Linus Torvalds (torvalds@linux-foundation.org) on 10/3/10 wrote:
---------------------------
>? (0xe2.0x9a.0x9b@gmail.com) on 10/3/10 wrote:
>>
>>1. [Marketing]: A big CPU/GPU company will make sure (you can bet on it) to implant
>>the consumers with the idea of "you need an even more powerful chip". There exist
>>numerous techniques of how to achieve such an effect.
>
>Did your mother drop you on your head repeatedly, or were
>you just born that way?

You are insane. (The fact that I told you to start with an insult does *not* mean that you had to. You had a choice not to start like this.)

>And yeah, I actually do kind of disagree. "Kind of",
>because I don't think you're necessarily wrong, but I
>think that marketing doesn't really care what they
>sell, and performance isn't realy the only thing to go for.

Imagine that you have a "dying big company". The sells are getting lower, nobody wants its products. What do you think the marketing minds in the company are going to do about it? I think they are going to look at what the company produces or might be producing (a GPU company produces GPUs or something GPU related), and sell it. Other companies in the industry will try to do the same, otherwise they are dead. The purpose of marketing in a dying industry is to create a new kind of product and a new kind of usage pattern that will appeal to people, so that those people think they are buying something they need and when using the product they should feel like the time spent using the product was actually worth it.

Maybe the previous paragraph sounds negatively. That is *not* what I am saying. What I am saying is that marketing is creative. It "creates stuff". (You can find the same statement in many books about marketing.)

The bottom line: What I am trying to say is that consumers should have *opinions* about products. This is perfectly logical, since many products are arbitrary creations. A consumer should always ask at least the questions "Is this what I really need?" and "Isn't there something hidden in this product that I do not want or do not approve of?". Once the consumer stops asking these questions, he or she is reduced into a mindless puppet in the hands of the market.

>Look at cars. Yes, performance is still one facet, but
>it's not a very big one any more. Milage, but even more
>so just quality is a bigger deal. And things like
>a silent cabin an a smooth ride (in the US that obviously
>means "mushy") is a much bigger selling point.

Pff ... you are comparing totally different things. I would say that using this analogy (GPU = car engine) kind of proves that you actually are mad (you know, mad people see analogy everywhere and in everything). Whoops ... that was an insult, but here comes the argument:

You *cannot* compare GPU and car engines "just like that". First and foremost, GPU don't need to follow roads. That may not seem important, but it is important. On the *same* road, there are *various* cars. Big, small, fast, slow, old, new. If you had a super-fast car, capable of 300 km/h, you wouldn't be able to use it on any road. Even if all cars on the roads were the same, roads aren't built for 300 km/h, especially in cities. In GPUs, there is no such thing as a road.

There is no major bottleneck which would prevent an individual from using a 10x faster GPU than others. This is why some people do have a 10x faster GPU at home than some other people. With car engines, this would be completely impossible. This is one of the major reasons why car marketing *cannot* concentrate itself on increasing engine horse power. If car marketing could make use of increasing the engine horse power, they would be already taking advantage of it, but they cannot, since it does not make much sense.

Considering this difference and other differences between GPUs and car engines, I don't think it is a good idea to make the analogy. Of course, I am not against making "GPU = car engine" analogies, I am just saying that if you do such an analogy, you should check that the analogy is sane.

Bottom line: Unless somebody presents a good reason that would *prevent* marketing from utilizing raw GPU performance as a selling point, I am going to maintain that marketing will try to take advantage of it.

Or do you think a GPU company is going to advertise like this: "If you optimize your software, you probably won't need any GPU at all!"

>And marketing people had absolutely no problem with it.
>The whole muscle car thing was replaced by reliability,
>safety, price etc with hardly a blip. The car manufacturers
>that still sell on "performance" are selling to a pretty
>small audience these days.

Have you seen a "totally inefficient car"? One that consumes 100x more gas than other ones? I don't think I have. But I've seen totally inefficient software.

So, "car performance" and "software performance" are somewhat different concepts, don't you think?

>And the same thing is happening in computers. In general,
>performance has become almost secondary to form factor.
>Even in the gaming world, a lot of it is done in very
>portable environments, with things like the iPhone and the
>3DS changing the environment.
>
>>2. [Programmer laziness]:
>
>Sure. It's always easier to just brute-force things
>and use big hardware over trying to be really clever.
>
>But the thing is, power use is a real issue.

Wouldn't a faster algorithm automatically consume less power?

>That's true even on the desktop. Have you played any of
>the first-person horror games ("survival horror").

Why should I involve myself in controlling a character in situations that have a predetermined outcome, are an artificial set-up, even with time limits?

(In my opinion, one of the best games is Sokoban for DOS. You don't need a GPU to play that one.)

>There's
>a lot of quiet sneaking around. To get the full experience,
>you basically want to do it alone at night - and without
>that leafblower running in your computer.

If I may ask, why do you think the time you spent playing this particular game was worth living? Isn't is because you thought you were doing something meaningful? The question is: where did this meaningfulness come from?

I do not mean this in a negative way. I just want to really know what are the factors and people that influenced you and ultimately "caused you to play the game".

>Yeah, you can use headphones, and people often do. But it's
>a negative.

Why don't you wait 5 years or something, and play the game on a computer that is much more energy efficient than the one you have right now? If you don't want to listen to the noise of e.g. Fermi, simply wait 5 years.

The question I am asking is: Why do you think you have to play the game *now*?

>>Oh my, I almost forgot: Don't you forget to start your response with the obligatory
>>"You are on drugs" phrase! If not that, please make sure the first sentence of your
>>response is an insult. It will be an insult to me if you don't start with an insult. I am dead serious!
>
>So I wonder if I should have insulted you more by
>not insulting you. It's a conundrum.
>
>Linus

Yes, it is a conundrum. But if you read Godel,Escher,Bach you would know that we have the ability to escape from such contradictory situations. In other words: there is always a 3rd option.