Support NeoGAF

[JCBossman]

I know this is going to sound stupid, but why not make just the "Best" CPU, same thing with graphics card manufacturers, why make all the different grades when the cost to make a chip is about the same no matter if the final product sells for $60 or $600, sure the $600 card has a better fan/heatsink and some extras here and there but thats pretty cheap stuff.I know not every chip is going to turn out perfect so you put all those as the "budget chips" or lower cheaper grades. Wouldn't be better for computing if intel made NOTHING but Extreme Quads, the good ones sell for a nice premium and the somewhat messed up ones as "Mainstream" parts.
I can understand for example why a car company doesn't make anything but "Luxury" cars, because things like Leather and fancy subsystems cost big bucks, but a chip is a chip.

 

[Rentahamster]

...what? I don't want to pay $1000 for a CPU that'll be obsolete in 6 months.

 

[The Interrobanger]

:lol

EDIT: Serious answer. Why do some drugs cost more than others? Why does an 8GB MP3 player cost more than a 2GB one? It's called business, my friend.

 

[numble]

The manufacturing cost difference between high end basketball sneakers and low end ones is pretty negligible. Why does it cost more for the ones with Jordan/Lebron/Kobe's name on it?

 

[Crushed]

Yeah a chip is a chip. There's absolutely no difference in manufacturing. That extra "technology" is just a bunch of whizzer-whazzers slapped on.

 

[speculawyer]

Wouldn't be better for computing if intel made NOTHING but Extreme Quads, the good ones sell for a nice premium and the somewhat messed up ones as "Mainstream" parts.

Well, the 'lower quality' ones are often exactly the same as the high-rated ones . . . hence people overclocking their systems.

 

[Steriletom]

The manufacturing cost difference between high end basketball sneakers and low end ones is pretty negligible. Why does it cost more for the ones with Jordan/Lebron/Kobe's name on it?

Marketing costs are substantially higher.

 

[quaere]

early adopters of expensive high margin CPU technology are paying for the r&d

 

[zoku88]

early adopters of expensive high margin CPU technology are paying for the r&d

Actually, mostly server chips pay for their R&D.

Anyway, I think the reason they can't do what the OP is suggesting is because of yields. Not quite sure, though.

 

[JCBossman]

Well thats a good point with the USB drives, but I always figured the smaller ones were just last years model, and all they are making NOW is 8gb ones, but do they really make the smaller chips? and the mid size, and the top of the line? Or Maybe a 8gb drive has 4 chips, a 4gb 2, and a 2gb 1 of THE SAME chip, that would make sense.

As far as the shoes, that's just marketing. Coach bags and Channel Perfume obviously don't cost much more than regular stuff, it's just mark up.

But I am talking PURPOSELY making old/obsolete stuff at the same cost as making state of the art.

It's even WORSE with software companies, no "private user" is going to go out and BUY Office 2007 for $500+, but alot of people would buy it at $100, so sell it for Non Commercial use for $100 and to businesses for a $1,000

 

[quaere]

Actually, mostly server chips pay for their R&D.

Anyway, I think the reason they can't do what the OP is suggesting is because of yields. Not quite sure, though.

uh what. server chips are the same as high performance desktop parts. the same technology, the same high profit margins. a CPU company doesn't care if it sells tons of server parts or tons of hardcore games chips. same economic result.
think of it this way. currently there are high margin performance parts and low margin budget parts. to recoup r&d costs if there was only one part, there would have to be a middle margin, middle cost CPU. sure, that is better for the high end purchaser and you arguably get more for your money. but it completely removes the option of paying less for those who want a budget chip.

 

[The Interrobanger]

Well thats a good point with the USB drives, but I always figured the smaller ones were just last years model, and all they are making NOW is 8gb ones, but do they really make the smaller chips? and the mid size, and the top of the line? Or Maybe a 8gb drive has 4 chips, a 4gb 2, and a 2gb 1 of THE SAME chip, that would make sense.

As far as the shoes, that's just marketing. Coach bags and Channel Perfume obviously don't cost much more than regular stuff, it's just mark up.

But I am talking PURPOSELY making old/obsolete stuff at the same cost as making state of the art.

It's even WORSE with software companies, no "private user" is going to go out and BUY Office 2007 for $500+, but alot of people would buy it at $100, so sell it for Non Commercial use for $100 and to businesses for a $1,000

Yes, we know this...isn't the point of the thread to teach you how basic manufacturing economics work?

 

[TransTrender]

Because of product development, engineering, R&D, testing, physical limitations, market expectations....and uhh you can't just shove a whole crapload of bleeding edge technologies inside one package or product and expect every feature to work exactly correctly from day one and if you do shove them all into one package from the beginning you'll be in debug for years before everything works correctly. Generally technologies progress over time because more features are integrated into the product/package so they can develop and work as expected before you add another feature. Hope this helps.

 

[JzeroT1437]

WHY DONT WE ALL HAVE THE SAME AMOUNT OF MONEY? WHY IS IT RAINY SOME DAYS, BUT SUNNY OTHERS?

 

[zoku88]

uh what. server chips are the same as high performance desktop parts. the same technology, the same high profit margins. a CPU company doesn't care if it sells tons of server parts or tons of hardcore games chips. same economic result.
think of it this way. currently there are high margin performance parts and low margin budget parts. to recoup r&d costs if there was only one part, there would have to be a middle margin, middle cost CPU. sure, that is better for the high end purchaser and you arguably get more for your money. but it completely removes the option of paying less for those who want a budget chip.

Except that's not true. They are NOT the same. I'm not sure why you would think that. In fact, not all server chips are even high performance! I'm not sure what you're talking about. :-/

And no, not all chips have high profit margins. Where did you hear all of this BS from?

 

[JCBossman]

OK than tell me this say you are Toshiba and spent Billions developing the technical processes for HD DVD, now the ACTUAL parts, don't cost that much more than a regular DVD player, it's the R&D thats the BIG bucks.Now here is the issue do you:
A:Sell/License your tech for pretty high price to recoup your EXTENSIVE R&D expenditures, but with the risk your format dies before it has a chance to blossom because the added cost slowed adoption?
B. Give the Tech away until you ensure your tech gets entrenched and thereby guaranteeing sales for many years?

I would pick B

 

[zoku88]

OK than tell me this say you are Toshiba and spent Billions developing the technical processes for HD DVD, now the ACTUAL parts, don't cost that much more than a regular DVD player, it's the R&D thats the BIG bucks.Now here is the issue do you:
A:Sell/License your tech for pretty high price to recoup your EXTENSIVE R&D expenditures, but with the risk your format dies before it has a chance to blossom because the added cost slowed adoption?
B. Give the Tech away until you ensure your tech gets entrenched and thereby guaranteeing sales for many years?

I would pick B

When you give the tech away, how are you going to make money? There's no guarantee that consumers will buy YOUR HD-DVD players (they might buy those from competitors) therefore you would have to invest more into make much superior HD-DVD players to ensure that you get money from developing HD-DVD in the first place.

 

[JCBossman]

Well I am assuming they get paid by ALL manufacturers as they developed it. And just because you give away something today doesn't mean you cant start charging tomorrow. Or maybe in this case you start charging AFTER you sell a certain amount of units, say 2 million.

You aren't going to make ANY money if your tech is DOA

 

[quaere]

Except that's not true. They are NOT the same. I'm not sure why you would think that. In fact, not all server chips are even high performance! I'm not sure what you're talking about. :-/

And no, not all chips have high profit margins. Where did you hear all of this BS from?

and I'm not so sure you know what you're talking about. anyway, i've postedthe economic explanation to this question. believe it, or just think the evil corporations are out to steal all your money with arbitrary pricing, it's your choice.

 

[Fisticuffs]

Marketing costs are substantially higher.

you fell for it.

 

[Furoba]

I'd rather have chips competing for efficiëncy in power consumption than performance though.
Notebooks got to a level where they can almost compete with desktop computers, yet battery life is still kind of constrained IMO.

 

[DonasaurusRex]

Alot of times the higher end chips are clocked higher at the fab which means fewer will actually be validated at those speeds hence lower quantities = higher price. Also many times they have more cache which is pretty much the most expensive memory there is outside of the memory for the registers on a chip. Other than that yeah you are paying for name and having the highest performing part at the time...how much higher varies by generation. Other than more on board memory, perhaps an unlocked multiplier, and higher factory validated clock speed there is no difference from the enthusiast chip and the performance or mainstream chip. EXCEPT in the fab the enthusiast chip came out working better, which is why waiting for process refinements is great for buying a cheap baseline chip and overclocking them. By the time a B2 or higher stepping is out most of the lower chips just have errors in their cache and so its disabled, otherwise you can clock em just as high as the enthusiast parts. There aren't any micro architecture differences.

To answer why they dont just release the best part is because they would price alot of people out of buying cpu's if they only released the best parts their yields would be horrible and would have to charge obscene amts to cover the price. Basically that IS what video cards manufacturers do at release of new card generations then quickly go to the cpu style price structure.

For example the 2600XT is a 2900XT with some shader units that didnt pass verification or memory that wouldnt clock high enough, so the answer is to take the base chip and disable the shader units, clock the memory lower, turn off some of the pipelines etc then you got a midrange card. If you guys remember after the 9700 process became so well refined you could flash a 9500 to a 9700 pro because it was the same card that was purposely disabled.

 

[Danj]

Wouldn't be better for computing if intel made NOTHING but Extreme Quads, the good ones sell for a nice premium and the somewhat messed up ones as "Mainstream" parts.

Um.. isn't this basically what they already do? The ones that are stable at the highest speeds get sold at a premium, and the lower speed ones are sorted in to lower speed bins and sold at regular prices?

 

[Borys]

How do I made CPU?

 

[Burger]

Alot of times the higher end chips are clocked higher at the fab which means fewer will actually be validated at those speeds hence lower quantities = higher price. Also many times they have more cache which is pretty much the most expensive memory there is outside of the memory for the registers on a chip. Other than that yeah you are paying for name and having the highest performing part at the time...how much higher varies by generation. Other than more on board memory, perhaps an unlocked multiplier, and higher factory validated clock speed there is no difference from the enthusiast chip and the performance or mainstream chip. EXCEPT in the fab the enthusiast chip came out working better, which is why waiting for process refinements is great for buying a cheap baseline chip and overclocking them. By the time a B2 or higher stepping is out most of the lower chips just have errors in their cache and so its disabled, otherwise you can clock em just as high as the enthusiast parts. There aren't any micro architecture differences.

To answer why they dont just release the best part is because they would price alot of people out of buying cpu's if they only released the best parts their yields would be horrible and would have to charge obscene amts to cover the price. Basically that IS what video cards manufacturers do at release of new card generations then quickly go to the cpu style price structure.

For example the 2600XT is a 2900XT with some shader units that didnt pass verification or memory that wouldnt clock high enough, so the answer is to take the base chip and disable the shader units, clock the memory lower, turn off some of the pipelines etc then you got a midrange card. If you guys remember after the 9700 process became so well refined you could flash a 9500 to a 9700 pro because it was the same card that was purposely disabled.

This is the correct answer.

You don't go and make a wafer of 500 chips and throw away 387 of them because they do not perform at 100%. The ones that perform at 90% are used, the ones at 80% are used and so on. It's all about the yield.

Obviously that way of producing chips also translates to a great business model.