How come intel/amd don't create an ultra cheap processor...

[Philippine Mango]

Like what they could do is build a processor on the latest micron process like .9 or .65 with as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics? Maybe even a 1GHZ processor but the idea is that since they're going with a small micron process, they'd be able to output tons of these much cheaper than just rereleasing those old processors since they take up a huge chunk of those wafers...

And yes, I'm assuming I'm going to hear a lot of posters call me names and tell me that I'm completely retarded and my idea lacks sense and yada yada yada, I DON'T CARE, it's an idea that I think would make economical sense, building a processor like this would allow intel to create sub <$200 computers I believe...

 

[SVT Cobra]

Originally posted by: Philippine Mango
Like what they could do is build a processor on the latest micron process like .9 or .65 with as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics? Maybe even a 1GHZ processor but the idea is that since they're going with a small micron process, they'd be able to output tons of these much cheaper than just rereleasing those old processors since they take up a huge chunk of those wafers...

And yes, I'm assuming I'm going to hear a lot of posters call me names and tell me that I'm completely retarded and my idea lacks sense and yada yada yada, I DON'T CARE, it's an idea that I think would make economical sense, building a processor like this would allow intel to create sub <$200 computers I believe...

you obviosly do not understand the industry, the economics, and the actual fabrication process envolved, not to mention the actual fabs both companies have
/thread

 

[TuxDave]

Something like the Intel xScale for pocket PCs?

 

[Pocatello]

My guess would be it's not profitable, since the cost difference to produce a fast processor and a slower on is pretty close. Intel and AMD make more profit by selling expensive processors, I remember the day when the Athlon XP processors were cheap, and AMD was losing buckets of money. I think VIA is the one who is selling very cheap CPU right now, and can afford to because of lower overall fixed cost.

 

[Philippine Mango]

Originally posted by: Son of a N00b

Originally posted by: Philippine Mango
Like what they could do is build a processor on the latest micron process like .9 or .65 with as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics? Maybe even a 1GHZ processor but the idea is that since they're going with a small micron process, they'd be able to output tons of these much cheaper than just rereleasing those old processors since they take up a huge chunk of those wafers...

And yes, I'm assuming I'm going to hear a lot of posters call me names and tell me that I'm completely retarded and my idea lacks sense and yada yada yada, I DON'T CARE, it's an idea that I think would make economical sense, building a processor like this would allow intel to create sub <$200 computers I believe...

you obviosly do not understand the industry, the economics, and the actual fabrication process envolved, not to mention the actual fabs both companies have
/thread

So what your saying is that since you don't know much about the industry but you do know that it's a complex industry, you assume that intel has thought of every idea plausable and has determined that wouldn't be economically feasible and it would make much more sense to sell processors that are overpowered for their use and cost more than they should for their intended use.. right, alright...

 

[Pocatello]

Originally posted by: Philippine Mango

Originally posted by: Son of a N00b

Originally posted by: Philippine Mango
Like what they could do is build a processor on the latest micron process like .9 or .65 with as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics? Maybe even a 1GHZ processor but the idea is that since they're going with a small micron process, they'd be able to output tons of these much cheaper than just rereleasing those old processors since they take up a huge chunk of those wafers...

And yes, I'm assuming I'm going to hear a lot of posters call me names and tell me that I'm completely retarded and my idea lacks sense and yada yada yada, I DON'T CARE, it's an idea that I think would make economical sense, building a processor like this would allow intel to create sub <$200 computers I believe...

you obviosly do not understand the industry, the economics, and the actual fabrication process envolved, not to mention the actual fabs both companies have
/thread

So what your saying is that since you don't know much about the industry but you do know that it's a complex industry, you assume that intel has thought of every idea plausable and has determined that wouldn't be economically feasible and it would make much more sense to sell processors that are overpowered for their use and cost more than they should for their intended use.. right, alright...

In general, a business does what is profitable, not to bring peace or the cheapest PC to the world, unless it is profitable to bring peace to the world.

 

[Wingznut]

Originally posted by: Philippine Mango
Like what they could do is build a processor on the latest micron process like .9 or .65 with as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics? Maybe even a 1GHZ processor but the idea is that since they're going with a small micron process, they'd be able to output tons of these much cheaper than just rereleasing those old processors since they take up a huge chunk of those wafers...

And yes, I'm assuming I'm going to hear a lot of posters call me names and tell me that I'm completely retarded and my idea lacks sense and yada yada yada, I DON'T CARE, it's an idea that I think would make economical sense, building a processor like this would allow intel to create sub <$200 computers I believe...

Do you think there's a market for a 1ghz sub-$200 PC?

I'll absolutely guarantee that if Intel/AMD/Dell/Gateway/etc thought so... They'd make it happen.

Via/Cyrix makes a low speed, low cost, low power cpu... And nobody really gives a rat's ass.

 

[igowerf]

Originally posted by: TuxDave
Something like the Intel xScale for pocket PCs?

That's what I was thinking too. xScale is used in a fair amount of small devices, not limited to PDAs and phones. My Axim x30 High is powerful enough to playback an HDTV recorded divx file and scale it to fit the 320x240 res screen.

 

[Neurorelay]

bring peace to the world.

Key factor right there.

 

[Philippine Mango]

Originally posted by: Wingznut

Originally posted by: Philippine Mango
Like what they could do is build a processor on the latest micron process like .9 or .65 with as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics? Maybe even a 1GHZ processor but the idea is that since they're going with a small micron process, they'd be able to output tons of these much cheaper than just rereleasing those old processors since they take up a huge chunk of those wafers...

And yes, I'm assuming I'm going to hear a lot of posters call me names and tell me that I'm completely retarded and my idea lacks sense and yada yada yada, I DON'T CARE, it's an idea that I think would make economical sense, building a processor like this would allow intel to create sub <$200 computers I believe...

Do you think there's a market for a 1ghz sub-$200 PC?

I'll absolutely guarantee that if Intel/AMD/Dell/Gateway/etc thought so... They'd make it happen.

Via/Cyrix makes a low speed, low cost, low power cpu... And nobody really gives a rat's ass.

Thing is, Via/Cyrix processors don't perfrom equally as a similarly clock PIII/PIV, they're slow as molasses...

 

[Wingznut]

Originally posted by: Philippine Mango

Originally posted by: Wingznut

Originally posted by: Philippine Mango
Like what they could do is build a processor on the latest micron process like .9 or .65 with as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics? Maybe even a 1GHZ processor but the idea is that since they're going with a small micron process, they'd be able to output tons of these much cheaper than just rereleasing those old processors since they take up a huge chunk of those wafers...

And yes, I'm assuming I'm going to hear a lot of posters call me names and tell me that I'm completely retarded and my idea lacks sense and yada yada yada, I DON'T CARE, it's an idea that I think would make economical sense, building a processor like this would allow intel to create sub <$200 computers I believe...

Do you think there's a market for a 1ghz sub-$200 PC?

I'll absolutely guarantee that if Intel/AMD/Dell/Gateway/etc thought so... They'd make it happen.

Via/Cyrix makes a low speed, low cost, low power cpu... And nobody really gives a rat's ass.

Thing is, Via/Cyrix processors don't perfrom equally as a similarly clock PIII/PIV, they're slow as molasses...

Yeah, but I don't think the demographic who is interested in a sub-$200 PC cares. It's not like you'd notice any performance difference between the two processors with the applications that would go along with that kind of setup.

 

[dmens]

as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics?

Wasn't one of those demonstrated at the 2005 IDC? There's already designs in progress to extend the xscale line in performance and power margins.

Why bother designing a seperate chip for cheap pc's. Just use the same chip and chop off the other cores and/or cache. The xscale line will never be able to run vista or similar software, so it will always cater to a seperate market.

 

[nerp]

There are more chip makers out there than AMD and Intel. Via, Samsung, TI, IBM and Motorola all build processors for a variety of applications.

 

[aka1nas]

It isn't much cheaper to build a less capable processor than a more powerful one if yields on both are acceptable. The vast majority of the cost of a CPU is in the fabrication facilities, which cost several billion dollars each to build. The marginal cost in materials etc to crank out an individual CPU is pretty low. I read once that the material cost to produce a pentium 4 is somewhere around $23 (wingnut or dmems or someone more familiar could probably give a better estimate). However, they have to amortize the tremendous cost of the fabs over the lifetime of it's production run.

 

[Philippine Mango]

Originally posted by: aka1nas
It isn't much cheaper to build a less capable processor than a more powerful one if yields on both are acceptable. The vast majority of the cost of a CPU is in the fabrication facilities, which cost several billion dollars each to build. The marginal cost in materials etc to crank out an individual CPU is pretty low. I read once that the material cost to produce a pentium 4 is somewhere around $23 (wingnut or dmems or someone more familiar could probably give a better estimate). However, they have to amortize the tremendous cost of the fabs over the lifetime of it's production run.

Yes but a P4 has significantly more transistors than say a Pentium III, I'm not sure though but I believe the P4 is larger despite it having a smaller micron process but imagine taking a relatively low transistor count of a Pentium III and creating it agian, say you get 100 yeilds from .15 micro process, well as you go down, you multiply those yeilds by some factor (I don't know) but it could easily be a factor of 5 so you get say 500 yeilds instead of 100, I think thats pretty damn good if you ask me, therefore reducing the price the processor. (I know these numbers aren't even close to being accurate or representative, but you get the idea).

Because you go to smaller fabrication processes, you are able to pack more transistors in the same amount of space, Plus you don't have excess transistors in the P4 relative to the Pentium 3 so it should techincally become even cheaper. It's not like they'd have to redesign the processor or anything, they can just use old tech sheets but make it smaller... It's possible that I'm just making it seem to be to simple...

 

[Wingznut]

Originally posted by: aka1nas
It isn't much cheaper to build a less capable processor than a more powerful one if yields on both are acceptable. The vast majority of the cost of a CPU is in the fabrication facilities, which cost several billion dollars each to build. The marginal cost in materials etc to crank out an individual CPU is pretty low. I read once that the material cost to produce a pentium 4 is somewhere around $23 (wingnut or dmems or someone more familiar could probably give a better estimate). However, they have to amortize the tremendous cost of the fabs over the lifetime of it's production run.

Actually sorry, no I couldn't... Those kinds of numbers are very confidential.

But here's another thought to chew on... Right now, Intel and AMD can sell pretty much every chip they make. They have no excess capacity. So, they wouldn't want to use that extremely valuable fab space to create a cheap cpu that has little demand, and would generate even less revenue.

 

[coldpower27]

Originally posted by: Philippine Mango

Originally posted by: aka1nas
It isn't much cheaper to build a less capable processor than a more powerful one if yields on both are acceptable. The vast majority of the cost of a CPU is in the fabrication facilities, which cost several billion dollars each to build. The marginal cost in materials etc to crank out an individual CPU is pretty low. I read once that the material cost to produce a pentium 4 is somewhere around $23 (wingnut or dmems or someone more familiar could probably give a better estimate). However, they have to amortize the tremendous cost of the fabs over the lifetime of it's production run.

Yes but a P4 has significantly more transistors than say a Pentium III, I'm not sure though but I believe the P4 is larger despite it having a smaller micron process but imagine taking a relatively low transistor count of a Pentium III and creating it agian, say you get 100 yeilds from .15 micro process, well as you go down, you multiply those yeilds by some factor (I don't know) but it could easily be a factor of 5 so you get say 500 yeilds instead of 100, I think thats pretty damn good if you ask me, therefore reducing the price the processor. (I know these numbers aren't even close to being accurate or representative, but you get the idea).

Because you go to smaller fabrication processes, you are able to pack more transistors in the same amount of space, Plus you don't have excess transistors in the P4 relative to the Pentium 3 so it should techincally become even cheaper. It's not like they'd have to redesign the processor or anything, they can just use old tech sheets but make it smaller... It's possible that I'm just making it seem to be to simple...

it doesn't work out this way, you also have packaging cost to add into the mix in addition to the die itself. You also can't jsut keep giving the consumer the same product over and over. IF you just make the processor smaller and not add squat your not increasing performance.

Pentium 3 performance is enough for web surfing, and office, but nowhere near enough for encoding, or gaming.

If you assumed 100mm2 for Pentium 3 on the 0.18 micron process, then on 0.13 micron it would be 60mm2 for a dumb shrink, on 0.09 micron your down 36 mm2, and 21.6mm2 once you reached 65nm as we have now. However the packaging would remain constant, so your not saving that much. You also need to increase performance over time, dumb shrinks alone can't do that.

 

[Philippine Mango]

Originally posted by: coldpower27

Originally posted by: Philippine Mango

Originally posted by: aka1nas
It isn't much cheaper to build a less capable processor than a more powerful one if yields on both are acceptable. The vast majority of the cost of a CPU is in the fabrication facilities, which cost several billion dollars each to build. The marginal cost in materials etc to crank out an individual CPU is pretty low. I read once that the material cost to produce a pentium 4 is somewhere around $23 (wingnut or dmems or someone more familiar could probably give a better estimate). However, they have to amortize the tremendous cost of the fabs over the lifetime of it's production run.

Yes but a P4 has significantly more transistors than say a Pentium III, I'm not sure though but I believe the P4 is larger despite it having a smaller micron process but imagine taking a relatively low transistor count of a Pentium III and creating it agian, say you get 100 yeilds from .15 micro process, well as you go down, you multiply those yeilds by some factor (I don't know) but it could easily be a factor of 5 so you get say 500 yeilds instead of 100, I think thats pretty damn good if you ask me, therefore reducing the price the processor. (I know these numbers aren't even close to being accurate or representative, but you get the idea).

Because you go to smaller fabrication processes, you are able to pack more transistors in the same amount of space, Plus you don't have excess transistors in the P4 relative to the Pentium 3 so it should techincally become even cheaper. It's not like they'd have to redesign the processor or anything, they can just use old tech sheets but make it smaller... It's possible that I'm just making it seem to be to simple...

it doesn't work out this way, you also have packaging cost to add into the mix in addition to the die itself. You also can't jsut keep giving the consumer the same product over and over. IF you just make the processor smaller and not add squat your not increasing performance.

Pentium 3 performance is enough for web surfing, and office, but nowhere near enough for encoding, or gaming.

If you assumed 100mm2 for Pentium 3 on the 0.18 micron process, then on 0.13 micron it would be 60mm2 for a dumb shrink, on 0.09 micron your down 36 mm2, and 21.6mm2 once you reached 65nm as we have now. However the packaging would remain constant, so your not saving that much. You also need to increase performance over time, dumb shrinks alone can't do that.

That is EXACTLY my point. Most people do simple tasks, a PIII 800/1GHZ is plenty for encoding as well, just not as fast as say your rig or my rig. The idea is to make this cheaper and cheaper to eventually these processors could cost around $10 or so, and if they do things right, they could make them faster even if they maintain the same clock speeds.

The packaging could be smaller and fewer pins than that of a P4. The general idea with smaller die sizes is: more transistors in the same area with out spending more money by using up more of the wafer space. If you keep making the same processor with maybe tiny increments in speed increase (either MHZ wise or efficiency wise) but the general idea is to make an ultra cheap CPU, something third worlds could benefit from and people within out own country..

 

[lyssword]

you can buy pIII 933 for like $40 retail box..

 

[coldpower27]

Originally posted by: Philippine Mango

Originally posted by: coldpower27

Originally posted by: Philippine Mango

Originally posted by: aka1nas
It isn't much cheaper to build a less capable processor than a more powerful one if yields on both are acceptable. The vast majority of the cost of a CPU is in the fabrication facilities, which cost several billion dollars each to build. The marginal cost in materials etc to crank out an individual CPU is pretty low. I read once that the material cost to produce a pentium 4 is somewhere around $23 (wingnut or dmems or someone more familiar could probably give a better estimate). However, they have to amortize the tremendous cost of the fabs over the lifetime of it's production run.

Yes but a P4 has significantly more transistors than say a Pentium III, I'm not sure though but I believe the P4 is larger despite it having a smaller micron process but imagine taking a relatively low transistor count of a Pentium III and creating it agian, say you get 100 yeilds from .15 micro process, well as you go down, you multiply those yeilds by some factor (I don't know) but it could easily be a factor of 5 so you get say 500 yeilds instead of 100, I think thats pretty damn good if you ask me, therefore reducing the price the processor. (I know these numbers aren't even close to being accurate or representative, but you get the idea).

Because you go to smaller fabrication processes, you are able to pack more transistors in the same amount of space, Plus you don't have excess transistors in the P4 relative to the Pentium 3 so it should techincally become even cheaper. It's not like they'd have to redesign the processor or anything, they can just use old tech sheets but make it smaller... It's possible that I'm just making it seem to be to simple...

it doesn't work out this way, you also have packaging cost to add into the mix in addition to the die itself. You also can't jsut keep giving the consumer the same product over and over. IF you just make the processor smaller and not add squat your not increasing performance.

Pentium 3 performance is enough for web surfing, and office, but nowhere near enough for encoding, or gaming.

If you assumed 100mm2 for Pentium 3 on the 0.18 micron process, then on 0.13 micron it would be 60mm2 for a dumb shrink, on 0.09 micron your down 36 mm2, and 21.6mm2 once you reached 65nm as we have now. However the packaging would remain constant, so your not saving that much. You also need to increase performance over time, dumb shrinks alone can't do that.

That is EXACTLY my point. Most people do simple tasks, a PIII 800/1GHZ is plenty for encoding as well, just not as fast as say your rig or my rig. The idea is to make this cheaper and cheaper to eventually these processors could cost around $10 or so, and if they do things right, they could make them faster even if they maintain the same clock speeds.

The packaging could be smaller and fewer pins than that of a P4. The general idea with smaller die sizes is: more transistors in the same area with out spending more money by using up more of the wafer space. If you keep making the same processor with maybe tiny increments in speed increase (either MHZ wise or efficiency wise) but the general idea is to make an ultra cheap CPU, something third worlds could benefit from and people within out own country..

It depends on what your encoding 10 hours is not an acceptable timeframe to me, for a DVD Rip, the packaging would be Socket 370. It still would remain constant cost that wont be reduced. And I think it has been mentioned, Intel is a coporation and not a charity, it in the buisness to make money. Also to keep in mind voltage would need to be reduce in each sucessive die shrink. So you would still need to design new motherboards over time.
It simply just isn't profitable to do it this way, any changes to efficiency would require changes in die size as your adding something.

$10 is not profitable at all, you would have to sell so many, and why go to the trouble when you make 10 times more from selling a single Celeron D and those are already sold in good quantities anyway. Not to mention new fabrication plants and process cost money even with exiting architectures. Intel is currently FAB limited now, it make no sense to produce such a rpoduct to take away from their office PC's.

 

[Philippine Mango]

Originally posted by: coldpower27

Originally posted by: Philippine Mango

Originally posted by: coldpower27

Originally posted by: Philippine Mango

Originally posted by: aka1nas
It isn't much cheaper to build a less capable processor than a more powerful one if yields on both are acceptable. The vast majority of the cost of a CPU is in the fabrication facilities, which cost several billion dollars each to build. The marginal cost in materials etc to crank out an individual CPU is pretty low. I read once that the material cost to produce a pentium 4 is somewhere around $23 (wingnut or dmems or someone more familiar could probably give a better estimate). However, they have to amortize the tremendous cost of the fabs over the lifetime of it's production run.

Yes but a P4 has significantly more transistors than say a Pentium III, I'm not sure though but I believe the P4 is larger despite it having a smaller micron process but imagine taking a relatively low transistor count of a Pentium III and creating it agian, say you get 100 yeilds from .15 micro process, well as you go down, you multiply those yeilds by some factor (I don't know) but it could easily be a factor of 5 so you get say 500 yeilds instead of 100, I think thats pretty damn good if you ask me, therefore reducing the price the processor. (I know these numbers aren't even close to being accurate or representative, but you get the idea).

Because you go to smaller fabrication processes, you are able to pack more transistors in the same amount of space, Plus you don't have excess transistors in the P4 relative to the Pentium 3 so it should techincally become even cheaper. It's not like they'd have to redesign the processor or anything, they can just use old tech sheets but make it smaller... It's possible that I'm just making it seem to be to simple...

it doesn't work out this way, you also have packaging cost to add into the mix in addition to the die itself. You also can't jsut keep giving the consumer the same product over and over. IF you just make the processor smaller and not add squat your not increasing performance.

Pentium 3 performance is enough for web surfing, and office, but nowhere near enough for encoding, or gaming.

If you assumed 100mm2 for Pentium 3 on the 0.18 micron process, then on 0.13 micron it would be 60mm2 for a dumb shrink, on 0.09 micron your down 36 mm2, and 21.6mm2 once you reached 65nm as we have now. However the packaging would remain constant, so your not saving that much. You also need to increase performance over time, dumb shrinks alone can't do that.

That is EXACTLY my point. Most people do simple tasks, a PIII 800/1GHZ is plenty for encoding as well, just not as fast as say your rig or my rig. The idea is to make this cheaper and cheaper to eventually these processors could cost around $10 or so, and if they do things right, they could make them faster even if they maintain the same clock speeds.

The packaging could be smaller and fewer pins than that of a P4. The general idea with smaller die sizes is: more transistors in the same area with out spending more money by using up more of the wafer space. If you keep making the same processor with maybe tiny increments in speed increase (either MHZ wise or efficiency wise) but the general idea is to make an ultra cheap CPU, something third worlds could benefit from and people within out own country..

It depends on what your encoding 10 hours is not an acceptable timeframe to me, for a DVD Rip, the packaging would be Socket 370. It still would remain constant cost that wont be reduced. And I think it has been mentioned, Intel is a coporation and not a charity, it in the buisness to make money. Also to keep in mind voltage would need to be reduce in each sucessive die shrink. So you would still need to design new motherboards over time.
It simply just isn't profitable to do it this way, any changes to efficiency would require changes in die size as your adding something.

$10 is not profitable at all, you would have to sell so many, and why go to the trouble when you make 10 times more from selling a single Celeron D and those are already sold in good quantities anyway. Not to mention new fabrication plants and process cost money even with exiting architectures. Intel is currently FAB limited now, it make no sense to produce such a rpoduct to take away from their office PC's.

Don't you mean AMD is fab limited? Intel is NOW having trouble selling it's processors because less people are buying them now. I'm sure intel could make more money if they just sold $300 processors like they were doing before, but the idea is to appeal to different markets, plus if they go cheap they could probably get some sort of government incentive or something. This isn't just an idea for intel as AMD makes cheaper processors in general so either one of them could make them.

For the most part, I've been thinking of that processor that goes into that $100 processor and how they could do so much more with it like I said before, lower voltage, small die, therefore making it cheap and widely available for the masses who need only a quick machine to do everyday tasks and not some sluggish 2.4 celeron with 128MB of ram, it's funny how a computer with 1.4GHZ less computing power could be so much faster with an upgrade of 128MB of more ram, or less bloat

 

[Fox5]

Originally posted by: Philippine Mango
Like what they could do is build a processor on the latest micron process like .9 or .65 with as many transistors as say a 600mhz or 800mhz PIII processor or (AMD equivalent of that), since it's such a small processor relative to the ones today, it could incorperate a memory controller and maybe graphics? Maybe even a 1GHZ processor but the idea is that since they're going with a small micron process, they'd be able to output tons of these much cheaper than just rereleasing those old processors since they take up a huge chunk of those wafers...

And yes, I'm assuming I'm going to hear a lot of posters call me names and tell me that I'm completely retarded and my idea lacks sense and yada yada yada, I DON'T CARE, it's an idea that I think would make economical sense, building a processor like this would allow intel to create sub <$200 computers I believe...

1. The current athlons and pentium ms are pretty much the P3 and Athlon, but with stuff added on. Most of the current cpus are cache; cut the cache down to 1998/1999 levels and you just considerably dropped the price of the cpu.
2. Both companies have had ideas to do as you say. Intel with Tinma, AMD with their Geode cpus and their little $200 Linux/Mobile Windows powered Internet gateway thing. (the little green box!)
3. Supply and demand, they try to maximize profit. Flooding the market with cheap chips may not do that.

 

[rchiu]

Originally posted by: Philippine Mango

Originally posted by: coldpower27

Originally posted by: Philippine Mango

Originally posted by: aka1nas
It isn't much cheaper to build a less capable processor than a more powerful one if yields on both are acceptable. The vast majority of the cost of a CPU is in the fabrication facilities, which cost several billion dollars each to build. The marginal cost in materials etc to crank out an individual CPU is pretty low. I read once that the material cost to produce a pentium 4 is somewhere around $23 (wingnut or dmems or someone more familiar could probably give a better estimate). However, they have to amortize the tremendous cost of the fabs over the lifetime of it's production run.

Yes but a P4 has significantly more transistors than say a Pentium III, I'm not sure though but I believe the P4 is larger despite it having a smaller micron process but imagine taking a relatively low transistor count of a Pentium III and creating it agian, say you get 100 yeilds from .15 micro process, well as you go down, you multiply those yeilds by some factor (I don't know) but it could easily be a factor of 5 so you get say 500 yeilds instead of 100, I think thats pretty damn good if you ask me, therefore reducing the price the processor. (I know these numbers aren't even close to being accurate or representative, but you get the idea).

Because you go to smaller fabrication processes, you are able to pack more transistors in the same amount of space, Plus you don't have excess transistors in the P4 relative to the Pentium 3 so it should techincally become even cheaper. It's not like they'd have to redesign the processor or anything, they can just use old tech sheets but make it smaller... It's possible that I'm just making it seem to be to simple...

it doesn't work out this way, you also have packaging cost to add into the mix in addition to the die itself. You also can't jsut keep giving the consumer the same product over and over. IF you just make the processor smaller and not add squat your not increasing performance.

Pentium 3 performance is enough for web surfing, and office, but nowhere near enough for encoding, or gaming.

If you assumed 100mm2 for Pentium 3 on the 0.18 micron process, then on 0.13 micron it would be 60mm2 for a dumb shrink, on 0.09 micron your down 36 mm2, and 21.6mm2 once you reached 65nm as we have now. However the packaging would remain constant, so your not saving that much. You also need to increase performance over time, dumb shrinks alone can't do that.

That is EXACTLY my point. Most people do simple tasks, a PIII 800/1GHZ is plenty for encoding as well, just not as fast as say your rig or my rig. The idea is to make this cheaper and cheaper to eventually these processors could cost around $10 or so, and if they do things right, they could make them faster even if they maintain the same clock speeds.

The packaging could be smaller and fewer pins than that of a P4. The general idea with smaller die sizes is: more transistors in the same area with out spending more money by using up more of the wafer space. If you keep making the same processor with maybe tiny increments in speed increase (either MHZ wise or efficiency wise) but the general idea is to make an ultra cheap CPU, something third worlds could benefit from and people within out own country..

Well, there are already pretty cheap processors on the Market. The value chips from Intel/AMD cost less than $100 bucks retail already, and I think some cost less than $50 to the OEM companies buying in bulk. But remember the cpu is not the entire cost of the PC. Even if you have a $10 cpu, the other cost like memory, mobo, HDD, video card, and especially adding up the MS windows still gonna cost you over $200~$300 bucks. Simply driving down the processor cost is not going to buy you much. Plus remember every companies that sells those components I just memtioned would like to make mony. Company that put those things together want to make money. After all the mark up each company gets, the consumer is gonna pay quite a bit for the final product.

That's why you see when companies like Dell who can use their volume and negotiate better deals on components and drive down the cost putting PC together. That's why they can price their PC to close to $200 level (without OS in some of their sever deal).

So what I am trying to say is you won't achieve your goal of low price PC simply through lowering the price of CPU. There is much more to it when manufacturing a PC.

 

[Wingznut]

Originally posted by: Philippine Mango
Don't you mean AMD is fab limited? Intel is NOW having trouble selling it's processors because less people are buying them now.

Where'd you get that at?


Intel Announces Record Revenue for Q3 2005

?In the third quarter, we achieved all-time records in company revenue and unit shipments across all of our major product lines,? said Paul Otellini, Intel president and CEO.

 

[aLeoN]

Originally posted by: Philippine Mango
[Yes but a P4 has significantly more transistors than say a Pentium III, I'm not sure though but I believe the P4 is larger despite it having a smaller micron process but imagine taking a relatively low transistor count of a Pentium III and creating it agian, say you get 100 yeilds from .15 micro process, well as you go down, you multiply those yeilds by some factor (I don't know) but it could easily be a factor of 5 so you get say 500 yeilds instead of 100, I think thats pretty damn good if you ask me, therefore reducing the price the processor. (I know these numbers aren't even close to being accurate or representative, but you get the idea).

Because you go to smaller fabrication processes, you are able to pack more transistors in the same amount of space, Plus you don't have excess transistors in the P4 relative to the Pentium 3 so it should techincally become even cheaper. It's not like they'd have to redesign the processor or anything, they can just use old tech sheets but make it smaller... It's possible that I'm just making it seem to be to simple...

When the processor is made smaller, they have less room (literally & figuratively) to screw it up. Using a brand new, non-established smaller micron process generally yields less successfully than a larger micron one that's been around for years. Plus, unless all the other computer hardware manufacturers would like to step back into time and stay there, why would anyone get today's system that's actually bottlenecked by the cpu?

Maybe one day your vision could come true if intel made the other key components of the computer as well.