[Phil]

Look again, heat isn't going up. TDP trends have been staying the same or decreasing, chip sizes stay similar. Integration is only brought in when it's offset by process gains. No one is really pushing against this. And there's still a fair amount of process improvement headroom with not all that much stuff that needs to be integrated. The motherboard cost isn't "offset because they are manufactured in bulk", motherboard complexity costs scale with volume. A northbridge with memory controller may not have cost $30-40 extra vs one without, in and of itself, but it definitely adds board complexity due to the additional routing that's necessary. And if you have to add say, 2+ layers to a board of that size, you absolutely feel it in manufacturing costs.

The motherboards didn't cost more just because Intel was controlling the prices, obviously I was looking at non-Intel motherboards where the market is still very competitive. Chipset costs from Intel vs AMD would have played a role, but I doubt that was all there was to the difference in cost between motherboards from the same manufacturer. Of course the AMD processors had better perf/$ too, but I wasn't even comparing those. Just the motherboard costs, not the system costs. If the motherboards cost similar amounts I would have built a Core 2 system. The Intel motherboards were well over $50 more expensive.

The CPU's of today are more efficient at controlling heat vs previous designs. But the more transistors you pack into a die, the hotter it gets. This is offset by using smaller dies (less voltage). But you can only go so small before the transistors are literally touching one another and electron migration becomes more pronounced.

Again, since motherboards adhere to a standard, costs drop as the amount manufactured goes up. If all boards were proprietary, then yes, if they had extra complexity cost would be affected in a significant way.

There is still a price discrepancy with Intel and AMD boards today, even if there are little or no actual differences between boards.

I run an AMD system myself, simply for the fact that for around $380 (parts sourced through Criagslist ) I have similar power to my old 2009 Intel system which cost me nearly $2,000.

 

[Exophase]

The CPU's of today are more efficient at controlling heat vs previous designs. But the more transistors you pack into a die, the hotter it gets. This is offset by using smaller dies (less voltage). But you can only go so small before the transistors are literally touching one another and electron migration becomes more pronounced.

I think you mean it's offset by using smaller transistors (and keeping die sizes similar). Of course there's a limit to how far you can shrink. We aren't there yet. There's also a limit to how much you'd want to integrate. After Haswell there's only a few more things left, and that's still 22nm.

Integration as we've seen it so far and will see it in the future is not increasing thermals, that's all that matters with regard to your question. The stuff they're integrating are not exactly thermal hotspots either so it's not pushing the power density in a really bad way.

Again, since motherboards adhere to a standard, costs drop as the amount manufactured goes up. If all boards were proprietary, then yes, if they had extra complexity cost would be affected in a significant way.

I don't think you understand (even though I keep repeating it..) that when I refer to costs from motherboard complexity I'm NOT simply referring to design costs. More complex = more layers = more expensive PER BOARD.

I also think you're over-estimating what's standardized. Add a new component, any new component, and the layout can completely change, even if your high level schematic barely does. Obviously most motherboards are not using the same physical layout, and this is what I mean by design complexity of having more chips that have hundreds of I/Os on them.

There is still a price discrepancy with Intel and AMD boards today, even if there are little or no actual differences between boards.

I run an AMD system myself, simply for the fact that for around $380 (parts sourced through Criagslist ) I have similar power to my old 2009 Intel system which cost me nearly $2,000.

It's difficult to make a good apples to apples comparison because AMD has been using the same socket for a lot longer, and therefore you can run newer processors in older boards that overall have worse features.

At any rate, focusing on AMD vs Intel price value comparison is getting WAY off my point.. I'm not going to bother trying to prove that the price difference in motherboards when I was shopping was solely down to integration so you can take what you want from it. But I think it's naive to suggest the integration had NO effect at all. And you were saying you don't see the point with more integration in desktop systems, period..

 

[Phil]

I said I see no point in a desktop (i.e. enthusiast system). A desktop by default obviously has more room physically then say, a mobile setup. But Intel is pushing integration because the desktop market is nowhere near as lucrative as the mobile market. And everything is pointing at the mobile market superseding the desktop market sometime in the future. That's why they are so bullish on TDP these days.

As for the motherboard thing, I see your point. Of course the board will be more expensive if certain chips are present. But an MC does not cost $40 or more. All I am saying is that the large price difference is not solely down to a particular board being more complex. If you look at Intel's CPU's for example, why are they so much more expensive than AMD? Because they are more complex? Sure. But hundreds of dollars (sometime thousands) complex? Of course not. Call it poetic license, the material tech version.

Meh your right, this is getting way off topic, back to Ivy.

 

[Exophase]

So do you refute my claims that improved integration makes things cheaper on its own, and do you refute that lowering power and size also reduce price as a secondary effect? It isn't really a question of how much, it's a question of anything at all, because any price drop is an incentive and you said you see no reason for improved integration with desktops. Or do you think price is irrelevant in desktop systems? Note that I don't think that the large price difference was solely down to complexity either (but seriously, once again, the difference in manufacturing price is not simply due to additional chips costing more, the cost of the motherboard PCB itself is a significant chunk of the BOM cost)

Note that desktop doesn't always mean enthusiast either.. I've never owned a real laptop myself, just a crappy netbook. Desktops are just cheaper than laptops for the same sort of system, if you're just going to end up using a standalone screen and keyboard/mouse since the desktop doesn't try to build that in. You don't have to be a gamer to appreciate that.

When I think of the trend towards smaller/cheaper via integration and lower end parts even on desktops I think of something like Raspberry Pi.. which is ostensibly not a mobile platform but is getting attention due to its very low price and small form-factor. While you can't really call it comparable to a desktop either I do know people who intend on using it like a normal computer.

 

[Princess Garnet]

I run an AMD system myself, simply for the fact that for around $380 (parts sourced through Criagslist ) I have similar power to my old 2009 Intel system which cost me nearly $2,000.

This is an unfair comparison. I had a more costly system than you did, plus peripherals, plus upgrades, and I still might not have spent as much. I also bought it earlier than 2009 when components were likely more expensive. Either something is amiss, or the water cooling (or something other than the Intel/AMD exclusive parts) is why, and if so, that sort of voids your point.

I said I see no point in a desktop (i.e. enthusiast system). A desktop by default obviously has more room physically then say, a mobile setup. But Intel is pushing integration because the desktop market is nowhere near as lucrative as the mobile market. And everything is pointing at the mobile market superseding the desktop market sometime in the future. That's why they are so bullish on TDP these days.

A desktop is not just an "enthusiast system". The vast majority aren't.

It's funny to see how detached some people get from reality by looking at sales trends. Sales can be telling, yes, but they aren't the whole story.

As for when and how it supersedes it, I'll have more fun watching it than listening to people claiming it. Microsoft/Bill Gates said tablets would be more powerful than PCs either before or by five years time... back in around 2002 when Windows XP was new!

The matter is, at this point, they are all PCs, just in different form factors, and they all have their roles. I can see the desktop going away as the prime one, but it's not the same as it becoming extinct (I'll get more on this below).

Regarding extinction, it's not just the "desktop" in question here anyway, it's laptops as well (and anything "non-touch"). If anything, the laptop is more irrelevant in my opinion than the desktop now. But hey, that's just my foolish opinion; you'd never guess by looking at all the laptops around campuses, just like you'd never guess the desktop is dead by looking at the business or even home sectors. Still, with tablets/smart phones, that leaves the laptop filling a much more niche space than desktops do (and were it not for campuses, I feel it would be much more so the case). Desktops still have the advantage of power and cost, but laptops are stuck between desktops and smaller touch devices. They still have a space, yes, but it's so much more niche than the desktop is. Laptops were one day heralded to replace desktops too, and now they may never do that or even outlive them. Keep that in mind with all these "forecasts" people give. The one interesting thing about tech is that you can't truly know.

The truth is, like I said, these are all PCs in different form factors, and right now, they all are serving a good enough use.

Desktops are for power, flexibility, and cost. They're the biggest, but there's smaller form factors for them too.

Laptops are for mobility.

Tablets/smartphones are more so (for mobility and size), and for touch. In the case of smart phones, they're also, you know... phones (although PCs can do this, few use one primarily for that).

The power per cost and flexibility/capability generally goes down with each form factor. As of right now, as I see it, none are in no real danger of becoming extinct tomorrow. I do think that if/when the desktop does, the traditional laptop will be with it anyway. By then, we'll have tablets/smartphones spiritual successors that we hook up to "docking stations" for the role of desktops, while they themselves remain mobile. You can never know though; that's just my own subjective guess that even I'm not too sure on. Either way, I think things will change more than it will be a case of anything "going away completely", even if some things as we know it change. As they say, the more things change, the more they stay the same.

Anyway, yeah... big digression.

 

[Phil]

So do you refute my claims that improved integration makes things cheaper on its own, and do you refute that lowering power and size also reduce price as a secondary effect? It isn't really a question of how much, it's a question of anything at all, because any price drop is an incentive and you said you see no reason for improved integration with desktops. Or do you think price is irrelevant in desktop systems? Note that I don't think that the large price difference was solely down to complexity either (but seriously, once again, the difference in manufacturing price is not simply due to additional chips costing more, the cost of the motherboard PCB itself is a significant chunk of the BOM cost).

Why would I refute it? Integration will likely lower the overall price of a system. I said mass component integration was not really needed for a desktop system from a hardware standpoint, not a financial one. As for the rest, see below.

Note that desktop doesn't always mean enthusiast either.. I've never owned a real laptop myself, just a crappy netbook. Desktops are just cheaper than laptops for the same sort of system, if you're just going to end up using a standalone screen and keyboard/mouse since the desktop doesn't try to build that in. You don't have to be a gamer to appreciate that.

When I think of the trend towards smaller/cheaper via integration and lower end parts even on desktops I think of something like Raspberry Pi.. which is ostensibly not a mobile platform but is getting attention due to its very low price and small form-factor. While you can't really call it comparable to a desktop either I do know people who intend on using it like a normal computer.

Maybe I should clarify what I have been saying.

We started this convo about me making a claim about a desktop CPU. I claimed that integration of IO components onto the chips was not needed for a desktop system.

There is a reason I claimed that.

Let's with a desktop. I classify as desktop as a system that runs the most powerful version of current hardware. But you brought up that desktops are cheaper than laptops for laptops of the same hardware level. True, but this gets a bit murkier.

Cheap desktops are a bit different these days. Go to your local computer store, and check the specs of a cheap desktop. You probably be surprised to find that *surprise* they run mobile processors and gpu's. Celerons and mobile Athlon's. Sandy Pentiums and Mobile Tri Core's. Heck, even some Imac's run mobile GPU's instead of "desktop level" chips. And more than 50% run mobile chips in a desktop environment, and that number is growing by the day. Your posts about integration lowers cost as an incentive to buy, and that is correct. Mobile chips crave and NEED integration. The more they can integrate, the better they will be.

I have absolutely NO problem with these. None. Nada. Zilch.


Now this is where the disconnect and the reason I posted that claim comes in..

Regular, full TDP desktop CPU's. CPU's that are fullsize, and only meant to be run in a desktop environment.

Let's take the i7 2600K. Very few store desktops haven this CPU in a prebuilt non-ordered off the shelf system. Likely less than 10% of all desktops have this chip.

The i7 2600K has a mobile GPU on it's die.

Why? What purpose does it serve? Who with a K level chip will use a weak integrated GPU? Less than 5%.

Again, on saving power. The "K" denotes it's unlocked. So you can be pretty certain it will be overclocked with increased voltage.

So why lower voltage requirements...not just lower them, but make them a prime incentive to buy? Why so bullish on TDP if the desktop user has more than enough power to dissipate the heat?

The answer is obvious.

The Sandy Bridge core was designed as a mobile chip first, a desktop chip second.

And that trend will continue. Integration does little to nothing for enthusiasts (who own desktop systems). It does everything for your average user, who more than ever know choose a mobile platform (mainly laptops) over a desktop. Laptops surpassed desktops in sales back in 2008, and have never looked back.

Intel knows this. They are pouring billions into R&D not for the enthusiast, but for the lucrative mobile market. There are nearly twice the amount of CPU's available to the mobile sector verses the desktop sector.

To be on point though, integration of components will have no net benefit for someone who overclocks or uses the system extensively for cpu intensive tasks. Maybe Intel will find a way that this can benefit an enthusiast, I don't know, but as it stands, integration only enhances a mobile setup to a significant degree, not a desktop based one.

A long time ago I surmised that Intel would soon make the mobile line the focus before the desktop. And it's happening now. Intel is pulling out all the stops to make sure they own the mobile market out of the gate, before the chips even hit the shelves.

To this end, that's why you see a useless video chip on a processor that likely will never fully utilize it. To keep R&D costs downs and yields up, we desktop users get integrated chips.

Now I am all for cheaper systems overall. But a highend or even competent motherboard are not going to lower the price per unit just because a few chips moved onto the CPU die. They will stay the same (or even increase if you are dealing with a hot just out the door chipset.)

So in closing I hope I cleared myself up. All what you claimed is valid, but only makes perfect sense if applying to a mobile platform. A modern desktop does not need such integration to stay relevant.

Unfortuantly, desktops are not the main market now. Yes they have the most power. Yes they are the most (arguably) stable market. But there's one thing desktops do not have: Money. Billions upon billions are being poured into the mobile market. More people than ever are choosing laptops over desktops. Who gets a desktop for Christmas? Who gets a laptop for Christmas? You can buy a decent brand new laptop for less than $320. You can also buy a desktop with that money, but funnily enough, it will have the same or slightly more power, but more expansion options. That's it. The days of powerful desktops costing less than laptops are over. Thin laptops are still expensive, as are gaming ones, but at the super cheap level, they are actually very evenly matched. And to be honest, most people will spring for the laptop over the desktop.

Example:
$329
http://www.microcenter.com/single_product_results.phtml?product_id=0382601

$329
http://www.microcenter.com/single_product_results.phtml?product_id=0373743

Similar specs. The desktop is faster, but not by much. And remember, you need a monitor, keyboard and mouse. So if going apple to apple, you need something like this:

$199
http://www.microcenter.com/single_product_results.phtml?product_id=0384826

In such a situation, most people spring for the laptop.

My point is more and more the real desktop systems, powered by desktop processors, are usually in the hands of enthusiasts. The market for high end OEM desktop systems are disappearing. Pretty soon most cheap affordable desktops will be powered by mobile chips do to the higher yield and lower cost.

This is an unfair comparison. I had a more costly system than you did, plus peripherals, plus upgrades, and I still might not have spent as much. I also bought it earlier than 2009 when components were likely more expensive. Either something is amiss, or the water cooling (or something other than the Intel/AMD exclusive parts) is why, and if so, that sort of voids your point.A desktop is not just an "enthusiast system". The vast majority aren't.

For the desktop bit, see above. As for the rest, I could have saved nearly $300 by using AMD instead of Intel. The price I paid is irrelevant and my assertion is still valid.

It's funny to see how detached some people get from reality by looking at sales trends. Sales can be telling, yes, but they aren't the whole story.

Before I get into this, when I say "Mobile market", I mean laptops, not tablets or smartphones.
Aside from that, the reality is that the money being poured into Mobile, Smartphone and Tablet R&D likely quadruples the money being spent on desktop innovation.

As for when and how it supersedes it, I'll have more fun watching it than listening to people claiming it. Microsoft/Bill Gates said tablets would be more powerful than PCs either before or by five years time... back in around 2002 when Windows XP was new!

See above. Tablets will never replace desktops. They are for media consumption only until a better interface is designed other than a simple capacitive touchscreen.

On that point, instead of looking at outright power, look at relative advancement..

In terms of advancement, the mobile trifecta has moved far more quickly than desktops. Some smartphones sport 2gb of DDR2 ram, dual core processors exceeding 1GHZ in speed, screens with pixel density that make even the most dense LCD look pixelated. And all this happened within 5 years. And it gets quicker. You can't simply ignore that smartphone's are on the bleeding edge right now in terms of innovation speed.

The matter is, at this point, they are all PCs, just in different form factors, and they all have their roles. I can see the desktop going away as the prime one, but it's not the same as it becoming extinct (I'll get more on this below).

Desktops will not become extinct, simply because they are the most cost effective way to test new ideas...before dumping them into the fast moving mobile sector.

Regarding extinction, it's not just the "desktop" in question here anyway, it's laptops as well (and anything "non-touch"). If anything, the laptop is more irrelevant in my opinion than the desktop now. But hey, that's just my foolish opinion; you'd never guess by looking at all the laptops around campuses, just like you'd never guess the desktop is dead by looking at the business or even home sectors. Still, with tablets/smart phones, that leaves the laptop filling a much more niche space than desktops do (and were it not for campuses, I feel it would be much more so the case). Desktops still have the advantage of power and cost, but laptops are stuck between desktops and smaller touch devices. They still have a space, yes, but it's so much more niche than the desktop is. Laptops were one day heralded to replace desktops too, and now they may never do that or even outlive them. Keep that in mind with all these "forecasts" people give. The one interesting thing about tech is that you can't truly know.

Laptops are irreverent? How so? At this point the R&D finance being used to make new video cards or other components for your system is largely paid for by the proliferation of laptops. Laptops will not replace desktops. But where do you think a companies R&D money goes these days? To desktops? Or the more lucrative mobile market?

When was the last time a groundbreaking innovation was made for a desktop? They don't happen, they tend to do so overtime, because a desktop is a collection of components, while a laptop (to a normal buyer) is a cohesive whole. And that image is what's fueling the mobile push now.

As for tablets, sure, they are popular, but everyone who has one has a normal PC.

The truth is, like I said, these are all PCs in different form factors, and right now, they all are serving a good enough use.

I agree.

Desktops are for power, flexibility, and cost. They're the biggest, but there's smaller form factors for them too.

Agree with power and flexibility, cost, not so much. We are not in 2007 anymore.

Laptops are for mobility.

More and more people use them at home only. Like myself, none of my laptops leave my room unless absolutely necessary.

Tablets/smartphones are more so (for mobility and size), and for touch. In the case of smart phones, they're also, you know... phones (although PCs can do this, few use one primarily for that).

Some smartphones have buttons. And I think that the majority being touch based is their biggest downfall, regulating them to light (very light) work only.

The power per cost and flexibility/capability generally goes down with each form factor. As of right now, as I see it, none are in no real danger of becoming extinct tomorrow. I do think that if/when the desktop does, the traditional laptop will be with it anyway. By then, we'll have tablets/smartphones spiritual successors that we hook up to "docking stations" for the role of desktops, while they themselves remain mobile. You can never know though; that's just my own subjective guess that even I'm not too sure on. Either way, I think things will change more than it will be a case of anything "going away completely", even if some things as we know it change. As they say, the more things change, the more they stay the same.

Anyway, yeah... big digression.

I never heard that any one of them will become extinct, except for Job's mantra. I'm still waiting for the flux capacitor attachment for my desktop though. I heard IBM was working on it....

 

[PsyMan]

Past experience with all sorts of systems taught me one thing. That thing is to stay away from non-expandable systems with custom parts. Laptops. net-books, cellphones, tablets and the likes all fall to the same category.

Broken sound chip? Good luck with those usb sound cards.

Broken GPU? I hope your laptop can swap GPUs... and you have the money to get overpriced newish GPU.

Broken lcd? Good luck finding one compatible with that specific laptop you have.


The same goes for almost all parts on laptops. Embedded systems are worse since every single part is non-standard if it's even available for purchase on any shop.


No matter how you look at it, if you want a system to be in a working order after some part on it breaks then you'll get a desktop.


And I didn't even mention overclocking yet.

 

[Phil]

True lol. I hate fixing laptops. With desktops, just replace the part, and move on. And desktops are more sexy.

Lol I think we are getting off topic a little again. Can't say I did it this time

 

[Exophase]

Phil, the big issue I have, that applies to your entire post, is this distinction you're making between "mobile chips" and "desktop chips." True, there are different product lines, but what it really boils down to is two things: TDP and package. Chips with TDPs get there partially by more aggressive binning, but also by capping features and clock speed. Chips with less features and clock speed cost less, so there's a correlation - and what you're calling "mobile" chips includes the lower end of the performance spectrum because at that point it makes less sense to differentiate between the two. And non-socketed chips make sense soldering onto cheap motherboards for desktops, because again, that improves integration/costs, at least for high enough volume.

On the other hand, if you look at prices for the higher end lower TDP SB CPUs they cost way more than the highest end higher-TDP ones (go check out Intel Ark). So there's definitely a big middle of the road market where you don't want a desktop using CPUs that are usually used in laptops. And it doesn't start with the likes of something like i7-2600K.

Not to mention that even for desktops lower TDP IS cheaper (as I said at length before).

Sandy Bridge isn't "designed" for mobiles either. What it's designed is with the principle that it only increases performance if it doesn't impact power consumption. But it's also designed with the principle that every generation increases peak performance. This wouldn't be the case if they were now making "mobile CPUs." This is further evident by the fact that SB CAN be clocked a lot higher than it is, at the expense of TDP, but without messing with voltage. It's apparent that the TDP limit is selected due to other parts of the hardware ecosystem, rather than the design of the CPU selecting the TDP limit. You can't call the TDP bins a CPU microarchitecture design philosophy.

Yes, the motivation towards lower TDP is influenced by laptops, but it matters on the desktop too. Case in point: both Pentium 3 (the original Coppermines anyway) and Pentium 4 had their highest frequency parts pulled from the market because they generated too much heat to run reliably with the cooling Intel supplied. And yet the TDP for both was under the 125W that AMD bins at. AMD might manage, but those 125W CPUs from AMD come with stock fans that are way too loud under heavy load. Neither scenario is very desirable.

And sure, I agree that someone buying an i7-2600K (or especially say, SB-E) and spending hundreds of dollars on a GPU won't care about saving a few bucks on integration. But these are the minority among desktop sales, and I'm not just talking about apart from cheap ones - I mean there's a whole middle range that doesn't include these "enthusiasts." It's no surprise you don't see desktops with i7-2600Ks on display at computer stores, because that perf/$ gains you get at that point quickly become ridiculous. Most people will barely benefit from an i7-2600K over an i5-2500K (if at all) yet it costs $100 more. These are MEANT to only got to a few percent, but that's not a sign that desktops are in general supposed to go to the few percent.

It is a fact that laptop sales do enjoy some economy of scale due to their undoubtedly being more interest in them than laptops, but I think your example is a little too hand-picked.. with the laptop enjoying a $100 price reduction vs the desktop with a $30 reduction.. maybe in this country it really is getting to the point where cheap laptops cost what cheap desktops do but that isn't going to be the case everywhere, and the market in poorer countries certainly matters too. As for having to buy a keyboard/mouse/monitor.. these days people tend to go through computers faster than they go through those. I know I last bought a new computer without buying a new monitor (and input devices) and I'm anything but a habitual upgrader. Monitor progress in particular has slowed down a lot, is even going backwards in some regards.

 

[Phil]

The "middle market" you speak of is eroding...fast. Actually, the only thing keeping the "middle market" afloat are small business and corporations that need advancements in the desktop sector.

Put it this way. The desktop is the common denominator. You basically need one to do anything. But with laptops that have similar real world power, desktops sales will continue to decline year by year. I have a theory that they will level off, but notebooks are the catered to market. And with that being the case, let's go back to Intel.

So all the latest tech in CPU's they are making, is it catered to the desktop market, or the mobile market?

If you were a venture capitalist, where would you invest your capital?

So tell me, if the sales for desktops are decreasing every year, and the sales of mobile pc's are posting exponential growth every year, what avenue do you take to cater to the demand?

When I say "mobile chip" btw, it's still a regular CPU with mobile benefiting attributes. Also it's rare to see a daughter boards in a desktop or even some laptops (outside Apple). Most desktops still use sockets, regardless of cpu. Maybe Ultrabooks may have them, but I have not worked on any yet so I don't know for sure.

1:Lower TDP:Advantage Mobile

2:Integral on board IO: Advantage Mobile

3:Onchip Graphics: Advantage Mobile

4:CPU Core Behavior: Single Core Turbo Overclocking: Advantage Mobile

All of those can and do benefit a cheap desktop. I do not dispute that.

But to say that those attributes were made for a desktop makes no sense. Why make mobile friendly innovation for a slowing and diminishing market?

That's why the certain features of the Sandy Bridge Architecture was most likely designed with mobile computing in mind, instead of desktop computing.

To make this easier to see, look at the new Sandy and Ivy Desktop CPU's as beefed up version of the standard CPU design. Then it starts to make sense.

What I am simply saying is that Intel is baking in more mobile friendly features to it's CPU's each new generation. And the drive to do that is not to simply make desktops more efficient.

As for my example being hand picked btw, it wasn't. Find any laptop for 300-400 dollars. On the same site, find a desktop with the same price, and compare specs. The results will be the same. I choose Microcenter because it's near my house, but any site will do.

As an aside, it's also because of something you mentioned earlier, and I mentioned before that: We are seriously hitting limits of advancement in nearly all aspect now. Intel's "Tock" has lesser impact than it used to, especially back in the early days with chips such as the Pentium Pro. I think we are finally getting to a point of diminishing returns.

I had an interesting conversation with a guy that used to work on CPU design (back in the 60's) for IBM. He said he was blown away about how different things are these days. But the main point of our conversation was that the reason we do not see increases of IPC and even outright speed is heat. He kept in contact with several people in the industry, and said that there were a number of radical designs on the table (he also mentioned Intel's Tri Gate transistors) but heat dissipation was the main obstacle aside chip integrity. Makes sense, I guess.

 

[Exophase]

I'm still not seeing how ANY of this is about designing for mobiles or designing for desktops. Nowhere am I saying anything is made for desktops, I'm saying that chips like SB are made to target several markets/form factors. No matter what you think of the desktop market, worldwide, you're forgetting a major third market - servers. All of the things you described as being designed for mobile, except for onchip graphics, are just as much server advantages. And on-chip graphics is only replacing on-motherboard graphics as a superior solution. Motherboard IGPs, although not as standard as on laptops, have still been very common in desktop sales for several years, where desktop OEMs with IGPs probably grossly outsold those with discrete.

I'm kind of wondering what you think an alternative desktop-focused CPU would look like.. For what it's worth, more cores isn't a desktop demand either.

The big reason why your example looked hand picked to me is because the desktop had a $30 discount off of original price, while the laptop had a $100 discount. I still think you should look at poorer countries (USA doesn't dominate world's computer supply), because it's hard to imagine that something that costs so much more to manufacture is really going to sell for the same price everywhere, where there are substantial markets that really don't need laptops (work cubicles being one of them, regardless of the size of the company)

 

[Phil]

I see what you mean. Servers are a different market (I was mainly focusing on consumer sector but yes, your are entirely correct on where those attributes would benefit servers.)

And speaking of servers...

I'm kind of wondering what you think an alternative desktop-focused CPU would look like.. For what it's worth, more cores isn't a desktop demand either.

Come on, do you even have to ask?

Sexiest processor there is.

Anyway, I'm done, moving on.