying that.

Performance, well look at the benchmarks. There are going to be large breaks between random numbers:

630 -> 635
635 -> 640
660 -> 670
675 -> 680

Don't you think not having any clear break even between the generations is at all confusing to people?

So let's summarise:

I said there was a major difference in power consumption
I said there were differences (but did not say major differences) in features and discrepancies in the performance gaps.

Power consumption may as well be God when it comes to notebooks, it affects weight, price, size and battery life. To have two different process technologies scattered amongst the different chips is absurd.

 

[taltamir]

You continue to ignore what others say to you to go on ranting.

I explicitly said I agree with all those points; in fact I am pretty sure I raised some of them before you did.

But they are utterly irrelevant to the question of "did you or did you not try to pass off power consumption as a feature after you explicitly placed it in its own separate category". And note that saying they are irrelevant to that question doesn't mean they are irrelevant to all. They are very relevant to the discussion of mobile keplar.

While you addressed the "significant features" (that was my mistake in reading your post, you said just "features"; I apologize and concur that it is a problem, even if the features are minor you should have clear separation). You have tried to claim "28nm" and "lower power consumption" as a feature.

 

[Meaker10]

You could have two chips with the same power consumption, but because one has a power saving feature it can last longer when you are on your desktop, so Kepler has an improved optimus where it idles better.

You are being staggeringly pedantic here, but if you want to get into that level then sure.

Your apology is accepted though, all I am trying to highlight here is Nvidia's very confusing naming structure and am slightly astonished that people don't seem to agree with this.

Going from a 635M to a 640M is going to improve your idle battery, then going from a 660M to 670M it's going to dramatically decrease again, then going from the 675M to 680M its suddenly going to increase again.

 

[SirPauly]

I think your points are balanced and fair!

 

[busydude]

Your apology is accepted though, all I am trying to highlight here is Nvidia's very confusing naming structure and am slightly astonished that people don't seem to agree with this.

Sadly.. that is the norm rather than the exception. Look at Intel's horrid naming scheme.. and AMD is doing the same thing. Every one here expects 6990M to be two 6970Ms bolted together. It is very very hard to keep up with all the numbers and names in the laptop space.

 

[Meaker10]

But within the mobile naming scheme it's the same tech and is not advertised as crossfire (and can bolt two together).

To get this screwed up requires effort.

 

[taltamir]

Your apology is accepted though, all I am trying to highlight here is Nvidia's very confusing naming structure and am slightly astonished that people don't seem to agree with this.

Not a single person disagreed with you on this principle, you keep on trying to "win" with this appeal to emotion.

It might shock you but people can agree with you about how distasteful this behavior of nvidia is, and yet those same people can still argue with you about your inaccuracies. Just because we aren't happy with nvidia's nonsense doesn't justify you being inaccurate about parts of it. Nor does it give you the right to accuse people of "taking nvidia's side, how insane! don't you realize that <insert fact we agree with like power consumption>" when they call you out on those inaccuracies.

 

[Meaker10]

Yet one of those things that I was "inaccurate" about was something YOU misread and I was very careful in the way I wrote it, to not say it.

The other point I have just shown to be false.

By all means correct a mistake on my part, but it helps if I have made a factual error.

I have been upgrading notebook graphics cards since it's been pretty much possible to do so, before that I ran a list of mobile graphics cards. While I do make mistakes there are very few people who have the same amount of knowledge on these mobile chips as myself.

So please, I am waiting for you to point out where I was wrong in what I said.

Let's go back to my original statement, I will clarify it if that would help.

The new range of Nvidia cards, while ordered correctly in terms of performance, is going to have differences between chips that matter much more in the mobile space than their desktop counter parts would. There will be large jumps in TDP at certain points (up and down) along with vastly different idle power consumptions. Beyond that there are feature differences between fermi and kepler (which we will learn more about at launch) and while correctly ordered in performance the jumps are very uneven at certain points.

 

[Riek]

Temps in F?

Ofcourse they are.

but 40-50 degree is not something you want on your lap or where you want to lay your hands on for gaming.

 

[taltamir]

Yet one of those things that I was "inaccurate" about was something YOU misread and I was very careful in the way I wrote it, to not say it.

For which I apologized.
Yet the other 3 were not.

The other point I have just shown to be false.

What? No you didn't.
You said it differs in performance, power consumption, and features.
When asked what those features were you said:
1. Lower power consumption
2. Improvement to optimus power consumption.
3. 28nm vs 40nm.

Those 3 things all fall into power consumption, hence the inaccuracy.

The actual feature difference is DX11.1 and 3D vision improvements, neither of which you mentioned.

 

[railven]

...

Did I just really read through that? Someone else mentioned it, he didn't, it was put out there, he still isn't wrong in what he said. He just isn't listing the features as you want him to.

All he has to do is ammend "feature set with" with "DX11.1/blah" and his point still stands. It doesn't detract from what he is saying.

Which sucks for consumers since now they'll really get screwed.

 

[pelov]

Could this be another case of Ivy? Great perf-per-watt and low power consumption but due to decrease in die size the heat dissipation is hampered? I don't think it's going to be a big deal on the desktop but heat constraints on a laptop are an entirely separate matter.

 

[Meaker10]

The reason I am not mentioning exactly, is we don't know exactly, yes there is DX11.1, but there could be others we don't know about yet.

I pointed out how features can differentiate in power chips that at full load use the same amount of power.

I am no longer going to respond to your posts unless you actually post something that makes sense.

 

[f1sherman]

I understood you well the first time - mixing Kepler with Fermi SKUs, and model numbers not being indicative of what arch. they are built upon.

And ofc. you can't answer exactly what Kepler features are, because we don't know that yet.

 

[taltamir]

All he has to do is ammend "feature set with" with "DX11.1/blah" and his point still stands. It doesn't detract from what he is saying.

Yes, it is all he has to do. Yet he adamantly refuses to.
As I said before I agree with his view about nvidia, my issue is with his inaccuracy in presenting the points which support the argument both he and I agree on.

Yet he insists on making it into me "taking nvidia's side" and acts outraged that anyone would.

Yes, ONE of those inaccuracies was me misreading, I apologized. The rest weren't due to any misreading though.