are the price differences between intel chips purely software/patent things ?

[ahmedabady2005]

hello, everyone

I know that in chip industry all chips undergo the same manufacturing processes in any given generation (Sandy bridge,ivy bridge,haswell)

say for example i3/i5/i7 haswell chips, all of them are made on 300mm wafers which has the same price/yielding rate /22 nm process. so, from my understanding the cost of making the entry level core i3 chip should be nearly the same as core i7 extreme edition, while in the real life the i7EE cost more than triple the price of i3
​

same logic could be applied on ARM based chips

I wonder if the price differences between different chips from any given maker is purely software/patent related(i.e. R&D of new techs and marketing) without actual difference in manufacturing costs

I know this question seems naive for some as I have just recently started educating myself in the field of semiconductors
so pardon my ignorance and my english

 

[SecurityTheatre]

hello, everyone

I know that in chip industry all chips undergo the same manufacturing processes in any given generation (Sandy bridge,ivy bridge,haswell)

say for example i3/i5/i7 haswell chips, all of them are made on 300mm wafers which has the same price/yielding rate /22 nm process. so, from my understanding the cost of making the entry level core i3 chip should be nearly the same as core i7 extreme edition, while in the real life the i7EE cost more than triple the price of i3
​

same logic could be applied on ARM based chips

I wonder if the price differences between different chips from any given maker is purely software/patent related(i.e. R&D of new techs and marketing) without actual difference in manufacturing costs

I know this question seems naive for some as I have just recently started educating myself in the field of semiconductors
so pardon my ignorance and my english

First, the i7 chip has a much larger die than the i3. In fact, I'm not certain (and it varies by model), but I think it's on the order of 3-4x bigger.

Now, there probably are some different lines (i7 vs i7 EE) that are simply "binning" differences. That is to say, the chips as they come off the line, are "binned" according to their speed capability. Those with the most precise build will run faster than those with slight imperfections and will be rated to higher speeds.

So, to be clear, the i3 has a real reason to be much cheaper than the i7, but to some extent, there are simply marketing and profitability reasons to charge different amounts. The yeild


Assuming all chips are identical and are all basically top-end i7 chips, they STILL have an incentive to sell them at different prices for corporate reasons. First, if a chip is too expensive, it will not sell well because competitors will eat away the low end of the market. However, in order to increase profitability, they want to sell some chips at as high a price as possible.

As a result, they create 'tiers' of chips. Some they sell for cheap at the bottom end (to sell more units and still make some profit) and some they sell at the high end (to sell their best units at maximum profit).

So between there being different lines (i3 vs i7) with different inherent costs (die size, yield due to complexity, easier design, etc) and different "bins" due to testing yield (tested speed, thermal design, etc) and finally, different tiers due entirely to marketing reasons (extreme edition, etc), they have a complete lineup of products.

 

[mjd]

Prices are generally driven by Economics, Supply & Demand and less by costs.
The consumers want the new, better, featured chips more than they want the older, slower, lower-binned chips.

So, your post is certainly on the right track. All those features and quality metrics drive customer demand. And then a business guy does some Economics 101 math to find the optimal price to maximize profit based on that demand and the cost. And that's the super-simple explanation of how pricing works. It's a little more complicated then that in reality, but that's the overview.

-M

 

[Sheep221]

First, the i7 chip has a much larger die than the i3. In fact, I'm not certain (and it varies by model), but I think it's on the order of 3-4x bigger.

Now, there probably are some different lines (i7 vs i7 EE) that are simply "binning" differences. That is to say, the chips as they come off the line, are "binned" according to their speed capability. Those with the most precise build will run faster than those with slight imperfections and will be rated to higher speeds.

So, to be clear, the i3 has a real reason to be much cheaper than the i7, but to some extent, there are simply marketing and profitability reasons to charge different amounts. The yeild


Assuming all chips are identical and are all basically top-end i7 chips, they STILL have an incentive to sell them at different prices for corporate reasons. First, if a chip is too expensive, it will not sell well because competitors will eat away the low end of the market. However, in order to increase profitability, they want to sell some chips at as high a price as possible.

As a result, they create 'tiers' of chips. Some they sell for cheap at the bottom end (to sell more units and still make some profit) and some they sell at the high end (to sell their best units at maximum profit).

So between there being different lines (i3 vs i7) with different inherent costs (die size, yield due to complexity, easier design, etc) and different "bins" due to testing yield (tested speed, thermal design, etc) and finally, different tiers due entirely to marketing reasons (extreme edition, etc), they have a complete lineup of products.

Yeah there are only 2 types of nodes, the i3 and i7, all others are just malfunctioned chips sold for lower price, e.g i5 is not stable with HT so it is disabled, the lower freq i5 is not stable on the same freq as top end i5 and the list goes on, the single core celeron is dual core i3 with one defective core, and the dual core celeron is defective i3 with HT malfunction and half of the working cache, some CPUs have no functional processor graphics and are sold as P models. It all goes just by this.

 

[ahmedabady2005]

thank u all for these informations

 

[Lepton87]

Yeah there are only 2 types of nodes, the i3 and i7, all others are just malfunctioned chips sold for lower price, e.g i5 is not stable with HT so it is disabled, the lower freq i5 is not stable on the same freq as top end i5 and the list goes on, the single core celeron is dual core i3 with one defective core, and the dual core celeron is defective i3 with HT malfunction and half of the working cache, some CPUs have no functional processor graphics and are sold as P models. It all goes just by this.

That's not entirely accurate. For example most 6950 unlocked their shaders from 1408 to 1536, a lot of Phenom X2 unlocked their cores etc. Intel even sells codes to unlock cache on some of their CPUs, so in a lot of cases it's just artificial marketing segmentation.

 

[PsiStar]

Prices are generally driven by Economics, Supply & Demand and less by costs.
The consumers want the new, better, featured chips more than they want the older, slower, lower-binned chips.

So, your post is certainly on the right track. All those features and quality metrics drive customer demand. And then a business guy does some Economics 101 math to find the optimal price to maximize profit based on that demand and the cost. And that's the super-simple explanation of how pricing works. It's a little more complicated then that in reality, but that's the overview.

-M

This. Fundamentally straight out of an Econ. 201 course. Supply and demand and profitability. Pay for your R&D in the initial introduction but be careful about putting the price too high. If it is too high then competition and alternatives will be drawn in more quickly.

Patents are part of this formula also. In many large hi-tech companies a significant investment and business strategy are their patents. Nothing about this process is inexpensive.

I think a problem that technologists have versus non-techs, is that to the non-techies it looks too easy.

 

[stockwiz]

Nvidia charge more for their stuff because they are perceived in a more positive manner and had the better card. AMD has done the same thing in the past. Basically competition is a good thing.

Intel is the market leader in the high end segment. AMD just has not been able to keep up since the Opteron days back before Intel launched Core 2 Duo. Thus at the high end their chips are priced much higher with not much higher production cost... same thing with Nvidia Titan.. when you are the market leader in the high end segment it gives you leeway to price accordingly.

Where else are you going to get your 6 core chip for video encoding that runs at 4+ GHZ that still uses soldering for the IHS and not crappy thermal compound? (that blows away the comparable AMD product in per core performance) Me? I gots me a 2600k around launch for $315 or so I believe. I got my money's worth.

 

[Anarchist420]

yes, imaginary property legislation is a large reason things arent as good as they could be. they result in malinvestment... intel has to have lawyers to protect their ip, for example. then they may make for inefficient r&d because they know that the price monopoly they will have will cover too many r&d mistakes. so not only do people pay often pay higher prices with IP, they have to pay taxes to get higher prices and they may even get a worse product.

think about how it wouldnt even help all that much if ip was reduced in length.

i guess maybe a reasonable reduction would be to only have ip to recover 10% of r&d costs. once the 10% thershold is exceeded, then there is no more legislative monopoly on that particular idea. and no i didnt think of that on my own.

maybe a good compromise bill would be 90% for the first year, then 50% for the next 5, then 10% for the last one, then repeal of all ip at the start of the 8th year.
of course, many big businesses would make another excuse. currently they say they need legislated price monopolies to recoup r&d costs. reduction based upon r&d would then prove that theyre making excuses if they then say they need it for something else. perhaps they should patent their lies instead.